https://homesignalblog.wordpress.com/feed/atom

From time to time, I enjoy browsing the LA Metro subreddit. A creature of the East Coast, I am forever fascinated by the Los Angeles’s preculiar mix of ambition and ambivalence around transit and urbanism, and will find myself spending hours dwelling on discourse from the basin. In a browse back in September, I came across a post lamenting reliability issues on the C and K lines.

This complaint piqued my interest. What might cause such a precipitous drop in reliability? Why is a fairly new rail line suffering so extensively from delays? As I dug in, I found that this Redditor had good reason to be upset. Between September and December, the C line was beset by delays, leading to irregular waits and missed connections for the line’s thousands of riders. The underlying cause of this disruption is an important one for transit. The C’s struggles are an almost too-perfect case study of how the fine details of schedule design can seriously harm transit’s ability to deliver reliably for riders and the communities it serves–or, in other words, how the hidden mechanics of our transit agencies inflect the service they deliver.

Why These Transfers Matter

LA Metro’s C line is something of an oddity. Borne of a consent decree imposed on CalTrans as they carved the Century Freeway through a densely-populated, working class, and overwhelmingly Black part of Los Angeles in the 1970s, most of the line runs down the median of the highway that created it. Unfortunately for its riders, this history means that the C’s stations are loud, dirty, and distant from the businesses and homes that drive travel demand in its corridor. It is not, by any stretch of the imagination, a well-designed transit route. Nevertheless, Angelenos use it: though ridership has dropped precipitously from its peak of 41,000 in 2014, its 2024 weekday ridership of 20,000 reflects the continued utility of the route.

Recently, the C line has been changing. When it was first built, the C connected stations along the Century Freeway to Redondo Beach by way of El Segundo. Though expansion to the north had been planned since the line’s conception, the hook-shaped service was initially designed to provide a way for workers along the eastern portion of the line to access manufacturing jobs–especially in aviation–closer to the coast.

Almost thirty years after the C opened, LA Metro completed the line’s northern connection. As a part of the agency’s ambitious network expansion plans, it built a new north-south light rail route (the Crenshaw Line) that connects the agency’s E line on the city’s west side to the C line by way of LAX. When this new line opened in 2024, the agency chose to reroute the C from its historic western terminus in Redondo Beach to the airport, with the new K line taking over the Torrance branch, and running northwards to the connection with the E.

The importance of transfers between the C and K stems from this change. The plan to modify the C’s routing was controversial: representatives from communities further east along the C wanted their branch to serve the connection with the E, while others worried about the loss of direct connections between Redondo Beach and the eastern leg of the C. To mitigate their concerns, Metro committed to scheduling quick transfers between C and K trains. Riders on a LAX-bound C have a 2-minute connection to a Redondo Beach-bound K at Aviation/Century, and will eventually have a 3-minute connection [link triggers PDF download] to an E line-bound K at LAX itself (the connections scheduled to date have generally been longer, in the 4-8 minute range).

Enter: Maintenance

Unfortunately, LA Metro inaugurated the C and K lines’ new service pattern in the midst of a disruptive series of service diversions. After thirty years of use, the C line’s overhead catenary is approaching the end of its useful life. To facilitate the requisite replacement work, the agency has been shutting down or single-tracking (running both directions of service on one track) segments of the C to give work crews unfettered access to tracks and equipment in need of upgrades.

The most recent of these diversions caused the transfer reliability issue on the C. Between September and December of this year, LA Metro’s C line upgrade project was working at the far eastern end of the line, in the vicinity of the Lakewood station. Work in this area forces trains to single track between a location known as Paramount Interlocking (about a mile west of the Lakewood station) and the end of the line at Norwalk. This restriction has two effects:

• Paramount interlocking and the Norwalk Station are about 3 miles apart. C trains cover that distance in around 6 or 6.5 minutes. Because trains now have to take turns passing through this area, the single tracking forces the agency to cut the frequency of C service from 8-10 mins to 13 mins, or twice the running time through the segment.
• The single track introduces an interdependency between the two directions of C service. If one trip is late arriving to the single track segment, it risks delaying the next train in the opposite direction. This linkage reduces the resilience of the C to delays: it’s easier for a small disruption to propagate across the whole line.

Here’s where operational minutiae become important. Agencies often design schedules for single-track operations to help prevent delays from propagating. There are several ways to do so. Some operators will run headways that are longer than necessary when single tracking to provide some “give” in the timetable for delays. Others will add a bit of recovery time upstream of the single track segment so that late trains have a chance to make up time before they reach the operationally sensitive work zone. Still others will extend layover times at terminals to ensure that late arriving vehicles and crews have a chance to get back on schedule for their departure. The slideshow below illustrates how these different mitigation options help reduce delays.

During their fall 2025 service change, LA Metro appears not to have employed any of these strategies to protect reliability. The 13-minute headway chosen for this operation approaches the limit of feasibility given the length of the single track segment, and is unprotected by any extra running time that could help stem the propagation of delays. To top it all off: because of the chosen headway and running time, infrastructure limitations at LAX forced schedulers to actually shorten layover times at the north end of the C when constructing the diversion schedule, providing only 4 minutes for turn-arounds at LAX. (If you’re curious to understand how scheduling choices 1 and 2 lead to this outcome, see the postscript). The outcome here? The C’s timetable became much, much less resilient.

From Broken Schedules to Broken Transfers

The consequences of those design deficiencies for the C were predictable and severe. With no breathing room in the single track segment and no chance to catch up elsewhere, the C was fragile. Service started each morning on time and then devolved as the day went on. A simple example: early on the morning of October 14th, a 4:42 AM C trip headed to Norwalk left LAX 3 mins late. The train maintained its lateness all the way to the single track segment, where it then forced a 5:08 C trip to leave Norwalk 3 mins late. That newly delayed train failed to recover its delay as it made its way to LAX and back, and when it returned, it delayed a 6:13 departure by 2.5 mins. A small delay thus produced 90 minutes of off-schedule C service, a testament to the sheer fragility of the diverted operation.

Needless to say, the timetable’s lack of slack showed up clearly in synoptic measures of the line’s performance. A 10-minute delay on the C used to have a controlled impact: the train experiencing the delay would catch up to its schedule after a trip or two, and service would otherwise run normally. Following the commencement of this single-track operation, the opposite was true. After a train reached that 10-minute mark, the overall lateness of C service consistently tended to grow, until the median C arrival was more than 5 minutes late. It was only 60-90 mins after the initial delay that lateness began to fall to more normal levels, usually following some extensive service management interventions by LA Metro’s dispatchers.

Once you multiply this challenge with slow delay recovery across dozens of incidents and operating conditions, you get a large drop in reliability. In July and August 2025, around 85 percent of C trains arrived at their station within 3 mins of schedule. In September and October, that fraction hovered around 60 percent, and at some stations, the rate of on-time arrivals fell below 30 percent. Though few C riders likely scrutinize their trip’s schedule, each of these delays represents longer travel times, extended and inconsistent waits for trains–and missed transfers.

This (finally) returns us to the Reddit user’s complaint. To facilitate the promised connections between C and K trains, northbound C trains to LAX are scheduled to arrive two to three minutes before southbound K trains to Redondo Beach (and vice versa) at Aviation/Century, providing an easy cross-platform transfer for riders who previously had a one-seat trip. The C’s spike in lateness broke this link. In September and October 2025, 20 to 30 percent of northbound C-to-southbound K connections at Aviation/Century involved a wait of more than 10 mins between vehicles, up from around just 4 percent in July and August.

Reliability

At a high level, what this country needs is more transit—more buses and more trains, with more supportive density along each important corridor. Returning to my initial reason for browsing this Reddit, few agencies have made more progress towards that goal in the last twenty years than LA Metro, whose growing network of rail and frequent bus routes represents genuine ambition for transit in a car-oriented milieu. But to fully realize the promise of those investments, the agency must protect reliability. To have any hope of serving the needs of its riders and offering a compelling alternative to the car, the transit LA builds must predictably deliver its promised speed and frequency. Lose that reliability, and transit suffers even without any service cut or shelved expansion plan.

If anything, the C exemplifies the abstruse chains of causality that often underlie those very reliability challenges. Delivering large, network-based services across often patchwork infrastructure is an inherently fraught task, one whose constraints and contingencies can often intimidate would-be reformers. But few of those complexities are truly unique, and most are soluble with effort. Much as policy researchers have begun to turn their attention to the dry mechanics of regulation and bureaucracy to improve the outcomes of their programs, reformers and agencies alike must familiarize themselves with the constellation of processes and channels of bureaucratic power that transform our goals for transit into moving trains and buses.

On December 1, the C’s single track operation ended. In its place, the C began a new diversion, in which half of trains are turning back at Willowbrook. This service plan appears to be performing more reliably than did the changes from the fall, but it nevertheless consistutes a fairly major reduction in service for riders headed to Norwalk. Thus continues the thankless work of upkeep: so long as the C line exists, it will require maintenance, and that maintenance will require service mitigation. It is my sincere hope that the next time Metro does repairs around Lakewood, riders will fare a bit better.

Postscript: Why Recoveries Got Shorter

During a single track operation, the primary constraint on timetabling is usually the single track segment itself. The headway between trains must respect that segment’s limitations (remember: that’s how we got to 13 mins) and trains in opposite directions must meet outside the segment. In the C’s case, that essentially forces a timetable where Cs in either direction run every 13 mins and pass just outside of Paramount Interlocking. The entire timetable is then built outwards from this constraint. If you run out that construction process, you create a timetable that puts C arrivals at LAX and C departures from LAX about 4 minutes apart–not a lot of time to turn a train around, and recover inbound delays!

Now, in theory, you could have a given C arrival turn for the second C departure, not the immediate next departure. In this world, a C train arriving at 1:00 would turn for the 1:17 departure, not the 1:04. Here’s where the infrastructure piece gets sticky: the C’s terminal at LAX is a single pocket relay terminal, i.e. there is only one track available to turn trains around. Having multiple Cs scheduled to occupy a terminal of this design could cause delays to K service, as there would be nowhere to stash the extra C train without blocking one of the through tracks.

In this particular scenario, schedulers might have been able to get away with longer turn-around times, as Ks were scheduled fairly far behind arriving Cs. However, doing so would have further exposed the K to the C’s performance challenges–so 4 minutes it was.

As always, all opinions expressed in this post are solely my own, and do not represent the opinions of any organization/employer.

Special thanks to Transsee and to Transit.land for providing easy access to historical data GTFS-RT and static GTFS data used to write this post.

All opinions in this post are solely my own, and do not represent the positions of my employer or any organizations of which I am part.

About two months ago, I found myself waiting for a 77 bus in a Boston rainshower. Like billions of others on this planet, I do not own a car, and rely on mass transit for travel. That day, transit service failed to meet my needs: despite the 77’s advertisement as having frequent service throughout the day, the next bus appeared to be over 30 minutes away. So, much to my chagrin, my bus trip ended up becoming a soggy walk through to a nearby train station.

The point of recounting this story is not to highlight my own inconvenience—my walk was not particularly far and I had no urgent constraint binding my arrival time. Rather, the significance of this story lies in the fact that my experience was deeply unremarkable. Every day, millions of people the world over board buses, and every day, millions of them are disappointed. A 2019 survey of American bus riders found that reliability was second only to frequency in its potential to increase usage; a 2018 study of transit satisfaction survey data from Santiago de Chile found that wait time reliability was the most important predictor of public transit user satisfaction. To make transit a reliable public service—let alone the bulwark against climate change and urban congestion that it needs to be—we need to make buses run reliably.

In recent years, discussions around bus reliability planning have focused on:

• Service management technology and techniques that would fine-tune the frequency and precision of dispatching interventions.
• Street designs that might reduce conflicts between buses and other road users (especially cars).
• Bus route design that could to reduce exposure to turns and road segments known to cause extensive delay.

This work has been impactful. Though the political landscape for buses remains challenged, many bus networks can point to material improvements in system performance related to their implementation of these practices. But in discussions around those general strategies, the details of how delays materialize and propagate can be lost. That day in the rain, I found myself wondering: how did this gap appear? 

In the weeks since my trip, I have often found myself playing with the MBTA’s bus performance data. The agency has made an admirable commitment to transparency by releasing raw bus performance data. These data allow interested researchers such as myself the opportunity to dig deep into the inner workings of MBTA service, and to take an honest crack at answering quandries such as the one I posed above about the 77.

Though these data capture only a small slice of the full complexity of the MBTA’s operations, the patterns one can find them are multiple and meaningful. Like many other bus routes, the 77 suffers a spate of headwinds endemic to transit operations in a car-centric transport regime. With limited priority and high exposure to environments dense with traffic, buses experience delay across their entire trip, and rapidly translate that delay into inconsistent service. But the 77’s performance data also reveal numerous operational factors that compound the effects of this structurally disadvantageous environment. Whether it be the design of the route’s limited bus lanes, its lack of schedule resiliency, or the consistent underperformance of handful of trips, riders’ experiences of the 77 are inseparable from the operational details of the route and the streets on which it runs. The 77 can thus begin to illustrate how the finer details of transit operations ramify the health of a service—and the networks of which each of them are a part.

Three notes:

• Due to my relatively greater level of knowledge on the topic, I am going to spend more time in this post discussing service management issues on the 77 than street design challenges. This is not meant as an implicit suggestion that streets factors in bus service are secondary to service management, but rather an attempt to spend my words where I think they might provide the most value, and to highlight the interconnectedness of the two problems.
• All analyses of operational performance data in this post focus on the weekday portion of the 77’s Winter 2025 timetable (December 2024-April 2025), which was the most recent completed timetable when I began writing this post in April. I focus on weekdays to simplify analysis—weekdays and weekends have different timetables, so writing this post to include all timetables would have made its length even more unwieldy while likely also being repetitive.
• The T’s open data is extensive, but not comprehensive. Innumerable institutional factors, pieces of operational context, and types of performance data are not available to the public. As such, while I have tried to dive deeply into the problems of the 77, my writings should be understood as a first look at an incredibly complex set of problems, rather than an authoritative account.

What is the 77?

The MBTA’s 77 is not a particularly interesting bus route. Like many of this country’s surface transit services, the route got its start as a streetcar connecting the Boston Elevated Railway’s Harvard subway terminal to the (then-peripheral) suburb of Arlington. Over the years, it has changed little; the route today is a frequent subway feeder for T’s Red Line running between a bus loop in Arlington Heights and Harvard Station (or rather, the bus terminal area adjacent to Harvard in Bennett Alley).

Unlike other MBTA routes, the 77’s timetable pattern is also straightforward. While many of the T’s routes share buses (“interline“), and some run trips that cover only part of each route (“short turn“), the 77’s schedule makes only limited use of either technique. Nor is its service profile particularly remarkable: though the scheduled end-to-end travel time of buses varies significantly with the daily profile of congestion and ridership, scheduled headways (transit slang for the time between vehicles) are relatively stable. This post analyzes weekday data from before the 77’s Spring 2025 service increase, so during peak hours the route was scheduled to run every 10-12 minutes (morning) or 12-14 minutes (evening). During middays and evenings, that frequency fell to 15-16 minutes.

Being a subway feeder route, the 77’s ridership patterns are also simple. Most riders board in either Arlington (heading inbound) or at Red Line stations (going outbound), and then ride through to the opposite end. While some riders do appear to make mid-route trips to and from schools and local shopping districts, the general rate of ridership turnover is low, while the long length of passengers’ trips means that bus loadings can be fairly high.

Operationally, the 77 is challenged. The MBTA’s service delivery policy defines bus performance in a two-part framework, focusing on service’s adherence to certain standards at “timepoints”—key bus stops used for scheduling—along each line. 

• On routes that run less frequently than every 15 minutes, buses must depart their origin 0-3 mins late, pass mid-route timepoints between 1 min early and 6 mins late, and arrive at their destination less than 5 mins late.
• On routes that run more frequently than every 15 minutes, buses must depart origin and mid-route timepoints no less than 1 scheduled headway, plus 3 minutes, after the preceding bus. They also must reach their destinations with an end-to-end trip time of less than 20 percent greater than schedule. 

The 77 falls into the former category, and sees middling performance: in 2024, only about 76 percent of the route’s weekday timepoint arrivals met the T’s standard, matching the average for the city’s frequent routes. Normal as it might be for buses in Boston, the line’s current level of performance leaves something wanting. During the route’s Winter 2025 timetable, a weekday rider at any one of the 77’s timepoints was liable to wait 1-2.5 minutes longer than scheduled, on average, due to trip cancellations, bunched buses, and variability—and one out of every twelve 77 trips followed a gap between buses at least 10 minutes longer than scheduled. Nor were trip times all that predictable, either: only about 81 percent of 77 trips meet the MBTA’s “scheduled runtime plus 20 percent” benchmark. The 77 is not a performant service.

Lateness, Variability, and Gaps

Understanding underperformance in transit services requires a working theory of operations. For most transit services, any such theory must begin with a route’s timetable. Timetables animate transit service: from policy, modeling, and budget inputs that determine the frequency, running time, and design of a transit network, timetables produce a detailed plan of action for daily service. They delineate how vehicles will enter and leave service, how they should move between terminals, and how they should be crewed. Yet important as timetables may be, few riders of frequent transit services like the 77 think about them. As I did that fateful afternoon, they turn up to a bus stop and expect the next bus to be some reasonable distance away. But here is the catch: in a timetable-based system like the MBTA’s, schedule adherence is what mediates regularity. 

On transit routes with even headways (as opposed to timetables that schedule bunching), there are essentially two varieties of service gap: cancellation gaps and variability gaps. Cancellation gaps are fairly self-explanatory—if you cancel a trip, you are doubling the time between buses. Variability gaps are slightly more complex. On most transit routes, lateness can be thought of as having a “trend” and “swing” component. Some line segments or hours will tend to see more lateness due to schedule design or service environment challenges that shift all trips’ schedule adherence. This type of structural misalignment is the trend component; in the 77’s case, an example of a lateness trend might be the fact that the median inbound 77 bus is about 4 minutes late at Porter Square. Lateness trends do not have direct gap impacts. After all, if every bus at a point is 4 minutes late, the rider experiences service whose spacing is approximately the same as it would be if they were on time.

The other variety of lateness is “swing,” or variability. From trip to trip, differences in loading, traffic conditions, operating style and more will impact just how late (or early) each vehicle is along its route. This type of variability can and will produce service gaps, especially on frequent routes. For example, on a bus route with an 8 minute headway (time between buses), one bus running 2 minutes early and the next running 2 late will create a 12 minute service gap—or, in other words, a 50 percent increase in headway impacted stops. Once seeded, these gaps tend to grow. A bus following a 12 minute gap is carrying 12 minutes worth of accumulated ridership, rather than 8. The extra load causes an increase in dwell time at stops, as both the volume of boarding or alighting riders increases, and the difficulty of squeezing onto a crowded bus grows. This effect progressively slows buses behind gaps down, eventually to the point where they bunch up with the buses behind them. Needless to say, this behavior is highly impactful to riders: not only do passengers awaiting these buses receive uneven service, but those aboard gap-following buses experience significantly lengthened travel times.

Dissecting the 77

If we accept that variability is important and that service gaps are critical drivers of both customer experience and vehicle delay, the path to understanding the 77 then lies through understanding its schedule adherence and gap rates. These aspects tell a remarkably simple story about the 77’s performance. Beginning with schedule adherence, the line clearly suffers from two major problems:

• In both directions, the spread of lateness values increases dramatically as buses proceed along the route. This is an indication of a lateness swing problem that would drive service inconsistency.
• Especially in outbound (i.e. Harvard to Arlington) service, the spread of lateness values is significant from origin terminals. In other words, buses are frequently being dispatched from Harvard late.

Gap rates tell a similar story. Breaking 77 headway data down to separate cancellation gaps from variability gaps, one can readily see that:

• Cancellation gaps are a non-negligible but relatively insignificant fraction of the overall gap challenge.
• Especially in the inbound direction, gap rates escalate as buses travel down the line.
• As one might expect given the lateness spread pattern in the schedule adherence charts, the 77’s outbound service suffers extensively from variability-related gaps that begin at Harvard.

These charts produce two simple questions: what causes the variability in 77 service? And what makes its terminals susceptible to dispatching service gaps?

Managing Traffic

The first step to understanding any bus route is understanding where and why its travel times tend to jitter. The 77’s challenges with variability are significant, and a structural feature of the line’s service environment. But, as ever, this general problem expresses itself through a handful of specific bottlenecks, whose challenges are broadly representative of the street planning tradeoffs that so commonly afflict bus service in urban environments.

Some of the 77’s variability challenges are straightforward products of traffic. Between Harvard and Walden Street, Massachusetts Avenue traffic volumes are high and bus stops tend to be busy. As a result, buses suffer traffic-related variability between stops and loading delays at stops. Similarly, at the northern end of the 77 in downtown Arlington, high ridership and apparent traffic interference from cars pulling in and out of local businesses increase variability. 

Alongside these simpler traffic volume delays, a handful of complex intersections wreak havoc in 77 service. The most impactful of these are the intersections of Massachusetts Avenue and Mystic Street in Arlington, and the intersection of Massachusetts Avenue and Alewife Brook Parkway on the Arlington-Cambridge line. In both cases, the high volume of turning traffic (and the attendant complexity of traffic light phasing and lane design) makes for variable levels of delay. Buses passing through these intersections are more likely to be cut off by other drivers, and are also more likely to be caught by a red light given that additional signal phases tend to extend the total cycle time of each traffic light. 

To the credit of involved municipalities, some of these congestion hotspots have received (or soon will receive) bus priority treatments. On the Cambridge-Arlington line, bus lanes carry 77 trips into the Alewife Brook Parkway intersection, and Cambridge is planning lanes for the lower portion of Massachusetts Avenue. While these bus priority measures have been unquestionably effective at reducing travel times, they are imperfect—and broadly emblematic of the North American circulation planning deficit so meticulously illustrated by Marco Chitti. 

Northbound 77 performance approaching Alewife Brook Parkway captures these challenges most clearly. During the morning rush hour, the Churchill Avenue-Gladstone Street segment of the 77’s journey has the unfortunate distinction of being the single highest-variability stop-to-stop segment anywhere on the line. The bus lane implemented in this area is a typical offset lane, buffered from the curb by a bike lane. Like many offset bus lanes, its exclusivity is regularly attenuated to permit right turns onto intersecting streets. While many of these permitted turns are low-volume (and therefore low-impact), the right turn flow onto Alewife Brook Parkway at the north end of the lane is not. Conflicts between this high turning volume and the flow of 77 service appear to cause extended and variable delays for riders, who often find their journeys muddled by a sea of turning cars. 

Terminal Troubles

With each additional increment of variability, the regularity of 77 service degrades. Buses run late, bunch, and strand riders at stops amid growing gaps. But bus service on the 77 and elsewhere is not always bad; a rough morning rush hour does not instantly lock the rest of the service day into chaos. Standing as one of the most important firewalls between one trip’s delay and the next’s timeliness are a line’s terminals, and the schedules that parameterize them. Just as en-route variability can determine the fate of a trip, terminals and terminal-related variability can tip the scales against a trip before it even turns a wheel—making them an essential (if often overlooked) element of the bus performance system.

As was immediately evident from schedule adherence and gap charts of the 77, terminals are a critical part of the line’s challenges. To be specific: 

• Over 35 percent of gap events on inbound trips and over 50 percent of gap events on outbound trips are associated with gaps that existed at a trip’s origin terminal.
• An additional 13 percent (inbound) and 10 percent (outbound) of gap events are associated with trips that left between 1.5 and 3 minutes after their scheduled headway—trips whose gaps may have been “seeded” at origin, in other words.

These data highlight the vulnerability of the 77 to ricochet, or the tendency for lateness and gaps to bounce through terminals. On most bus networks, schedules are assembled in “blocks” that define the movement of equipment on a line. Critically, vehicle blocks at many agencies are tightly bound to crew runs—drivers may change buses around their lunch breaks (to give one example), but otherwise will often work with a single vehicle for their entire shift. Crewing bus service in this manner simplifies operations and can reduce costs, but it also transforms terminals into simple input-output processes. If a block’s schedule provides (say) 5 minutes of recovery time between trips, and an inbound trip arrives more than 5 minutes late at the destination terminal, the outbound trip will necessarily leave late in the opposite direction, and may actually recover little of its inbound lateness at the terminal when involved drivers need to utilize their break between trips to use the bathroom or otherwise.

On the 77, this challenge with recoveries is particularly pronounced at the line’s Harvard terminal. Turning buses around at Harvard requires a fair bit of movement through congested streets, meaning that minimum turn times and turn time variability are high. Conversely, the volume of buses terminating at Harvard means that scheduled turn times cannot run much longer than minimum turn times, lest buses laying over between trips on the eight routes terminating at Harvard’s busways saturate all the street space in Harvard Square. The result is a terminal that is scheduled for stress. Even buses that arrive at Harvard early are liable to leave late, and those unfortunate enough to arrive at Harvard more than a few minutes late will almost certainly leave late as well.

As a result, the 77’s ability to recover delays and gaps on inbound trips is limited. Almost 30 percent of all trips departing Harvard are both leaving more than 90 seconds late and are doing so because of a late arriving inbound bus. In gap terms, this means ricochet: excluding gaps from cancelled trips, 50 percent of service gaps that pass Porter Square going north throughout the day correspond to a southbound gap that bounced back from Harvard. The kicker? 40 percent of those southbound gaps originated at the north end of the line in Arlington.

The Outliers

Terminal performance and runtime variance challenges are structural features of 77 service—they impact all trips, all day. Yet like most such phenomena, their impact is uneven. The 77’s gap rates readily display the lumpiness of service degradation: ten trips in the line’s schedule (approximately 5 percent of the day’s scheduled service) are responsible for 13 percent of all service gaps. While these trips’ impacts are extreme, their challenges are generally pedestrian; understanding them can help draw out how operational challenges compound global issues with variability and terminal performance to produce inconsistency.

It does not take long to notice that the 77’s outliers are patterned. Though they are spread across the day, the structure of outlier trips on the 77 reflect the route’s tendency towards ricochet, with outlier trips often related to previous outliers. This effect is much more clearly viewed on a stringline chart (a time-distance view of performance) than a heatmap, and is even more legible if one expands the list of outliers to include the top 15 and not just the top 10. The 77’s outlier trips not only pair over a round trip, but also display tight sequential linkages throughout the day. In the Winter 2025 timetable, eight of the fifteen worst performing trips on the 77 were part of the same vehicle block, T77-121.

Representative of block 121’s troubles is the 3:17 PM departure from Harvard to Arlington Heights, the second worst-performing trip in the 77’s Winter 2025 timetable. 56 percent of its arrivals at measurement followed service gaps, or more than twice as many as the 77’s average gap rate.

As with all outliers, the trip’s poor performance was related to lateness variance. Where the preceding outbound trip from Harvard (the 3:04 departure) generally left Harvard on time, the 3:17 tended to leave 4 minutes late, effectively guaranteeing a persistent gap in service.

The 3:17’s poor departing performance is, in turn, a study in compounded impacts. The trip is scheduled to turn from an inbound trip that arrives at 3:11, but unfortunately, that trip tends to arrive at Harvard about 3 minutes late. Under favorable conditions, it takes about 4.5 minutes to turn an inbound bus back for outbound service at Harvard; with 3 of the 15:17’s 6 minutes of turn time generally lost to lateness, the bus’s schedule ends up forcing the trip into a late departure from the Harvard terminal. To put some numbers to this pattern: in the Winter 2024 timetable, two thirds of all 15:17 trips leaving Harvard more than 90 seconds behind schedule arrived at the terminal sufficiently late to make a late outbound departure inevitable.

If outbound performance is driven by unrecoverable inbound lateness, inbound performance appears to be a product of some murkier dispatching problems in Arlington. In keeping with the 77’s general theme of terminal performance challenges, the inbound trip that feeds the 15:17 (the 14:36 from Arlington) develops a significant portion of its lateness at its origin terminal. Interestingly, however, the trip’s late departure rate does not appear to be a (direct) function of inbound lateness. While the 14:36’s feeder trip is also a poor performer, the trip has 17 minutes of turn time in Arlington; theoretically, a turn of that length should provide adequate margin to absorb the 6 minutes of lateness that generally attends the arrival of the previous trip.

This pattern is, unfortunately, representative of a broader challenge in the outlier block. As its day goes on, this block’s timeliness leaving its terminals falls. The first inbound trip of the day in this block leaves Arlington late 8 percent of the time; the final one does so 62 percent of the time. While some of this deterioration is a function of accumulating delays throughout the day, it is difficult to tell exactly what is driving the underperformance of this block’s 14:36 and 16:07 trips at Arlington. Perhaps some local ridership generator tends to let a wave of people onto this bus, causing boarding delays; perhaps the operator of this block changes over around these two trips; perhaps these trip are often held. Whatever its cause, the pattern and its effects are clear: the 14:36 leaves late, gets later, and arrives at Harvard with insufficient time to turn around before its next trip.

A Day of Gaps

Up until this point, we have been discussing the 77’s challenges in aggregate terms. Months-spanning analyses of variability, trip performance, gap rates and more are indeed essential to understanding how routes break down—but making these observations actionable requires an understanding of how they ramify real buses, rather than an imaginary average vehicle.

April 2nd of this year happens to have exemplified the 77’s issues. There was no immense breakdown of service that day, nor any rash of cancellations—and yet, putatively frequent 77 service that afternoon contained gaps close to half an hour in length. Behind those sizeable service gaps lies the sort of ricocheting service deterioration on which this post has focused.

The sequence of events on the 2nd appears to have been as follows (see the animated stringline below):

• Due to some traffic or loading-related variability, block 112’s 13:39 trip from Arlington to Harvard arrived at Harvard about 4 minutes late (1)
• With only 4 minutes of scheduled turn time at Harvard, that late inbound late trip produced an even later outbound trip. The 14:17 trip from Harvard back to Arlington left Harvard 8 minutes late, and (predictably) lost time on its journey back north. By the time it reached Arlington, the bus was 13 minutes late. (2)
• The 13 minutes of inbound lateness on block 112’s trip does was not absorbed by the block’s 8 minutes of turn time at Arlington. The outbound trip left Arlington 11 minutes late, and ended up behind a pair of buses the 77’s timetable inserts at Appleton St, seemingly to handle dismissal crowds from Arlington’s high school. The resulting gap in 77 service ahead of the school trips ensured that they would perform poorly: the first of them (part of block 113) ended up arriving at Harvard about 8 minutes late (despite a punctual departure from Arlington), while block 112’s inbound trip from Arlington struggled along with almost 15 minutes of lateness arriving at Harvard. (3)
• Predictably, these late inbound trips produced lateness outbound. Block 113’s 15:53 outbound trip left 6 minutes late, while 112’s 15:42 trip left 19 late. Both trips maintained their lateness to Arlington, where they arrived 9 and 15 minutes late, respectively. [The missing stop records for Arlington Heights appear to be a data error] (4)
• Upon these trips’ arrival at Arlington, they finally caught up to their schedules. Both trips made roughly on-time departures back to Harvard, ending the propagation of the gap. (5)

The relevance of this day of service is similar to that of outliers. Each day’s traffic, loading, and dispatching conditions will vary, but the basic weaknesses of a route will generally stay the same. That the trips involved in April 2nd’s gaps performed well on April 3rd tells us little; that gaps on a different set of blocks ricocheted through Harvard that evening tells us lots.

A Better Bus

On April 6 of this year, the 77’s Winter 2025 timetable ran for the last time. When it started the next day, the 77’s Spring timetable brought a significant service increase, improving travel for the thousands who use this route each day. Unfortunately, the improvement in service did not accompany an improvement in reliability. In this new timetable, the 77’s problems appear to be fundamentally the same: the route still suffers from terminal-driven gaps and extensive variability.

This point bears emphasis as incitement. 77’s problems are complex and persistent, but fundamentally tractable. There is no entropic malaise cast across the route, nor is there a lack of information about its failures. Improving the route requires some tinkering with street and bus schedule design, and ambition with service management. 

The easiest way to fix the 77 is to not break it in the first place. To succeed, the route needs a street environment that minimizes traffic interference to limit service variability. Cambridge and Arlington’s existing bus lanes may not be perfect but constitute commendable steps in the right direction; they produced significant reductions in travel time when installed. Upcoming work to redesign the remainder of Massachusetts Avenue in Cambridge promises to bring similar improvements. These projects will significantly expand the scope of bus priority on the 77, and will come with an accompanying build-out of floating bus stops that should reduce delays 77 trips experience when pulling out of and into stops.

Despite these measures, challenges with traffic conflicts will persist. Cambridge’s transit priority plans are laudable, but are neither comprehensive nor continuous: they preserve most right turns across bus lanes, maintain complex traffic flows at intersections, and (prospectively) involve part-time bus priority measures. If the controversy over the existing spate of improvements is any guide, a more ambitious traffic circulation redesign for Massachusetts Avenue may have been politically infeasible. Yet maximizing the reliability of surface transit inevitably means confronting these sorts of tradeoffs. Even if on a limited scale, our understanding of bus priority must evolve to include a greater emphasis on controlling the conflicts inherent to complex urban traffic patterns.

Once variability sets in, it becomes the MBTA’s responsibility to manage it. While the T has made strides to improve bus dispatching and routing in recent years, the 77’s performance highlights the some of the agency’s potential avenues for further enhancement. On the 77 and elsewhere, the T has an opportunity to develop a contextual service management playbook that aligns schedule design, ridership patterns, and disruption response strategies to ensure that riders receive the best service possible.

Many contemporary bus service management discussions revolve around the tradeoff between two different models of service regulation:

• Schedule regulation. Many transit agencies provide bus operators with detailed schedules that tell them what time they should leave various stops along their trips (timepoints). If they arrive at any of these timepoints early, they are generally asked to hold until their scheduled departure time. On low-frequency services where riders are reading a timetable before catching their bus, this practice ensures riders do not miss an early bus; on medium- or high-frequency services where riders look for even headways, holding reduces total variance from schedule by controlling earliness. Schedule regulation has the advantage of being simple and largely passive (as in, implementing it requires little more than a paper card with a timetable)—but it can both be a blunt and costly instrument. If traffic happens to be light one day, all buses will end up holding for earliness; if it is heavy, the schedule provides little spacing regulation because all buses will be late. Schedule regulation’s efficacy as a gap management mechanism also hinges on there being sufficient slack in the schedule for poor-performing buses to recover delays, which can lead to excessively-padded timetables and therefore higher costs.
• Headway regulation. Contra schedule regulation, headway mangement proposes a more selective approach to service regulation: dispatchers make every effort to depart buses from terminals at a set headway, and to re-space buses for evenness at mid-route points irrespective of their timetable deviation. As the theory goes, this more selective (and active) approach helps align service management incentives with riders’ interests on high-frequency routes where people are not likely to be waiting for a specific scheduled bus, and also helps cut the amount of padding in timetables. Operations that have switched to headway management have generally (though not universally) seen improved reliability on high-frequency routes. However, the practice can impose significant demands on agencies’ dispatching staff and IT budget due to the need for more active interventions in service and (often) more extensive technological supports for service management.

The MBTA’s network is currently managed on a blended schedule and headway basis, and with a light touch. As discussed previously, the agency’s metrics are headway-based on frequent routes. In practice, however, it appears that the agency’s service management strategy is built around schedule management. Unlike other schedule-driven agencies, the T makes limited use of passive schedule regulation; at least as of 2018, operators did not get instructions to hold at mid-route timepoints. The absence of regulatory holds means that service is only managed by dispatchers and by the capacity of terminals to absorb delays, which helps explain why variability on the 77 increases monotonically between terminals.

Ironically, the 77’s ridership patterns underscore the import of how one operationalizes either schedule or headway management, more so than they provide a resounding endorsement of either model. The 77’s largely end-to-end ridership profile means that it is important to ensure evenly spaced service at origin terminals—and, conversely, that any service regulation actions which slow buses down mid-route may inconvenience more customers than they help. Because of its high frequency and considerable variability, it likely makes sense to emphasize headways over lateness on the 77. But in practice, the literature that offers the greatest actionable input for the 77 service is that which describes ways to optimize the placement of “control points,” the stops in the schedule or a headway regulation scheme where buses will be more frequently held for alignment. On the 77, these key points are the line’s termini. While the T certainly should monitor mid-route service quality to help build intra-line ridership, improving 77 service in the short run probably means focusing on improving terminals, reducing operations-side sources of en-route variability, and making surgical applications of active service management.

The obvious place to start in any such effort is with scheduling. Many of the problems discussed in this post hinge on the difficulty of recovering from delays at the 77’s terminals—especially Harvard. While longer turn times have a cost in crews and equipment, the extent of their impact on the 77’s performance suggests that this is a cost that should be borne to ensure better alignment between promised and delivered service. In line with accepted practice elsewhere in the industry, the T might evaluate terminal turn times long enough to absorb normal levels of variability, with a margin for operator breaks. Explicitly tying turn times to variability could both help reliability and demonstrate the benefits of bus priority measures: every increment of reduced traffic interference from bus lanes can produce faster bus trips and less slack time at terminals.

Even as timetables change, the nature of distributions means that a few trips in every schedule will cause outsize performance problems. To the extent that these trips can be quickly identified and then managed with operations changes (e.g. different instructions to operators, or regularly allocated dispatcher time to a period with a concentration of outliers) they should be. Not all of these challenges will be solvable, but investigating outliers equips agency teams with valuable information on which scheduling, operations, crew performance/training, and traffic factors tend to create them. In the complex information environment of a control center, knowing what to watch and where to focus efforts can help leverage resources to great effect.

Lastly, the T likely needs to evaluate the resourcing of its bus dispatch function. The T relies on control center staff and strategically located field supervisors to control bus service. Recent data are not available, but as of 2018 the T’s control center had one of the higher buses-per-bus-dispatcher ratios in the American transit industry. This statistic is significant because most bus gaps are recoverable with active service management. Active service management contrasts with passive management (e.g. holds to schedule) in that active strategies are implemented on an as-needed basis, and that they can change a bus’s movement plan by (for example) having a trip skip some stops or make an unplanned short-turn. Given the 77’s tendency to acquire snowballing lateness and its ridership’s sensitivity to mid-route holds, providing more active supervision may be especially beneficial for the route. Recalling the example from April 2nd, running one of the impacted buses express for a few stops could have helped prevent further delays—but having the bus do so would have required a dispatcher’s intervention. If this staffing discrepancy still exists, increasing the number of control center dispatchers (or the level of automated support available to them) may yield a strongly positive return on investment.

Conclusion: Putting it Back Together

The 77’s story is not unique. Those who have marinated in bus dispatching literature will probably find my observations a tad unoriginal—and that is because they are. For decades, practitioners have identified street design, terminal operations and schedule structure as key drivers of bus performance, and for decades, these issues have persisted. Yet the universality of these issues is both a challenge and a strength, because it means that operational remedies for the 77’s problems may be transferable. One need look no further than the rest of the MBTA’s network for evidence of the same. In 2022, the busiest bus route in the T’s network was the 28, a workhorse route through Mattapan, Dorchester and Roxbury. Plotting its service gap charts reveals ailments similar to the 77’s—mid-route variability, and a high volume of service gaps originating at the line’s terminals.

At their core, then, the 77’s challenges both reflect the unending tyranny of operational details across bus networks, and these details’ entanglement with much larger strategic and policy choices around scheduling, street space, and more. It is both true that the incremental work of addressing the tangle of operational, schedule, and traffic performance problems that produce outliers like the 15:17 can have outsize impacts on any bus route’s performance–and that such work is essentially Sisyphean absent a policy decision to invest in realistic schedules and build an extensive network of good bus lanes. Conversely, addressing the fact that the T (pre-COVID, at least) had fewer dispatchers per active bus than most peer agencies would achieve less than one might hope without a finely tuned sense of where and how different bus routes tend to break down. To generalize: producing good outcomes for the users of complex systems requires clear and well-crafted policy goals, intelligently designed network strategies, and strong operational execution. Critically, information must flow in every direction across this stack. Strategies and policies developed independent of tactical knowledge will fail; operational plans produced without any strategic or political framework will not cohere.

The importance of this feedback loop is perhaps my main point in this post. Transit faces innumerable political challenges in this country—funding crises, a political refusal to confront road space allocation tradeoffs, fickle support for network expansion, and more. There is no way to operate one’s way out of these challenges, as they are endemic to this country’s political economy of urbanism and transport, and thus require political advocacy. And yet, within this challenging landscape, and in the face of ever more attractive alternatives to bus service, leveraging the controllability of details to ameliorate service provides an accessible (and often cheap) lever for agencies to achieve ridership and revenue improvement. Many more riders might avoid a walk through the rain if agencies choose to pull it.

Over the past year, American railroads seem to have discovered the value in growth. After over a decade of Wall St-fueled cost-cutting and service reductions, regulatory and financial pressures around the industry seem to have turned its attention to traffic—sparking hope among industry commentators and critics alike that the 2010s’ focus on capital discipline and traffic demarketing may be over. 

Common across all these new growth plans is an emphasis on intermodal traffic, the business of carrying containers and trailers on freight trains. CSX, one of the two major railroads in the eastern United States, suggested to investors suggests they can grow intermodal traffic faster than industrial production; the recent merger of Canadian Pacific and Kansas City Southern was sold to regulators, in part, on promises of significant traffic growth from new all-rail intermodal services between Canada, the Midwest, and Mexico. Nor are these aspirations just bluster, either: intermodal traffic has accounted for over 100 percent of railroad volume growth since the 1980s. In a country whose increasingly service-driven economy demands fast and reliable freight service, the simplicity, speed and flexibility of combined rail-truck service continues to allow historically sluggish and unreliable railroads to remain relevant players in the freight market. Or, at least, this is how the orthodox narrative goes.

But is rail intermodal service all it is said to be? Yes, intermodal traffic has tripled since the 1980s; yes, the only way your Amazon package ever meets steel wheels is in a container train. But a closer look at traffic reveals a more nuanced reality. The intermodal market today is small—railroad intermodal carries only 6 percent of the long-haul truckload freight market—and highly concentrated. Over half intermodal traffic is associated with foreign trade; eighteen percent of all North American intermodal volume moves between the American Southwest and Midwest. Intermodal today is not a comprehensive alternative to trucking, but rather a specialized service that has managed some incredible successes while failing to provide a comprehensive answer to trucking’s offer.

At the core of these mixed successes is an inability to effectively manage complexity. Railroads have spent the last 100 years simplifying their networks and operations in pursuit of productivity at almost any cost. Far from being an exception to that general trend towards market retrenchment, intermodal’s history is a product of these same tendencies. Railroads’ productivity-oriented operations philosophy has produced a service offering that reproduces their strategy for carrying bulk commodities while supposedly attempting to win service-sensitive flows of high-value consumer goods. These features—a near-exclusive focus on long trains, low network complexity, and high-volume corridors—make intermodal efficient and effective in markets flush with freight traffic, but crimp its growth beyond them. We should not minimize the significance of intermodal’s successes, but its limitations force the question: if railroads are going to hitch their futures to intermodal, how can it be made to compete?

Railroad Operations 101

Railroads are in the business of aggregation. A train is just a temporary union of many shippers’ freight; the strength of any given railroad’s economics hinges on the productive and temporal efficiency of that aggregation process. Since the 1960s, railroads have gone about assembling traffic into trains in two ways. In the traditional merchandise freight model, they gather small quantities of traffic from dozens of shippers, and then group it into trains at classification yards to move traffic to its destination. In the unit train model, they work with large shippers to run dedicated trains that shuttle back and forth between (for example) a coal mine and a power plant. 

Each of these approaches has advantages and limitations. Running merchandise freights allows railroads to act as a general-purpose transportation mode: anyone who can fill a railcar with their wares can ship in this network, provided they have access to the rails. However, sorting and re-sorting merchandise freight introduces delays and ineffiicencies into rail networks. Sorting yards are expensive to run, and complexity tends to come at a cost to reliability and speed—freight cars in merchandise service spend the majority of their life sitting still.

Unit trains offer a more attractive service product. In this operating model, a shipper fills an entire train with a single type of freight, and then the railroad hauls it to a destination. This incredible simplicity avoids the expensive and delay-creating sorting steps inherent to the merchandise model, and can provide something of an ideal for railroads and cost- or service-concious shippers alike. Unfortunately, they are only viable insofar as there exists large volumes of homogenous traffic to fill them. A unit train that must run with only a handful of cars makes poor use of labor; one that runs once a week (or less) wastes the potential productivity of the involved railroad equipment.

Railroad traffic does not come with some preordained sorting into “unit” or “merchandise” freight. The rules of the sorting game are prescribed by railroads themselves, and are currently driven by their mathematics of train length. If you take the shippers’ needs as constant, there is a rough tradeoff between average train length and network complexity: running shorter trains allows railroads to capture more traffic in their unit train network and handle the remaining merchandise freight with point-to-point trains that bypass intermediate sortings. However, while this strategy may produce better service and reduce terminal expenses, it increases train-related costs. Whether running 10 cars or 200, each train requires two crew members and consumes track capacity. The incentives of railroads’ managerial regime tend to favor economies in train service over network cost reduction or market-centric planning. As a result, railroads today run unit trains only where they can accrue hundred(s) of cars of traffic at a time, and handle the remaining freight in a series of behemoth merchandise trains. 

Intermodal’s current challenges stem from the fact that it tries to refuse this tradeoff while dealing in service-sensitive freight. Flows of intermodal-friendly traffic web the United States, connecting factories, ports, warehouses, farms and cities. And yet, railroads try to make these flows fit into point-to-point networks built in the image of unit trains. Intermodal freight is not carried on high-complexity trains, or shorter trains that can serve smaller markets without extensive sorting. Instead, intermodal service is the near-exclusive province of long and (relatively) simple trains to minimize the costs of sorting on one hand, and the number of trains they need to run on the other.

What results from this model is a sharply uneven set of freight services. Intermodal shippers get truck-like railroad options between North America’s largest cities—but outside those corridors, the situation is dramatically different. In many markets, intermodal simply does not exist: the railroads industry’s focus on simplicity means that intermodal can only be offered between city pairs that generate sufficient traffic to reliably fill a good portion of a miles-long train. The resultant sparseness of the nation’s network is hard to overstate. The only places to which one can ship a 53 foot container from the entire state of Nevada are Chicago (from Reno) and Los Angeles (from Las Vegas); the only place from which one can reach Nevada is Chicago. 

Where intermodal is offered on lower-volume lanes, service is often poor. Due to railroads’ focus on train length, service in these smaller markets tends to suffer a host of speed, frequency and reliability challenges that keep shippers on the highways. It takes less than half as long to ship a container from Chicago to Northern California by rail than it does to move one from Denver; it takes more than an entire day longer to move a container from Atlanta to Philadelphia than it takes to move one to New York. The tension these examples capture is at the heart of intermodal’s struggles. Without a more effective means of serving lighter-density markets, its prospects are limited. At this juncture, two questions arise: where did this focus come from? And how, if at all, can it be fixed?

Where it Began

The root of intermodal’s problems, ironically, lies in its successes. Intermodal’s growth took place because the rapid, late-20th century growth of international container trade enabled railroads to grow even while entrenching and then maintaining their dual focus on simplicity and volume. Thus soothed into complacency, the industry has not had to face the strategic incoherence of intermodal service.

These problems can be traced back to the 1960s, when carriers realized that their newfangled intermodal services (then known as “piggyback”) did not actually make them much money. Highly rated by shippers as an ideal blend of trucking’s flexibility and railroads’ economies, piggyback had grown from 150,000 loads in 1954 to 890,000 in 1964. However, just as this traffic gained a meaningful foothold in high-value freight markets, it became clear to railroads that piggyback’s growth was something of a double-edged sword. As carriers adapted their accounting mechanisms to the markedly different cost structure of intermodal freight, they realized that the margins on intermodal were low—sometimes even negative. This caused something of a panic in the industry, and rapidly focused railroad managers’ attention on two key issues.

The first was investment. Railroads had taken to piggyback cautiously, and their initial piggyback infrastructure reflected a focus on cost minimization. However, as intermodal grew, these decisions proved to be false economies. Take terminals, for example. Railroads initially set up intermodal facilities in any city of significance that they served, leading to a veritable explosion of intermodal offerings—the United States had about 1,400 intermodal terminals in 1971. However, these terminals suffered from two major issues. First, most of them handled very little traffic, which led to high costs as the basic maintenance on the facility had to be paid for by only a handful of movements a day. Second, even where successful, these terminals suffered a lack of mechanization. Before railroads bought specialized rail cars and cranes, each intermodal load had to be driven off of its train car individually in a process known as “circus loading.” A circus terminal could be erected anywhere at almost no cost—it was little more than a track and a ramp—but unloading more than a small number of trailers in this fashion was slow, labor-intensive, and inflexible. At 50 trailers per day, circus terminal costs just slightly exceeded those of mechanized operations; at 300, they added 25 percent to the cost of unloading a train.

The other and perhaps more intractable problem was service. Piggyback promised to reduce the inconsistency of merchandise service by replacing some parts of a railcar’s trip with a truck, and by consolidating freight loadings at points where they could be hitched directly to premium trains. In practice, realizing these benefits within the confines of an ad-hoc merchandise operating philosophy required dedicated intermodal trains. One railroad which studied its intermodal operations in the mid-1970s found that shipments handled as a part of merchandise volumes spent over a day waiting in yards, averaging only 16 miles per hour between origin and dstination—compared to more speedy dedicated runs which could average over 30. 

The challenge with dedicated trains was that they required large volumes of freight. In those early years of intermodal railroading, volumes were low and were spread across a larger number of terminals in a more fragmented railroad network. What resulted was a volume trap. Adding a few cars to an existing merchandise freight was easy and almost costless; running the same cars as a dedicated train was often unprofitable. And yet, the only way to attract more cars to those trains was to shift them out of the merchandise network—so railroads needed to find ways to centralize and consolidate volumes.  

Saved (?) by Trade

Facing these challenges, railroads mechanized and simplified their networks. Circus terminals rapidly gave way to centralized intermodal yards with cranes and forklifts, thinning the ranks of intermodal terminals from over 1,000 in 1978 to only 360 by the mid-1980s. These terminals were connected to each other by dedicated trains of specialized intermodal cars, each of which would carry traffic for just a handful of other points to limit switching. While not exactly unit trains given that they usually carried freight for multiple points, this “new” intermodal product was created in unit trains’ image—and critically, was implemented without any significant changes to train lengths. Even as industry commentators heralded experiments with shorter intermodal trains on point-to-point runs as potential trailblazers of a new approach to intermodal operations, railroads remained focused on train length. Insofar as the industry was struggling to recover intermodal’s profit margins, there was a logic to this approach: amortizing the fixed costs of operating a train across a larger amount of traffic would indeed improve profitability. Yet in this decision lay the fabric of intermodal’s strategic constraints: the cost-management approach to rail operations shrank the industry’s room to grow. 

It would stand to reason that intermodal growth would have leveled off—and perhaps even reversed—in the face of such service reductions. It did not. In the last two decades of the 20th century, railroads transformed intermodal from an interesting curiosity of a freight market to a core pillar of American railroad traffic: by the year 2000, about 35 percent of all railroad loadings were intermodal. The source of this growth? The tide of import-laden containers sweeping eastwards across the American continent, providing something of a deus ex machina for growth during a time of network shrinkage. 

Maritime container traffic is something of an ideal for railroads. Funneling nontrivial fractions of domestic consumption through a half-dozen ports creates freight flows of incredible density and geographic simplicity; attaching those flows to sluggish container ships blunted their sensitivity to speed and service. To capture this traffic, railroads simply had to provide terminals, competitive freight rates, and efficient equipment. Railroads made those investments eagerly, and then rapidly reaped the benefits. In 1988, North America’s railroads moved about 5.5 million intermodal units, about 2 million of which were containers moving to or from ports. 18 years later, in 2006, they moved 14.2 million units of intermodal traffic, about 8.5 million of which were international boxes. Though the 2010s brought stronger domestic traffic growth, intermodal remains dependent on trade: in 2024, half of all intermodal traffic was international boxes, with a sizable fraction of the remaining domestic traffic consisting of the repackaged contents of international containers.

The incoherence so long permitted by the golden handcuffs of international traffic has begun to show itself for what it is over the last 15 years. In 2008, railroads gained pricing power for the first time in decades; through the 2010s, investors pressured railroads to structure their strategies around this newfound leverage. Branded as “Precision Scheduled Railroading,” the industry spent the 2010s building its long history of self-destructive productivity management to implement rate hikes, infrastructure rationalization, and operational simplification. 

Intermodal traffic rapidly fell victim to these new principles of management. However important it might have been to the industry’s growth over the previous forty years, financial analysts lambasted intermodal for its low(er) profit margins and high sensitivity to service quality. Railroad fell in line quickly, and by 2017, they had begun slashing their intermodal networks. In some cases, they cut lower-density services to simplify operations; in others, they saddled flagship intermodal trains with other traffic to help meet lofty productivity targets. The impacts of these cuts were real and rapid. Comparing intermodal schedule data from each year shows that between 2014 and 2023, Norfolk Southern—one of two major railroads in the Eastern United States—cut over a quarter of the metro area-to-metro area lanes they previously offered. Of the remaining lanes, over half saw slower service versus just 8 percent with faster service.

This strategy worked, for better and worse. Railroad profits and share prices surged while traffic fell. Once the growth engine of the industry, intermodal volumes stagnated from 2017 onwards, falling far below the growth rate of consumer goods spending. Though carriers have made recent noises (backed, to varying degrees, by tangible actions) about a pivot to growth after their decade of austerity, the harsh reality of the industry is that its footprint in the American freight market is at a low.

Where We Go From Here

There was a great irony in the 2010s’ intermodal cuts. Lambasted as they were as exemplars of financial markets’ ever-increasing interest in capital discipline and short-term returns, they nevertheless mirrored the exact sorts of intervention that (supposedly) made intermodal viable in the first place. What were the 1970s’ reforms to the intermodal network if not a series of cuts to lower-density and higher-complexity services? Rather than seeing PSR’s impacts as a shocking reversal of intermodal’s fortunes, it is perhaps more appropriate to see it as a predictable consequence of intermodal’s current strategic alignment. If railroads stake their fortunes solely on their ability to fit intermodal traffic into a productivity-oriented model of railroad growth, it is a small wonder that investors steeped in railroads’ productivity culture do not ask them to crank up the dial (as it were) even more.

Putting these pieces together, you find that the railroad industry’s efforts to gain intermodal traffic have led into a strategic dead end. Any market cost railroads might have suffered due to their 1970s decision to run a low-complexity intermodal network was entirely masked by the rapid growth of international container traffic, whose spatial patterns matched such a network perfectly. Rewarded in their decision to choose this path, railroads have continued to walk it—despite the contradictions inherent to it highlighted by the battles over PSR and falling growth rates. To be sure, thinning the list of intermodal terminals was necessary to pull rail intermodal traffic out of the red, but since then this form of improvement has overwhelmed alternate approaches to the network to become something of a dogma. If railroads want to achieve anything more than incremental volume growth (itself an open question) they will need a new approach.

In the mid-1970s, the Federal Railroad Administration released a study on intermodal service. It began by striking many of the same notes railroads did at that time, discussing the fact that intermodal’s growth had stagnated and that “carriers have raised serious questions about the profitability of the traffic.” And yet, in the face of these challenges, the study did not turn to long unit trains as the salve for intermodal’s problems. Rather, it proposed a radically different approach to the network. Rather than there being several railroads competing for traffic in each city pair (reducing the traffic density available to any one operator), there would be a single national intermodal carrier which would run dedicated intermodal trains between 130 cities across the US. These trains, designed to strike a balance between productive efficiency, equipment utilization, and market competitiveness, would be shorter and faster than the norm—but would allow the railroads to win significant market share from trucks with their flexibility and their ability to effectively serve smaller and shorter-haul markets. 

Unsurprisingly, these proposals went nowhere. Centralizing intermodal marketing ran against the industry’s competitive and fragmented structure, and does not appear to have garnered any serious consideration based on a review of industry press. Railroads did experiment with short and fast intermodal trains, but their success was mixed and the industry’s commitment to the concept was limited. Early attempts at implementing such schemes on long-haul routes were unsuccessful, as were those which sought to pilot alternative intermodal technologies. Efforts to compete with trucks in short-haul markets with short and fast trains met with sometimes-considerable success, but generally were watered down with time as track capacity constraints grew following line rationalizations and traffic growth. Ironically, in at least one case, the end of short/fast intermodal service came when the success of these trains at growing traffic forced volumes above what smaller trains could handle. The lesson of this example evidently was not learned. Save for some elements of the internal structure of intermodal marketing departments, the model adopted by railroads for intermodal was quite nearly the opposite of that which the study suggested.

In today’s environment, suggesting such a radical departure from current practices would do little more than discredit the speaker. The 2020s are not a period of railroad industry crisis in any way similar to the 1970s. And indeed, even if this were not the case, railroads have made extensive investments in terminals, yards, passing sidings, signal systems and more that assume long freight trains. This is to say, the inscription of current operating practices in this infrastructure would make a transition to the 40-car freights envisaged by the FRA study either extremely costly or incredibly inefficient.

Nevertheless, the fundamental insights of that old study remain relevant today. Railroads brag to their shareholders about the length of their trains, their pricing power, and (sometimes) their on-time performance. But they do not dwell extensively on their intermodal product’s overall return on investment, or indeed whether the tradeoffs they make in the name of train-level efficiency actually make economic sense. Perhaps the keenest observations of the 1970s report were on this very topic: in arguing that train lengths should be determined by “market size and service requirements; operating speeds; and the number of intermediate stops,” it proposed a radically different approach to network planning and intermodal growth than the one taken today. It is altogether unlikely that we will become a country of short and fast freight trains, but allowing the stranglehold of the 10,000 foot freight to loosen slightly could pay dividends. A railroad network in which a 4,000 foot train is acceptable is one in which intermodal service might reach many more markets, and is one where the quality of intermodal service itself might improve. 

Conclusion: Networks and Industrial Policy

In April 2024, Norfolk Southern announced it would be ending intermodal service on 20 lanes to and from East Coast ports. While the railroad justified its actions by citing past underperformance in those markets, it is difficult not to see this action as a sign of things to come. If current tariffs hold, American international trade volumes are likely to drop precipitously—and with them, railroads’ fortunes. 

Even when faced with a threat of this magnitude, railroad managements are unlikely to admit that their economic philosophies are wrong. For over a century, industry insiders have decried the way railroading’s focus on productive efficiency blinds the sector to revenue and growth consideriations. And, for over a century, the charade has continued, sustained by a sufficiently widespread conviction that productive efficiency is the right way to think about railroading. Until analysts stop asking about operating ratio and train length benchmarks on earnings calls, or executive bonus schemes are restructured to deemphasize productive efficiency, or (perhaps) a crusading railroad leadership team manages to somehow prove the incoherence of this model, we are stuck with the problematic incentives of today. There simply is not much room for experimentation or deviation from the mean in an oligopolic industry dominated by publicly held (as in, by shareholders) companies. 

In the last third of the twentieth century, a scholar named Thomas Hughes wrote a series of essays and books that proposed the concept of “technological momentum” as a heuristic for understanding the dynamics of technological change. In these works, he argued that large technological systems tend to experience periods of changefulness in their youth (or following a crisis), followed by those of stasis. In the former, society tends to act upon technology; in the latter, technology tends to constrain social choice through the inertia of scale and status. Today’s railroads are, by all measures, large technological systems that have acquired momentum—and thus, relative stasis. They do not face profitability and maintenance challenges akin to those of the 1970s, nor is there much internal dissonance within the industry about the trajectory of technological standards and practices as there was in the closing third of the 19th century. The question—whose reach extends far beyond the realm of intermodal marketing—then becomes: what might open the door to change?

There seem to be two possible paths towards a new moment of flexibility. The first is technology. A series of start-ups and signal equipment manufacturers are currently attempting to develop autonomous trains. These projects come in many forms: some hew closely to the existing locomotive-and-train model, while others utilize self-powered freight cars which can behave either like a train or a rail-based truck. Deployment of any of these technologies would radically alter the economics of rail freight: driverless locomotives would reduce the fixed cost of running a train, while autonomous freight cars would allow complexity to be managed without the time- and cost-intensive process of switching freight cars. The near-term promise of these technologies appears murky, however. While some of these projects have made their into limited on-rail testing, none has made it beyond a fairly controlled pilot environment. Whether it be concerns around these technologies’ safety, challenges fitting autonomous vehicles into an infrastructural system with significant amounts of manually operated equipment, or simply the manifold institutional challenges of gaining railroad, shipper, and labor agreement about new standard types of railroad equipment, it does not appear that autonomous trains are simply “around the corner.”

The other path lies through the government. For a variety of reasons—ranging from climate change to economic efficiency to highway decongestion—it is increasingly well-understood that increasing the share of freight that travels by rail is in the public interest. However, despite the clear relevance of freight rail to broader public policy goals, the government’s role in freight rail discussions has been circumscribed since the industry’s late-twentieth century crisis. Agencies like the FRA have contributed relatively little to strategic discussions in freight railroading since the 1970s, even as a motley assortment of state and local actors have funded wide-ranging improvement efforts in the name of road-to-rail mode shift. Ironically, many of the most significant investments made through such programs have been around intermodal, whether they be portside intermodal infrastructure, clearance projects for container trains, or otherwise. At the moment, change to this reality is unlikely: even if the nation’s political trajectory was to reverse tomorrow, the continuing evisceration of regulatory bodies’ institutional capacity would make intensive engagement with the railroad industry difficult. However, in the medium-to-long run, it may be time for state to step back into the fray. Irrespective of where current conversations around regulation, mergers, or state ownership land, reasserting even a loose strategic framework for freight rail growth could help shift the problematic incentive structure in which carriers find themselves trapped. Whether through further support for capital investment, nationally-coordinated planning for improvements in corridors with high intermodal growth potential, the reintroduction of some basic intermodal service regulations, or even just an explicit normative vision for the role of freight rail in the American economy, government has several levers at its disposal to improve intermodal service. 

As always, the opinions represented above are mine and mine alone. They do not represent the positions of my employer, or any other organization of which I am/have been a part.

Special thanks to Sandy Johnston for help editing this post!

For the last century, the central question of American transit planning has been how best to serve our suburbs. As people moved out of urban cores, bought cars, and began living decentralized lifestyles, the transit industry has struggled to compete. To a great extent, this history has been one driven by factors outside the direct control of agencies: scarce funding, low population density, cheap gas, racism and a litany of veto points have complicated nearly every effort to get Americans off the road. And yet, across this century of decline and struggle, the transit industry has, at times, presented remarkably coherent answers to the suburban ridership question. As it so happens, one of these moments surrounded the origins of our national obsession with light rail.

“Light rail” is a famously ill-defined term. Ranging in form from overgrown streetcar routes to commuter trains run with streetcar equipment, light rail has been the cornerstone of our transit investment policy for the last 40 years. Seen as a cost-effective means of providing high-capacity transit for American cities, light rail now makes up a majority of fixed guideway (i.e. not an on-street bus) route miles in the United States, and for better or worse has become the assumed default when planning new rail networks.

America’s light rail networks encapsulate the challenges of planning transit in a suburban nation. While some cities have successfully leveraged their light rail networks to support significant overall increases in transit ridership, others have built extensive systems that get little use. Despite its ubiquity, light rail riders comprise only 5% of transit ridership. The classic critiques of American planning practices (correctly) link this underperformance to these rail lines’ planning. All too many American systems run on highway or railroad alignments that miss major activity centers, are surrounded by low-intensity land uses, serve weak downtowns, and exist in regions where highway capacity remains abundant. Many, too, are extremely slow, running on sinuous alignments or on streets without any sort of transit priority. 

These challenges are often mediated and aggravated by the fact that our light rail systems are often surrounded by weak and poorly integrated bus networks. All too often, riders arriving at one of this country’s poorly located rail stations have no options to reach their destinations aside from driving. The great irony of this reality? America’s love affair with light rail can be traced to a moment when rail was seen as a natural extension of changing bus networks. Rather than being a supposed panacea to America’s suburban mobility woes, we once took inspiration from international examples, and built rail networks to multiply the value of well-organized bus systems. To grow transit ridership, it is to that planning philosophy that we must return.

This post owes an immense intellectual debt to Gregory Thompson and Jeffrey Brown. As those who click on the links in this post may notice, their and their collaborators’ work on the history and shortcomings of American suburban rail planning was instrumental to the making of this argument.  

Looking Out, and Abroad

The 1970s were a difficult time for American transit. On the one hand, the industry seemed to be growing. The gas crisis had produced ridership gains, while funding increases and the opening of new urban rail systems had expanded the reach of transit service. On the other, the industry’s fiscal health was failing. American transit had been profitable until 1962. While net deficits had been small through the 1960s, they grew rapidly in the 1970s as inflation took its toll. This trend both forced service cuts, and put operators at the mercy of fickle local and national subsidy politics.

In the face of these competing pressures, transit agencies looked abroad for answers. To survive in their fast-changing environment, they would need ways of serving sprawled regions within their limited budgets. This search led the industry to a new planning regime and a new technology: timed transfers and light rail.

The first of these two innovations to arrive on American soil was the timed transfer. In the 1960s, most cities’ transit networks closely matched the layout of turn-of-the-century streetcar systems. In practical terms, this meant that midcentury bus systems were radial, which is to say they were focused on bringing people to and from downtowns. In the era of urban decentralization, these networks made service unusable for many would-be users who wanted to travel crosstown in the suburbs—and for the increasing number of transit riders who commuted from urban homes to suburban jobs located far from suburbs’ main roads.

Unsurprisingly, the desire to extend more comprehensive bus service to suburban corridors often ran up against the realities of suburban travel demand. Not only did the generally lower density of development spread riders over a wider area than in urban cores, but the decentralized commercial structure of suburbs made travel patterns significantly more complex and multidirectional. This left agencies in a bind. To minimize the burden of transferring, which travel in the suburbs’ decentralized environment would often require, planners wanted to run high-frequency routes. But frequency had a cost, and the low overall density of suburbia—to say nothing of suburbs’ high rates of car ownership and generally car-oriented land use—provided few guarantees that those investments in service would be paid back in fares. While some cities successfully deployed high frequency service in suburbia, more often agencies balked at the task: providing fast, attractive, and highly flexible transit service in an environment with weaker demand seemed impossible. Enter here the “timed transfer network.”

The basic concept of timed transfers was simple: planners would schedule bus routes running every half-hour or hour to converge at the same time on some central location—usually a mall or an office park. Buses would wait for a few minutes at these hubs, allowing passengers an opportunity to transfer between routes, and then would proceed on their way to the next focal point, or into the suburban periphery. When executed well, this strategy mitigated the frequency-transfer penalty tradeoff, providing flexibility without the costs of running high-frequency routes. 

The proliferation of timed transfer networks across North America began in Edmonton. In 1962, the city annexed a neighboring suburb, and hired a Dutch transit planner to help design new transit routes to serve it. This planner, one John Bakker, ended up recommending a much wider-ranging set of transit expansion and reorganization plans, including vastly expanded use of timed transfers. Historically, American cities had used timed transfers sparingly, usually synchronizing buses at a single downtown transit hub. Edmonton’s innovation—one inspired by European experience—was to schedule several transfer points throughout the region, allowing easy multidirectional travel throughout the urban periphery. As the city progressively rolled out this new network design, ridership responded. After falling before through the 1960s, Edmonton residents’ per capita use of transit stabilized, and by the 1970s was growing. 

Seeing Edmonton’s success, other cities also began to experiment with multifocal timed transfer networks. Throughout the 1970s, the Urban Mass Transit Administration (the predecessor of today’s FTA) provided funding for transit improvement research and experimentation with new service designs. Timed transfers were one of the beneficiaries of the agency’s largesse, as the success of a few early adopters generated broader industry interest. With these examples in hand, adoption spread: by the early 1980s, dozens of bus networks had been redesigned with suburban timed transfer hubs. While some implementations were much more successful than others—the finer details of running time calibration and route design mattered a lot—the turn away from radial route structures successfully began transit’s process of suburban adaptation. Whether in Portland, Vancouver, or Nassau County, timed transfer networks promised a better future for suburban transit. 

Light rail made it way across the Atlantic a few years after timed transfers. For decades, it had been clear to transit planners that having some sort of fixed-guideway network was essential for transit’s success in the age of highways. Without higher-speed services running independent of traffic interference, transit would always be slower than driving, and would struggle to retain–much less gain–riders. To this end, the 1960s had seen the design of several new suburb-oriented subway systems in the US, including San Francisco’s BART and Washington’s Metro. While these new heavy rail networks would eventually meet with immense success, they were extremely expensive to build. Given the scarcity of transit expansion funding and the pressures of austerity and inflation, transit engineers began seeking cheaper alternatives.

Here, again, European experience seemed to hold relevant lessons. As the continent rebuilt its cities in the 1940s and ‘50s, many transit agencies faced the same challenge of managing decentralization in a constrained funding environment. As in the US, some expanded subway systems, but many more upgraded their legacy tram networks (which by and large had not been eliminated as quickly as their American counterparts) with larger vehicles, better infrastructure, and small sections of tunneled or elevated rights-of-way to bypass major traffic bottlenecks. In so doing, those European agencies managed to avoid major capital expenditure on subway construction while ensuring their cities’ transit networks were still competitive in the postwar world.

To many American planners, these “light rail” networks appeared to present a cost-effective solution to the transit expansion problem. Not only were they cheaper than building subways, but their use of trams dovetailed well with the 1970s’ aesthetic and political focus on human-scale urbanism; streetcars, after all, had fed most American cities’ once-walkable downtowns. As with timed transfers, the federal transit bureaucracy played a pivotal role establishing the technology as an option on the national stage. Throughout the 1970s, federal transit officials convened conferences and commissioned reports on light rail to help disseminate knowledge of its use. These efforts were remarkably successful: within ten years of the publication of the first DOT report on light rail technology, one Canadian and one American city had opened new lines, while three others had projects under construction. 

The First Networks

It is telling of the tight linkage between high-quality suburban bus networks and the advent of light rail service that the first of these “new age” light rail networks was built in Edmonton. The same transit official who had drawn up the plans for the city’s timed transfer network had also put together a long-run plan for the city’s mass transit system. In this vision, rail service would replace buses on routes connecting timed transfer hubs to the city’s downtown core, providing high-capacity, high-speed service for those trips. As agency staff laid out the city’s new bus system, they had been careful to hew closely to that long-run plan, with transfer hubs strategically placed along rail corridors. When the windfall from the 1970s oil crisis landed in Alberta’s coffers, Edmonton’s planners leaped at the opportunity to build a starter rail line—and succeeded, beginning the light rail era.

Without any additional context, Edmonton’s first light rail line would appear to illustrate all that is wrong with North American light rail planning. To ensure easy and cheap construction, the new line followed an old freight rail corridor leading northeast out of the small city’s downtown. While stopped at an expo center and a stadium, there was little else of note along its route. The rail corridor had no history of passenger service, so it abutted industrial tracts, the backs of low-density residential neighborhoods, and an assortment of highway interchanges. Indeed, the only structures approximating a high-rise along the line were the silos and machinery of flour mills and factories. 

And yet, the line was a massive success. When it opened, the sheer volume of bus connections filled the trains up rapidly: 62% of rail users in 1978 reached stations on a bus. Despite the fact that it missed many key suburban activity centers, the new route managed to—literally—redraw the map of transit demand in Edmonton. At opening, the line saw 18,000 riders a day. After a short extension further into downtown, ridership stood at 36,000. Analysis after the opening of the light rail line found that transit patronage had risen in total in the sectors of the city served by the new line, indicative of the route’s capacity to speed trips and more tightly integrate the city’s transit network.

San Diego was next to build light rail. The city had studied rail transit in the early 1970s, but had found the cost of subway construction to be prohibitive. To the region’s frustrated planners, light rail appeared to provide a solution, as its low construction costs meant that the agency could start a new service without major investment. As plans evolved, they took strikingly similar form to Edmonton’s. This was not a coincidence: San Diego actually hired planners who had worked in Edmonton to shape their rail investment strategies, and unsurprisingly wound up with a network plan highly attuned to the need for bus integration. In San Diego’s planning environment, the integration task was significantly more complex than in Edmonton. Transit service in San Diego was operated by seven different agencies. Rail integration planning would have to navigate the differing priorities, incentives, and funding levels of those operators—and would have to then maintain those relationships in the decades to come if it was to last. Difficult as they might have been, the challenges of a fragmented planning environment were handled well: the regional planning coordination body implemented a transfer and transit pass system that would ease transfers between buses and trains operated by different agencies, and helped redesign cities’ bus networks to better connect with the new rail line. 

When completed in 1981, San Diego’s rail line was a near-instant success, surpassing ridership estimates after the agency surmounted initial reliability challenges. Critically, rail investment achieved its stated objective of catalyzing a broader-based increase in suburban transit ridership. Illustrative of that linkage were the experiences of two southern San Diego suburbs, Chula Vista and National City. In the ten years following the opening of the rail route, the transit operator in Chula Vista saw its ridership jump by 158 percent against an overall service increase of only 42 percent. In neighboring National City, ridership grew 178 percent against a service increase of 25 percent. These systems’ growth alongside the rail network was not coincidental: both agencies chose to pursue close service integration with the new rail line, restructuring routes and schedules to ensure access to the train. And indeed, within two years of the LRT line’s opening, both systems were drawing about 30 percent of their riders from the LRT. As San Diego opened more LRT lines into the 1990s, this pattern repeated, with coordinated feeder bus services helping build light rail ridership and transit use beyond the rails. 

Unlearning Complexity

Together, Edmonton and San Diego proved the viability of light rail: the cities had built effective networks at much less cost than new subways. Others rapidly followed, and by 1990, six American, two Canadian, and one Mexican cities had opened new systems. But in these heady days of expansion, light rail’s entanglement with suburban bus networks weakened. Reagan-era budget cuts and changing agency priorities reduced funding for service experiments and the overall industry focus on network design. Through the 1990s, the careful planning required to ensure good connections between trains and buses in the suburbs gave way to a less cohesive regime in which transfers were often onerous, if available at all.

Emblematic of the shift in philosophy was Sacramento. The city opened its first light rail line to great fanfare in 1987, only the 4th city to complete a new line in the US. Inspired by San Diego’s experience, Sacramento’s initial light rail plan focused on connecting timed-transfer hubs with the city’s downtown to make a multidestinational regional network. As in San Diego and Edmonton, the city’s initial light rail line was a success. Between 1987 and 1992, transit ridership in Sacramento grew by 47% against a total service increase of only 28%. And as in those other cities, those ridership gains were not limited to the new trains. Despite the new rail system diverting several former bus riders, bus ridership in Sacramento actually grew 4% between those years, indicative of the degree to which bus and rail functioned as an integrated unit in the city.

Unfortunately, in the 1990s and 2000s, Sacramento’s transit planning philosophy shifted. Noting high transit ridership on radial routes serving the city’s core, planners decided to deemphasize their network’s orientation towards rail transfers in the suburbs in favor of direct, radial bus routes to downtown. By the mid-2000s, the region’s bus network had lost a considerable amount of its rail-feeding crosstown service. On remaining routes, the system’s historic emphasis on timed connections weakened. Planners began shifting the 15-30-60 minute frequency pattern that ensured timed connections with trains to a more random assortment of service levels. Busy, every-15-minute routes often were cut to 20-minute headways during midday hours, while many former 30-minute routes were restructured with odd frequencies (e.g. a bus every 34 minutes) to save money and simplify scheduling. 

While these changes had roots in real budgetary constraints and ridership observations, they had a corrosive effect on the system’s health and planning. The weakened links between rail and bus, and indeed between previously well-connected bus routes, made it harder to use transit. After peaking in 1999, the productivity (measured in riders per vehicle-mile) of Sacramento’s bus service began to decline. By 2005, each mile of revenue bus service was carrying 18% fewer passengers than it had in 1999. The rail network fared little better. When the agency began opening extensions to their rail network in the early aughts, the shift in planning philosophy meant that they came with few of the bus network improvements that had made the original line so successful. As a result, these new extensions never lived up to their ridership potential, dragging the overall productivity of the region’s transit system downwards. Save for 2009, when a spike in gas prices drove people onto transit across the country, Sacramento’s rail system has seen fewer boardings per mile or hour of revenue service in every year since the new routes opened. 

Sacramento was hardly the only city to make the pivot away from coordination. As other agencies began building light rail networks in a planning environment less attuned to the importance of integration, backsliding proliferated. Through the 1990s and 2000s, several new rail corridors opened with limited bus integrations—sometimes due to a lack of emphasis on connectivity in the planning process, and other times because rail was being built into regions where bus service simply did not exist. What’s more, some cities began to eschew the “regional overlay” model for rail construction entirely. Rather than building LRT lines that would move riders quickly between far-flung parts of sprawling regions, some agencies chose to build lines that behaved more like enhanced buses, running at lower speeds and on shorter lines. To be sure, some of these new projects and approaches were successful. Houston’s small light rail network, for example, sees 43,000 trips a day despite being slower than the average bus route in the city, and only extending a few miles from the city’s downtown. But in this continent of sprawling cities, these slower or less-well integrated networks can only be so effective. Providing a comprehensive alternative to auto travel means offering transit whose speed and flexibility can compete with highways.

Conclusion

In 2006, a pair of researchers (whose work heavily inspired this post) were tasked with identifying service design decisions that seemed to increase rail transit ridership. After eleven US cities with rail networks of varying scales, they came to the conclusion that successful rail transit systems are those which, in their words:

1. Articulate a clear, multidestination vision for regional transit; 
2. Rely on rail transit as the system’s backbone; 
3. Recognize the importance of the non-CBD travel market; 
4. Encourage the use of transfers to reach a wider array of destinations; 
5. Recognize that rail transit alone is not enough to guarantee success; and 
6. Recognize the importance of serving regional destinations

Though prolific, those researchers have hardly been the only ones to underscore the importance of network connectivity for rail transit’s success. Whether they are seeking to explain variance in stop-level ridership or the use trends of entire transit systems, researchers time and again find themselves concluding that connectivity drives ridership. In a country of sprawling cities, rail (or even rapid bus service) will not be able to reach every corner of every suburb. But designing networks that can move people quickly between different parts of the broader region in concert with well-routed, well-scheduled, and (hopefully) frequent buses has been shown time and again to provide success. 

Make no mistake: the future of American rail planning need not be a return to the philosophies undergirding the first light rail networks. Timed transfers are not always appropriate or feasible, nor are region-spanning light rail systems always a more effective or equitable investment than alternate modes or network designs. Too many transit agencies have prioritized speed and suburban access over all else, leading to alignments that entirely bypass dense urban neighborhoods to the benefit of a smaller base of wealthier and whiter base of suburban riders. Conversely, some Canadian agencies have had great success attracting ridership by running high-frequency bus services in the suburbs, relying on short wait times rather than schedule coordination to facilitate connections.

Rather than being a prescriptive recipe for the future of transit service design, the true lesson of light rail’s history has to do with the nature of transit systems. The core insight of San Diego and Edmonton’s rail planning was that bus and rail networks must be planned together and then closely coordinated. Put differently, the two cities’ success should underscore to us that transit is a business that lives and dies on its ability to build networks. No single rail line or frequent bus service can ever unlock broad-based transit ridership gains; the structure of urban travel demand is almost never linear. Connectivity is key, and as American cities confront budget crises, changing travel demand patterns, and the overwhelming pressure to do something in the face of transportation’s contributions to climate change, it is past time for us to fall back in love with network planning. 

All opinions expressed in this post are my own.

In February 2023, two thousand feet of Chicago’s South Normal Boulevard disappeared. The street’s end came at the hands of Norfolk Southern’s 47th Street intermodal terminal, a facility which moves thousands of containers a year between trains and trucks. For over a decade, the railroad had been trying to expand the yard through a over a residential neighborhood on Chicago’s South Side. When they finally prevailed over vigorous opposition in 2022, they closed another chapter in the complex and intertwined histories of freight and community in Chicago. But lurking behind those threads was another story: 47th Street Yard is a facility which need not have grown.

To understand Englewood’s rail terminal, you must focus on the trucks leaving it. Follow them, and you might end up at a nearby warehouse or factory. But unlike almost anywhere else in the United States, those trucks’ trails may also lead to another railroad’s yard. A significant portion of the containers unloaded 47th Street are in Chicago not to deliver goods, but to transfer between different railroads as they make journeys across the continent. This use of trucks to bridge gaps between rail carriers is, on its face, unexpected. Chicago is a city famous for its lattice of tracks which link American railroads together. It begs the question: why does this practice exist?

At its core, this story is one of fragmentation. Unlike in much of the rest of the world, ownership and operation of the United States’ rail network is decentralized. Rather than having a national railroad, or centrally-managed tracks open to use by any company, our railroads are private and vertically integrated—they own, maintain, and run their own lines, and compete largely on the strengths and weaknesses of their networks.

The resulting rail system is today dominated by six companies: two Canadian carriers whose lines reach into the American Midwest, two railroads serving the Eastern United States, and two serving the West. Yet neat as those divisions might be, flows of freight generally do not respect such contrived boundaries. Shippers want to move billions of tons between the east and the west, and to do that on rails, they must use multiple carriers. Railroads directly interchange all their carload freight—the boxcars, tank cars, flat cars and hopper cars that carry lower-value bulk and manufactured goods around the country. One railroad brings those cars to another’s yard, and they move onwards on their journey across these invisible lines.

For intermodal freight, matters are more complicated. Over half of rail freight today is comprised of containers and trailers, mostly carrying goods for shipping lines and more service-sensitive domestic shippers (like Amazon and Walmart). Rather than being interchanged with other railroads directly, this traffic often makes the jump across railroading’s dividing lines on the back of a truck. Almost all of those “rubber tire” or “crosstown” interchanges happen within the metropolitan limits of Chicago, the hub of the American rail network. In 2002, the last time anyone publicly published statistics on the scale of this problem, over one million trucks each year (about 20 or 25 percent of the total intermodal volume in the region) made trips across the city hauling railroad-to-railroad freight—each one of them adding to Chicago’s air quality, pedestrian safety, and traffic congestion problems. Since then, intermodal volumes have grown by 44 percent.

In the dogged persistence of Chicago’s crosstown trucking—and, indeed, in the history of 47th Street Yard itself—lies a story of American railroading’s ambivalence. Neither a fully integrated network nor a series of independent, competing companies, the industry’s history has been shaped by its leaders’ changing approaches to cooperation within their fragmented reality. Waxing and waning through time, and forever constrained by the innate path dependency imposed by railroad infrastructure, the trajectory of railroads’ efforts to improve the flow of freight through the Chicago gateway highlights not just the limits of our balkanized railroad network, but also our national disengagement from freight policy. As railroads confront important decisions about the future direction of the industry, and as policymakers consider transport policy options for the age of climate change, the persistence of crosstown trucking stands as a local environmental problem and national supply chain bottleneck of immense significance. We need more active measures to fix it.

Density

If I had to pick one, my favorite book would be William Cronon’s Nature’s Metropolis. If you have not read it, you should. It details the evolution of America’s industrial economy in a process that both made Chicago a center of trade and manufacturing and the Great Plains the nation’s agricultural heartland. Spread across its pages are stories about the power of density in early industrial America. Chicago’s economic import emanated from its concentration of markets and factories, each of which functioned on economies of scale and proximity—giant slaughterhouses which could cut beef with maximal efficiency, commodities markets which could facilitate trading between far-flung firms, and so on. Of course, those industries did not appear from nowhere: as Cronon so forcefully argues, Chicago’s growth owed much to the access provided by its railroads.

Railroads’ intertwined history with Chicago’s urban growth is, on some level, unsurprising. Trains are an artefact of density: each of them aggregates dozens if not hundreds of cars of freight or people into a unit that can be moved cheaply over long distances. Add in the fact that railroads have high fixed costs, thanks to all the track and equipment they must maintain whether they run trains or not, and the industry is one defined by the inverse relationship between its traffic volumes and unit costs. In simpler terms, the more traffic a railroad can concentrate on a given line, the less it spends to carry each bit of it. Railroading is thus an industry which thrives on urbanization. Cities’ scale created intense flows of people and goods that slake railroads’ thirst for density.

More than any other city in the United States, Chicago’s history reflects railroads’ tendency to seek and produce concentration. Carriers first converged on Chicago to access Lake Michigan’s ships, and later to reach the city’s collection of markets and factories. The resulting hub-and-spoke rail system in the Midwest afforded Chicago’s manufacturers unprecedented market access, its merchants incredible market power—and its railroads immense profits. As Chicago evolved into the hub of the nation’s railroad system, its railroad companies came to enjoy a new benefit of Chicago access. Just as the city’s commodity exchanges made it easier for merchants and financiers to trade with each other by putting everybody in the same room, Chicago’s concentration of rail carriers made it easier for railroads to hand off traffic to each other by putting all those connections in one place. The city became the endpoint of more than two dozen different railroads’ networks, a clearinghouse for the freight and passengers of an industrializing nation.

Congestion

Unfortunately, however rapid Chicago’s growth might have been, economies of scale and proximity do not work as neatly in practice as they do on paper. As the nation’s network grew, American railroads built thousands of miles of duplicate lines as different companies sought access to the same markets. In Chicago, this tendency reached an apex. Most of Chicago’s twenty major railroads built their own lines into the region, each of which terminated at railroads’ individual yards and a series of six passenger stations ringing the city’s downtown. 47th Street once encapsulated the process of fragmentation: the site that would become Norfolk Southern’s intermodal terminal housed no fewer than four different railroads’ yards.

The irony of this splintered system was that carriers had driven their tracks to Chicago in part so they could link with other railroads. To achieve this, carriers built a warren of connections between each other’s tracks. In the six miles between the Pennsylvania’s yard at 55th Street (their name for their part of 47th’s future site) and Chicago’s Union Station, the railroad’s main line made connections with over a dozen other railroads, allowing the carrier to offer direct interchange service to points across the west. Yet while these links provided much-desired connectivity, they also compounded the costs of fragmentation. A bit less than four miles north of 55th Street, for example, the Pennsy’s main line met three other railroads’ tracks at 21st Street, a crossing that became famous for its sheer complexity. With hundreds of trains a day passing through that and countless other junctions across the region, Chicago’s railroads rapidly became plagued by congestion.

It fell to the crews of Chicago’s “transfer” freights to manage the consequences of this fractured development pattern. Alongside the local crews who delivered cars to the city’s thousands of industries, these transfer crews were Chicago’s sinew. Trains from across the country converged on railroads’ respective Chicago yards, where their cars were sorted for delivery to other carriers. Transfers then moved those cars to those railroads’ yards, from where they could continue their journeys. With so many railroads, so many yards, and such complex infrastructure in Chicago, this system aggravated the city’s congestive chaos. In 1927, Chicago’s railroads ran nearly four times more transfer freights within the region than they did road freight trains in and out of the region. Moving at an average speed of five miles per hour, each of them helped gum up junctions, block streets, and complicate sorting at yards.

But even without congestion, Chicago’s infrastructural organization all but guaranteed delays to interchange freight. Railroad cars are liable to spend half their lives or more sitting in yards awaiting sorting or their next train; railroading is as much the art of building traffic density and moving trains quickly as it is simply getting cars out of terminals. The fragmentation of yards in Chicago turbocharged those delays to freight. At best, freight cars had to wait in a yard twice: once to be passed from their inbound train to an interchange transfer, and again to be handed off from that transfer to their outbound trip. And that was at best. In 1922, the average freight car moving through the region was yarded 2.8 times on its trip through the city. Of the thirty-one hours it took cars to move through the Chicago region in that year, twenty-three were spent sitting in yards.

Reform

As early as the 1880s, dissatisfaction with Chicago’s railroad system had begun to grow. Shippers bemoaned the delays suffered by their freight; railroads lamented the cost of running such a complex network; urban residents organized against the pollution and accidents that followed the trains wherever they went. By the turn of the twentieth century, then, railroads had begun to seek ways around Chicago’s manifold problems. Their efforts had two prongs. First, they built shared yards and crosstown railroads—called “belt lines”—to speed interchanges. And second, they started assembling solid trains of traffic for movement through Chicago onto another railroad without any sorting.

Belt lines offered railroads a compelling proposition: rather than operating a complicated network of transfers through Chicago, they could dispatch their trains to a gargantuan shared yard at the edge of the city that would act as a sort of massive mixing bowl for railroad traffic. In these yards, freight from all different railroads would be consolidated, and rearranged into trains ready for movement to points in receiving carriers’ networks. By consolidating the multiple stops involved in the traditional operating model into a single sorting at a central yard, and leveraging the resulting traffic density to efficiently build trains, belt lines would save railroads and shippers significant amounts of time and money.

Despite the sound logic of belt lines, railroads were initially resistant to using their services. Because the costs of slow and unreliable interchange service were partially externalized among other carriers, car owners and shippers, it was often cheaper for carriers to limit their use of belt lines to their least-used connections, and instead continue running cars slowly through the transfer freight network to avoid paying the belt railroads’ switching charges. As a result, these shared facilities handled only 44 percent of Chicago region interchange freight in the late 1920s, and less by the 1940s.

More impactful, perhaps, were the efforts railroads made to speed freight within their own networks. Around when the region’s belt railroads were completed in the late 1910s, American railroads began to experiment with ways of organizing trains that would reduce the need to sort traffic in the first place. For years, carriers had moved freight with little planning: yardmasters grouped cars headed in roughly similar directions into trains, and sent them to the next yard down the line. But as railroad reformers and managers sought to reduce delays and congestion in the rail network after World War I, they began investigating centralized planning of car movements. Its logic was simple and compelling. Rather than dispatching freight ad-hoc, railroads would study traffic flows across their networks, devise ways of grouping cars heading in similar directions into “blocks,” and then assemble those blocks into trains. By pre-sorting freight for distant points, shipments could bypass intermediate yards. This saved railroads time and money, while at the same time improving chronically-poor service reliability.

Initially gamed out within the confines of individual railroads’ networks, carriers slowly realized that they could speed interchanges by applying this philosophy to Chicago. Rather than sorting traffic in the city’s cramped yards, trains would arrive in the region with their cars already “blocked” for delivery to other carriers—or, better yet, for specific yards in that next carrier’s network. Doing this would require railroads to trust each other and coordinate, but slowly, this sort of operation spread. Before the 1920s, the Pennsylvania had relied on their 55th Street yard in Chicago to sort traffic for the region. When the Pennsylvania Railroad opened a new yard in Crestline, Ohio in 1928, this began to change. Immediately, they began pre-classifying blocks of cars for direct delivery to the city’s belt lines. By 1941, Crestline was also blocking traffic for direct delivery to the Chicago & Northwestern and Milwaukee Roads, two important western connections in Chicago. And in 1974, after the Pennsylvania had been merged into the ill-fated Penn Central, new, supersized regional classification yards as far east as Pittsburgh were building blocks for direct Chicago interchange, while the railroad’s modern sorting yard in northern Indiana classified freight for points as far west as California. Railroads began running pre-blocked trains straight through the city, linking regional classification yards in Chicago’s hinterland with high-speed, high-reliability freight service. In this world of extensive and careful pre-sorting, yards like 55th Street assumed a more minor role, primarily serving the city’s fast-dwindling industrial base.

By the 1980s, then, railroads had begun to get Chicago right. The larger and more modern regional yards they built around the Chicago gateway allowed more extensive pre-blocking and more frequent run-through trains. At the same time, railroads finally increased their use of belt railways, whose yards handled interchange traffic that could not move in pre-classified blocks. In conjunction with Chicago’s deindustrialization, traffic losses to trucks, and the increasing use of single-commodity and single-shipper “unit trains” for moving bulk goods by the trainload without any en-route sorting, these changing operating practices allowed the retirement of several of Chicago’s older urban terminals. Freight moved through the city faster—but its changing patterns left weed-choked tangles of steel in its wake.

Intermodal

For some of Chicago’s urban railroad yards, abandonment really was the end of their story. Many a Chicago strip mall is built on land which not forty years earlier would have been a morass of freight cars waiting for a train east or west. But for many others, their retirement as classification yards merely closed one chapter in their stories. The 1960s and ‘70s were the dawn of the “intermodal age,” when railroads began putting truck trailers and shipping containers on their trains. In this convergence of old infrastructure and new shipping technology lay the roots of South Normal Avenue’s transformation into a parking lot.

Intermodal railroading as we know it got started in the 1950s. As industries decentralized away from railroads, and as railroads failed to improve their slow and unreliable carload services for the demands of contemporary shippers, carriers turned to the roads. The thought was simple: customers would bring trailers to terminals in towns and cities across a railroad’s network, from which railroads would take the trailers—loaded on flat cars—to another terminal near their destination. By making it easier to use rail without a line directly to each shippers’ plant, intermodal allowed railroads to serve decentralizing factories; by reducing the complicated sorting and gathering steps inherent in carload railroading, it made railroad service better. When railroads started offering intermodal options, shippers flocked to them, seemingly validating these theories. Between 1954 and 1964, trailer and container volumes on American railroads grew from nothing to 890,000 annual loads. 95 percent of shippers in a 1959 survey rated intermodal as being superior to traditional carload rail, and 85 percent responded that they had shifted freight back from trucks to the rails thanks to the new services. Railroads had a winner—or so they thought.

What railroad accountants realized in the late 1960s was that intermodal profit margins were low. Intermodal services had to be priced affordably enough so that they could absorb the added costs of trucking loads from shippers’ facilities to a railroad terminal (and vice versa at the other end). At the same time, intermodal rates had to cover the cost of loading shipments onto trains, and running the fast, reliable service demanded by intermodal’s target shippers. When railroads added up those expenses, it rapidly became clear that their growing intermodal business was often actually a financial drain.

In the eyes of most railroad managers, intermodal’s complexity lay at the heart of its profitability woes. During intermodal’s early years, railroads sought to offer comprehensive services across their networks. They carpeted the map with thousands of intermodal terminals, and offered services linking all of them to each other. While the extensiveness of these early networks kept trucking distances to rail terminals short, and allowed railroads to compete for nearly all flows of freight, the complexity of these services added to intermodal’s costs. Shipment volumes outside major shipping corridors were light—maybe a few loads a week. To aggregate these small volumes of freight into trains, railroads were forced to extensively sort intermodal traffic, which slowed supposedly-speedy intermodal shipments while adding to their handling costs. Moreover, each of those small intermodal terminals was itself expensive. Larger cities could justify the expense of mechanized cranes and improved terminal layouts; smaller places often could not, especially in that era when American railroading teetered at the edge of financial ruin.

Through the last third of the twentieth century, railroads consequently slashed their intermodal networks. Thousands of smaller terminals closed as railroads consolidated their facilities in major cities and restructured their service offerings around dedicated intermodal trains offering relatively fast service in a tightly limited number of high-density shipping markets. These reforms, in conjunction with new pricing freedoms afforded by deregulation, helped improve profit margins. But simultaneously, they began to lock railroads into a rather inflexible operating model. Intermodal equipment designs congealed around permanently coupled sets of train cars, which reflected an assumption that traffic would move only in large blocks. At the same time, intermodal terminals lost sorting capacity as they expanded over old classification yards, or moved to new terminals at ports and in the suburbs whose designs often skimped on marshalling tracks.

Managing Complexity

As intermodal grew further into the 1990s and 2000s, this system architecture—capable of building large point-to-point trains quickly and efficiently, but poorly suited to handling freight in lighter-volume markets—would become one of the industry’s biggest problems. By 1990, rail volume growth was intermodal traffic growth; the rising tide of containerized foreign imports and time-sensitive domestic freight helped buoy rail volumes despite widespread deindustrialization. But though railroads could make killings hauling traffic in dedicated trains between New York and Chicago, or Los Angeles and Texas, growing high-quality intermodal service outside those major city pairs required reinserting complexity into their plans. So long as railroads remained committed to operating long trains (a philosophy which planners challenged in the ‘70s for these reasons), the volume of freight moving between, say, Cincinnati and Syracuse would never justify its own run. If railroads wanted to handle it, they would have to sort intermodal traffic so they could build intermodal trains with traffic from multiple origins headed to multiple destinations. Railroads thus found themselves caught in a balancing act. To continue intermodal growth, they would need to build out more complex operating plans and more extensive sorting infrastructure, but not so much as to erode intermodal’s already-thin profit margins.

The tension between railroads’ drive for simple networks and shippers’ demands for more extensive intermodal service came to a head when railroads tried to interchange intermodal freight. Negotiating complexity within a single carrier’s network was one thing, but interchanges between railroads brought a new level of intricacy to railroads’ recently-simplified intermodal services. Creating inter-carrier services could mean adding dozens, if not hundreds, of additional origin and destination points to service plans, while also exposing intermodal to railroads’ age-old struggle to coordinate and build for efficient cargo handoffs.

Railroads initially refused to confront those challenges. Into the 1970s, shippers generally had to truck their trailers from one railroad terminal to another when using multiple railroads’ intermodal services. While crosstown truck interchange avoided complexity for railroads, it added significantly to shipping costs. Unloading trucks from trains, moving them to another railyard, and then putting them back onto trains could add hundreds of dollars to freight bills, deterring shippers from the rails. When railroads finally began offering widespread all-rail interchange service in the late 1970s, shippers seemingly had reason to celebrate. With their newfound ability to move a container clear across the country on a train, they shifted thousands of trucks off Chicago’s roads, and swelled transcontinental rail volumes with containers full of Asian imports. Thus, even though direct rail services still left a large portion of interchange traffic on highways, railroads seemed to have created a winner.

If one took a closer look, however, Chicago’s rail intermodal interchanges were tightly constrained by their own history. In Chicago, as in most other cities, railroads had built their intermodal terminals in the footprints of old urban freight yards. The Pennsylvania, for example, replaced their 55th Street classification yard with an intermodal facility in the 1960s as urban traffic volumes fell and intermodal grew. This yard helped the Pennsylvania cheaply expand intermodal capacity while affording them good access to urban industries and highway networks. But when the carrier’s successors began trying to directly hand off intermodal traffic in Chicago, their efforts were constrained by the site’s original use. That railroads like the Pennsylvania had built their new Chicago intermodal yards on the sites of the region’s first generation of rail terminals meant that intermodal traffic was doomed to suffer all the old problems of Chicago interchange. Fragmented, railroad-owned terminals foreclosed centralized sorting and forced the extensive use of transfer freights to hand off intermodal traffic. Meanwhile, railroads’ broader ambivalence about the appropriate level of complexity for intermodal crimped the development of sorting infrastructure, which made building pre-blocked or run-through intermodal freights difficult outside the busiest of markets. Had railroads truly internalized the lessons of their last century, they might have corrected one or more of these deficiencies–but despite the clear relationship between intermodal’s problems and the age-old issues of Chicago region railroading, carriers largely failed to confront the failures inherent in fragmented infrastructure.

Thanks to these constraints, rail shippers were forced to contend with a messy three-tiered system for intermodal interchange. Illustrative of this segmented approach to interline service was a train Conrail ran in the 1990s called TV-10, which carried traffic from 47th Street Yard to over a dozen destinations across the Northeast. Much of the train’s freight came from trains operated by the Santa Fe (predecessor to today’s BNSF) from the west and Texas. The heavy flow of cargo from Los Angeles headed to Massachusetts and northern New Jersey ran through onto TV-10 off of a dedicated Santa Fe train with almost no delay in Chicago; the Santa Fe assembled the train in California so it was pre-blocked for Conrail’s terminals. But loads moving from Los Angeles to smaller eastern cities, or from smaller western cities like Houston or Dallas to the east were not so lucky. Without adequate sorting infrastructure, those cars could not be pre-classified by eastern destination; they arrived in Chicago in blocks that mixed them with other eastbound traffic. Santa Fe transfer crews would deliver those cars to Conrail anywhere from six to sixteen hours before TV-10’s scheduled departure, so Conrail could painstakingly sort them at 47th Street and the railroad’s nearby Ashland Avenue yard for their destinations. And if you were trying to ship from an even smaller city on the Santa Fe—say, Albuquerque—onto Conrail’s network, your only option was crosstown trucking. Santa Fe and Conrail did not wish to bear the costs required to sort such small volumes of intermodal traffic at their constrained yards. Direct intermodal interchange might have been simpler and cheaper than trucking freight across town, but its complexity, sliggishness, and dependence on old infrastructure—one whose costs were felt with particular strength in smaller markets—made it hardly an optimal fix. Crosstown trucking persisted, and as intermodal traffic grew, its growth resumed too.

Modernization?

The resurgence of American rail traffic volumes in the 1980s and 1990s rapidly increased the strain on Chicago’s infrastructure. During those decades, city’s junctions and yards repeatedly devolved into congestive chaos under the strain of merger-wrought shifts in traffic patterns, unexpected traffic peaks, and blizzards. In response to these and the city’s other persistent service problems, regional planners and railroaders spent the early 2000s developing an immense rail infrastructure improvement plan for Chicago. Known as the Chicago Region Environmental and Transportation Efficiency Program, or CREATE, the 2003 proposal consisted of over a hundred investments in flyovers, junction reconfigurations, signal improvements, and capacity expansions designed to improve the flow of trains—both passenger and freight—through the region. Though funding shortfalls have thus far prevented the completion of CREATE’s wide-ranging plans, CREATE has been a success even in its incomplete form. Thanks to its investments and a series of institutional changes to the way railroads coordinate their Chicago-region operations, Chicago’s rail riders and freight shippers enjoy more and faster trains than they did twenty years ago.

But CREATE had one major flaw. While the plan provided an admirable roadmap for relieving train congestion and improving passenger rail service, it fell short of offering a fix for Chicago’s rubber tire interchange problem. Amid its proposals for gargantuan flyovers and rerouted commuter rail routes, CREATE left the fragmentation of Chicago’s intermodal yards untouched. To some extent, this shortcoming was likely just a product of the project’s institutional origins. Being an effort premised on voluntary cooperation between regional governments and freight railroads, major changes to the way railroad firms operate their networks would likely have been difficult to sell to its stakeholders. Nevertheless, without shared intermodal terminals, centralized sorting hubs, or better developed peripheral intermodal classification yards, the future CREATE promised for the Chicago region was one of only partial salvation. The hundreds of carload freight trains and dozens of intermodal transfers that plied the city’s rails would move faster, yes—but the crosstown trucks left to fill in the gaps in the intermodal interchange network would remain on the city’s streets.

In the past five years, the unaddressed fragmentation of Chicago’s intermodal yards has become a locus of rail shippers’ and Chicagoans’ troubles with a changing railroad industry. After the Great Recession, railroad capacity constraints and Wall Street pressures led to a resurgence of productivity-oriented managerial strategies in railroading, a trend which culminated in the late-2010s rollout of cost-cutting operating plans across most of the railroad industry. Theoretically, these new operating plans, billed as “Precision Scheduled Railroading,” should have improved interchange operations by increasing the predictability of train arrivals in Chicago. However, the underlying managerial focus on profit and productivity improvement compromised that goal. Not only have today’s longer trains added to railroading’s long-standing reliability problems, but PSR’s search for productivity improvement came at great expense to intermodal. Railroads redoubled their emphasis on simplicity, as each cut dozens of their service lanes and hundreds of all-rail interchange offerings to achieve their high targets for their overall operating margin without having to make the capital investments required for efficient operations.

One might think that simplification efforts would yield service benefits for the lower-volume lanes which did not get cut—but as of yet, those dividends have largely failed to materialize. Five years after a major round of cuts in 2018, shippers using Union Pacific and CSX to move freight across the country are still presented with uneven service options. It takes 21 hours longer to get a container from Los Angeles to Buffalo than it does to get one to Northern New Jersey, thanks to the additional sorting that the lower-volume Buffalo lane requires. Weakened and shrunken, American rail intermodal volumes remain 7 percent off their 2018 peak, despite surging demand for freight service during and after the pandemic. Productivity-oriented managerial philosophies and ineffective infrastructure have combined to cause rail’s retrenchment in a time of climate crisis; the only winners in this era have been investors.

What We Should Do

To solve Chicago’s interchange problem, railroaders and planners will need to return to the essence of railroading, and indeed Chicago itself: density. Reconciling the clear demand for intermodal service linking smaller markets and railroads’ struggles to manage complicated flows of container traffic will require building interchange and sorting infrastructure capable of efficiently consolidating the manifold flows of intermodal freight that move through Chicago.

There are basically three ways of creating that density. The first is to build shared interchange yards for intermodal traffic. Imitating the logic of belt lines, shared intermodal terminals would allow railroads to consolidate their lighter-volume interchange flows into blocks (or trains) for a belt terminal, where loads could either be re-sorted, or trucked to a different train within the terminal. By eliminating transfer freights and miniaturizing the crosstown transfer process, these terminals would cut travel times and costs. Shared terminals are hardly a new proposal, but the fact that nobody has yet to attempt building one speaks to their complexity. Railroads and policymakers would have to not only not design and fund the project, but also make long-run commitments to the use of such facilities. What’s more, they would have to find somewhere to put the gargantuan terminal, no small challenge in as urbanized a region as Chicago. With the right alignment of actors and land, the next generation of infrastructure sharing could be revolutionary for Chicago—but getting to that point may require decades of effort that we do not have.

The second, and perhaps more feasible, way of fixing Chicago’s bottleneck is by building regional classification yards for modern intermodal networks. Every railroad entering the city already has some sort of intermodal sorting yard they use for swapping blocks of traffic on trains headed into the Windy City’s terminals and interchanges. But except for traffic moving through purpose-built facilities along CSX and Union Pacific’s east-west mainlines in Ohio and rural Illinois, respectively, most sorting takes place in old, constrained terminals. Norfolk Southern sorts their eastbound traffic in a Chicago yard originally laid out for livestock movements in the 1880s; BNSF’s trains from the southwest use an old and small division point yard in Iowa; Union Pacific’s loads from Texas are simply shunted around in a yard on the city’s south side. Along with the limited yard space available at intermodal terminals themselves, these facilities allow for a basic level of classification—but not the sort of quick, flexible, and high-volume type of sorting required to operate an extensive interchange network. If regional policymakers and railroads were to build out a better-equipped set of intermodal sorting yards on their main lines feeding the Chicago gateway, they could more easily and cheaply consolidate small flows of traffic into larger blocks for interchange, or even solid run-through trains. Terminals like 47th Street could stop growing, and the volume of truck traffic on the city’s streets would shrink.

The third—and most fraught—solution is mergers. Fundamentally, the city’s rail coordination problems trace their roots to the North American rail network’s balkanization. Natural as those lines might seem to American railroaders, they need not be permanent. Other cities that suffered from interchange-induced truck traffic—like Cincinnati, once a gateway between Midwestern and Southern carriers—now see none, because mergers have integrated the networks that once met in those cities. Tearing down those geographic walls without aggravating monopolization issues would likely require new sorts of regulatory protection for shippers, or even new structures of rail ownership akin to those seen in Europe and Asia. Nevertheless, mergers would give railroaders an opportunity to assemble intermodal networks across the Chicago gateway, entirely eliminating the interchange problem. In this age of intercontinental supply chains, it may well be that assembling railroad networks of even grander scales than today’s might provide the best route to achieving transportation and environmental policy goals in Chicago and beyond.

Conclusion

These solutions will be too late to save South Normal Avenue, of course. As of 2023, Norfolk Southern’s bulldozers are slowly transforming their slice of Englewood into a parking lot. This is perhaps the project’s final irony: South Normal Avenue is disappearing to create a relief valve for freight’s continued fragmentation, a storage space where containers can wait until their shippers are ready to move them to a loading dock or another intermodal terminal. Parking is undoubtedly important to a well-functioning intermodal yard, but it is fundamentally a salve for unreliable rail service and unpredictable supply chains, rather than a cure for any of freight transport’s woes. In Norfolk Southern’s plan, then, lies a condensed history of railroads’ ambivalence towards the myriad coordination problems that surround their infrastructure. Each individual carrier does their best to work within their constraints—and in pursuing their fragmented efforts, cements a dysfunctional system of interchange.

There may still be hope for a corporate fix for Chicago’s woes, but awaiting private solutions to complex collective action problems can lead to disappointment. In these times, we do not have the luxury of patience: as much as anything else, 47th Steeet’s story is a call for civil society to grapple with freight. Since the 1970s, policy efforts to grapple with the fragmentation of freight systems’ governance have dwindled. Today, goods’ place in public debates extends little further than its ability to justify highway projects, and present the occasional crisis for policymakers. More than anything else, Chicago’s example demonstrates the limits of that model. We must begin treating freight and the array of actors which govern it as important stakeholders and instruments for advancing our policy goals. Whether it be in the realm of climate, economic development, or otherwise, freight configures so much of our lives; we are poorly served to ignore it.

A year ago, the New Jersey Turnpike Authority released a plan to widen the New Jersey Turnpike Extension through Jersey City. Estimated to cost over $10 billion, the plan has earned the support of a series of construction industry groups, unions, and the state’s governor — and has run up against vigorous resistance in Jersey City itself, where the widened road will displace residents in working class communities of color while adding to the area’s air quality problems. Needless to say, this proposal is inadvisable. Time and again, research has shown that highway widenings rarely reduce congestion, and instead just drive the number of cars on the road — and with them, emissions and accidents — upward. Highway widening’s history of failure should be enough to dismiss this project out of hand, but analyzing the specific rationales for the project put forth by the NJTA make the case against the road even stronger. Rather than being the solution to any of the region’s pressing transport problems, this proposal is a $10 billion dead end. 

The “Need”

In their environmental assessment, the NJTA says the following about the need for the project:

These investments are necessary to maintain structures comprising the NBHCE and accommodate recent and future increases in travel demand along the corridor. This increased demand is associated primarily with the new deep water port operations and freight handling facilities along the Bayonne waterfront and redevelopment of major sections of Jersey City and the City of Bayonne. The anticipated increase in commercial vehicular traffic, as well as existing and growing passenger vehicle traffic, will place new travel demands on the entire length of the NBHCE [Newark Bay-Hudson County Extension, the full name of the Turnpike Extension] mainline and Interchanges 14A, 14B, and 14C that are critical to accessing Hudson County and New York City.

The statement identifies two major rationales for widening the highway: the growth of the port in Bayonne, and the growth of Jersey City. It is true that both the port and the city are growing—but it does not then follow that we need to widen the highway that feeds them.

Reality

Traffic volumes on the Turnpike extension have remained stagnant. While traffic volumes did increase temporarily during the protracted Pulaski Skyway reconstruction project, they actually were in decline beforehand, and have stabilized at around their 2011 levels after the Skyway project’s completion—despite Jersey City and Bayonne growing by a combined 17 percent between 2010 and 2020.

When looking at the individual flows of traffic on the Turnpike, the image of highway traffic gets more complicated, but no more supportive of the NJTA’s vision. Again, contrary to the idea that population growth begets more driving, vehicle volumes at exits 14B and C—those which serve rapidly growing neighborhoods of Jersey City, as well as the Holland Tunnel—are at or below their low points for the last decade. Traffic increases, such as they exist on the Turnpike Extension, are coming entirely from Exit 14A, the connection that links the Turnpike to the port, Bayonne, and Staten Island.

While it might be tempting to read 14A’s growth as a vindication of the NJTA’s claims about port traffic, 14A’s recent growth has been driven almost entirely by cars using the exit. Between 2018 and 2022, the number of autos moving from 14A to points north along the Turnpike Extension rose by about 30 percent, while traffic volumes between 14A and points south and west on the highway network remained near their Skyway shutdown highs through the pandemic. Few simple explanations exist for that growth. Some of it is likely related to the completion of capacity-constraining construction projects at exit 14A itself in 2018 and the Bayonne Bridge in 2019. And, to the NJTA’s credit, some of it is likely related to new transit-poor developments along the Bayonne waterfront. But even with these increases, the fact remains that the aggregate number of new users at 14A is small; their increasing numbers barely offset declines in traffic further north. 

Remarkably, throughout this period, the number of trucks using 14A to access Bayonne’s port facilities and industries has remained small. Despite a 45 percent increase in tonnage at Port Jersey in Bayonne, trucks’ share of all traffic using Exit 14A remains well below 10 percent. This is to say: given that past freight growth has had relatively minor impacts on overall highway traffic volumes, the NJTA’s “anticipated” freight traffic increases are unlikely to add enough trucks to significantly stress the highway. Make no mistake: ensuring efficient access to the port is a matter of great importance, and planners must understand the freight flow that those trucks represent as an important stakeholder in the region’s transport network. Nevertheless, the notion that truck traffic growth justifies highway widenings here seems flawed: in both absolute and incremental terms, demands on the Turnpike’s capacity today come overwhelmingly from cars. 

Whatever the NJTA’s claims might be, then, traffic data simply do not support highway widening. Indeed, there is good reason to expect that the next decade will bring traffic declines in Hudson County. With the recent approval of congestion pricing in New York, ongoing efforts to calm traffic in Jersey City, and the trend towards building less parking in Hudson County housing developments, the usefulness and user base of the Turnpike Extension will shrink. Traffic models are famously bad at predicting stable traffic volumes; here, again, they seem to have erred in their estimates.

Alternatives: Mass Transit

We do not need to widen the Turnpike. But if Northern New Jersey is to continue growing at anything approaching its current rate, it will need better transportation. Unlike Turnpike traffic, mass transit use in North Jersey is growing, with Hudson County’s PATH and Hudson-Bergen Light Rail system posting ridership increases through the 2010s. Nevertheless, there are barriers to greater use. While the region currently enjoys relatively good rail and bus links to and from Midtown Manhattan at peak hours, travel at other times of day and to other destinations can be difficult, pushing people into their cars. Offering better transit alternatives to would-be drivers will require remedying those problems as the region continues its growth. 

Possibly the lowest hanging fruit among alternatives for highway widening is increasing off-peak transit frequencies in Hudson and Essex Counties. Historically, mass transit planning in New Jersey has catered primarily to peak-hour commuters heading to regional business districts, leaving riders destined for other points at other times of day with fewer options. Especially in the post-COVID environment of reduced commuting and increased leisure travel, that deficiency is important to correct. Matters are especially dire near the Turnpike’s corridor in Hudson County. Since 2006, when Jersey City and Bayonne had considerably fewer people than they do today, off-peak rail transit frequencies have fallen precipitously thanks to service reductions during the Great Recession, long-term construction projects, and COVID. Despite significant investments in longer trains on both the Hudson Bergen Light Rail system and PATH, and new signaling on PATH, not a single rail route through the county runs as many off-peak trains on weekends as it did before 2008. Nor is the situation much better on New Jersey’s commuter trains. Off-peak service to and from Hoboken terminal (just north of downtown Jersey City) generally runs once an hour or less on each line, and access to many important destinations on the commuter network requires transferring to trains from Penn Station—all of which are currently frequency-limited by repairs to Amtrak’s tunnels under the Hudson River. The agency’s buses provide moderately better service, but do so on a network of routes which both focuses its energies on carrying passengers to and from New York, and dilutes high overall levels of bus service with nearly endless route permutations. 

Especially given that the PATH network is controlled by the Port Authority, which is legally forbidden from using tax dollars to fund its operations, providing funding for more off-peak trains across the PATH and NJT may require broader changes to transit and public authority legal structures. Nevertheless, for the sake of the entire region, better off peak transit is critical: by making transit an attractive alternative for all types of trips at all times of day, leaders can help shift New Jerseyans away from driving.

Of course, simply running more buses and trains can reach a point of diminishing returns if one has underinvested in infrastructure. While North Jersey seems primed for major commuter rail expansions with the Gateway Project’s new Hudson River tunnels and a long (albeit only partially funded) list of accompanying infrastructure improvements, those improvements will do comparatively little for the Turnpike Extension’s user base. Only 21 percent of the highway’s traffic destined for the Holland Tunnel during the morning peak; the transit improvements required to shift drivers away from the road are really those to and between Hudson County points. Some of these fixes might be relatively cheap. Expanded or discounted transit service to Newark Airport for workers, new bus lanes, better integration between PATH and New Jersey Transit services, and Hudson-Bergen Light Rail speed improvements could all likely be implemented for a small fraction of the cost of a new highway to immense benefit for regional riders. But advocates should not let their ambitions for transit remain so small. Whether it be extending the light rail system to Staten Island, a Springfield Avenue rail transit route in Newark and Irvington, or even a subway down the spine of Hudson County, the Turnpike Extension’s hinterland is rife with opportunities for investment in better transit. 

Alternatives: Freight

While truck volumes over the Turnpike Extension have not achieved the levels or growth rates that would make them truly credible motivators of highway expansion, that does not mean their problems are unimportant. Lurking behind the Turnpike Extension’s planning process is an upcoming decision point in the future of the Northern New Jersey port. In their master plan for the future of the port, the Port Authority outlined two broad visions for the port’s future. In the first, the Authority would focus its expansion efforts in Newark and Elizabeth, at the historic core of the port. In the second, it would shift the port’s center of gravity to the docks at Bayonne, adding terminal capacity in an area with fewer navigational constraints. The authority has yet to select from between its plans, but if they do proceed with the Bayonne plan, truck volumes on the Turnpike from 14A to points west will increase: only about 10 percent of the port’s containerized cargo leaves the region on trains. Given those potential truck traffic impacts on the Turnpike, port planners should make their decisions while paying close attention to highway capacity: the future growth of the port should not create the need for additional highway expansion in densely populated areas.

Irrespective of what plan the Authority ends up choosing, there is more that officials could do to mitigate truck traffic around the docks. The lion’s share of the freight moving by truck off the docks is destined for nearby warehouses which feed the New York consumer market. While in most cases, these volumes will be difficult to convert to other modes, the port has seen some success subsidizing short-haul barge services to carry containers around the harbor. One such barge shuttle already links Brooklyn with Port Newark; if container volumes at Bayonne continue to grow, the Port Authority could conceivably set up another such service between Bayonne and points elsewhere in the region. Port planners might also consider funding short-haul rail services between the port and the fast-growing warehouse clusters around Allentown and Harrisburg in Pennsylvania. New York’s notoriously expensive trucking market already makes slightly longer rail hauls (to Worcester and Syracuse) possible without government support; while moving containers to Allentown and Harrisburg might require some subsidy, the involved costs will almost certainly be smaller than that of adding more highway capacity across New Jersey. 

Conclusion

For over seventy years, the default response of American transportation planners to congestion has been to add highway capacity. And for over seventy years, Americans’ daily travel has been getting more difficult. The Turnpike Extension’s proposed widening is just one in a long list of similarly inadvisable proposals under consideration today; if we are to save the environment—let alone enhance the function of our cities’ transportation networks—we need to learn new planning habits before it is too late.

Every day, hundreds of thousands of New Yorkers lurch into dusty darkness as their Manhattan-bound 4 or 5 train leaves Brooklyn’s Franklin Avenue station. As the blurred fluorescence of the station cuts to the brown-grey textures of tunnel walls, their trains speed up, and tilt uphill, following the lines of tortured topography bequeathed on Brooklyn by a glacier 18,000 years ago. And then, the train slows down. And then it speeds up. And then, again, it slows down. See-sawing back and forth between fast and slow while screeching around the curve which carries trains onto the broad, downwards-sloping diagonal cut that is Flatbush Avenue, these trains’ path from Franklin to Atlantic Avenue is one whose variability is a story of signaling, speed control, and a subway system locked in a slow-moving crisis.

In 2018, Aaron Gordon, then working at the Village Voice, wrote an article on the subway’s performance woes. The system of 2016-2018 was in crisis; on-time performance had plunged below 75 percent and ridership had begun to fall. Up until that point, most mainstream press coverage had ascribed the system’s reliability struggles to old signals, deferred maintenance, and other, similar stories of disinvestment. What Gordon, and the dozens of articles written on the subject after he published his piece, showed was that the subway’s problems had a lot more to do with the low-level erosion of capacity and speed across the system than they did some meltdown of technology. Years of speed control signal installations, scared operators, and increasingly restrictive operating rules had taken their toll on the system, and as ridership and maintenance loads ticked upwards, the system devolved into a crisis. Important and revealing as this coverage might have been, it did little to illuminate the deeper roots of those signaling decisions. Why did the agency add all these speed restrictions? Why were train crews so scared of moving fast? Why, in other words, did the subway slow down without seemingly anyone working to stop it?

Signals, 101

The subway’s signal system exists to keep riders, crews, and trains safe. As electrified, grade separated rapid transit systems evolved into a category of urban railroading distinct from streetcars during the last quarter of the nineteenth century, signal systems became a key means of organizing and bounding the increasingly speedy movement of trains and people. And, as death tolls rose, accomplishing safety seemingly required these signal systems not only to advise train operators on whether they could proceed safely, but also to enforce those advisements. In that mandate to ensure life and limb lies the origins of the speed crisis.

The working theory of the subway’s signal enforcement begins like this: provided a means of activating the emergency brakes on a train passing a red signal (in New York, a trip stop), trains must be kept far enough apart that a speeding train passing a stop signal will not collide with its leader. That distance varies extensively, dependent as it is on the maximum potential speed at which a train can pass a signal, the slope of the track, and so on – but this commitment to maintaining one ‘safe braking distance’ between trains generally does not.

To achieve that braking safety, the subway’s signal architecture works around the principle of ‘two-block control.’ Every track on the subway is cut up into blocks, electrically isolated sections of line that can detect occupancy by sending current through the axles of passing trains. In two-block control, the signal system keeps at least one unoccupied block – really, however many blocks you need to provide that braking distance with a margin for error – between one train and the next. In signal-ese, the signal’s control line – the distance ahead of it which, if occupied, will make that signal red – extends one safe braking distance, plus one block, ahead of the signal. At junctions, similar rules apply. A switch lined against a train, or a train crossing in front of another is treated by the signals in the same way as is occupied track. So, you can just put these elements together and get safety. Right?

Not so fast. Two-block control is one of those inventions that makes for nice theoretical explanations but horrifically complex practical implementation. To understand why, let’s imagine a short subway line. Our imaginary railroad has four stops, two tracks, an S-curve, and a downhill slope. Remember: the question signal engineers ask themselves is “how far ahead of [this signal] do I need to keep clear so a speeding train can stop before hitting the train in front.” If you treat this question very literally, you might imagine a train that leaves the line’s north terminal and just keeps accelerating. After traversing the downhill across the whole line, it would be moving very fast by the last stop, forcing braking and train separation distances of incredible length. But could it even get to those speeds? The curve in the middle is, let’s say, good for only 25 miles per hour. By that point in its trip, our imaginary train would be cruising along at 60 – which is to say that a signal system for this line designed around constant acceleration would be one which ‘bakes in’ a derailment.

To make signaling work, engineers thus needed to handle speed. That meant figuring out ways to enforcing limits, and to deal with long train separation distances. It also meant establishing a set of basic assumptions about the behavior of trains and their crews.

The New York subway uses a type of signal known as a “grade timer” to regulate speed. Its logic is simple. Signal engineers know that a train moving at 50 feet per second (about 34 miles per hour) would take 10 seconds to traverse a 500-foot section of track. So, they installed timers in signals, set to the amount of time a train moving through the block of track in front of the signal should take to pass through it. Timers act as red signals until their countdowns complete, so trains passing such signals are stopped if they passed before the timer completed – which is to say, at a higher-than-desired speed. Later timers would give train crews two tries (“shots”) at getting that speed right, making operations a bit easier on the operators. But in all their forms, these devices allowed engineers two key capabilities: controlling speeds, and thus separation distances, on downhills; and enforcing speeds in areas of risk-laden track. On our imaginary subway line, timers could bridge the gap between assumed constant acceleration and the curve and separation issues that result from it: they would make this route’s service safely possible.

Grade timers, however, only got engineers so far. Subway signal systems might exist to keep subways safe, but lest they compromise the effectiveness of the subway service, they also need to provide capacity. A two-block control system designed to stop speeding trains does not always do a great job of that, even if you add lots of signals. Continuing with our example, imagine we put in a signal system that assumes trains make stops at stations. It would have more lenient control lines than even our timer-controlled constant-acceleration design, but would still wind up with a few signals enforcing long train separation distances — for example, the signals with three-block long control lines approaching the rightmost stop.

Now, imagine two trains following each other closely into that last stop. When the first train is stopped in a station, the second train is held well back from it, as the signal system protects the leader from a follower travelling at speed. But the follower here is not moving quickly; it is, in fact, not moving at all. Why not let it get closer, provided it keeps to a low speed? Engineers’ response to this once again involved timers: some signals approaching stations and junctions will allow trains to pass if they are moving at a speed slow enough that the next signal has enough space to stop them if they begin to accelerate. These devices are known as “station timers.”

Station timers and grade timers are, at the end of the day, instruments. Whether and how they are ever used depends on a set of engineering assumptions, which directly concern the expected behavior of trains and crews, but really are about the limits of socially acceptable risk, and socially desired service. The most basic and immediate one in our example is about whether trains are assumed to stop at stations. But you can (and have to) go beyond that. Can we assume low train frequencies to save on more complex signal arrangements with more signals? Are we willing to assume that our train operators are alert and attentive? Do we design the signal system around that assumption too, baking signed-but-not-enforced speed limits into the assumed speeds of trains passing signals beyond it? What about switches – do we need to enforce trains’ speed over them, or do we assume that operators are paying attention when taking a diverging route? Or, better yet, station time signals. If an operator has their train well enough under control to reduce a trains’ speed to 20 miles per hour and clear a station timer, is it not then possible to rely on their awareness, rather than signal design, to prevent collisions with trains in front?

Eventually, you reach this list’s conclusion, which lies in an unassuming technical question about the trains themselves. After you have figured out what they will look like, and how fast they will run, you need to give signal engineers two facts about them: how fast they can accelerate, and how quickly they can stop. And you need to make sure that your car equipment engineers keep those values the same until you replace the signal system. Remember, the whole point of having signals is keeping a train going at any given location’s maximum possible speed enough braking distance from its leader to prevent a collision. If your gain acceleration performance or lose braking power without rebuilding your signal system, the whole pyramid of assumptions comes crashing down.

In a nutshell, what created the speed crisis on New York’s subway was that engineers and managers at New York City Transit changed almost all of those assumptions ahead of the signal redesign process. When the mess they created was made violently, tragically visible, people freaked out.

Making a Crisis

In 1948, New Yorkers excitedly greeted a new fleet of train cars. As they were barged across New York Harbor, they received a water cannon salute from a fireboat, and were adorned with a massive banner calling out to spectators along the city’s waterfront “New Yorkers: LOOK YOUR NEW SUBWAY CARS.” And indeed, these were special cars. The first new equipment delivered to the system since the ‘30s, they were hailed by advocates and management alike as the first step into a new, better transit system.

Perhaps these cars most important legacies, however, lay between their wheels. Unlike previous subway cars, the R10s were equipped with more powerful motors, giving them starting acceleration rates of 2.5 miles per hour per second, versus 1.75 on older cars. They also had a new braking system, which while mechanically more advanced, was slightly less powerful than those on older cars. The R10s’ predecessors could stop in 230 feet from 30 miles per hour; the R10s and their successors needed 250. It remains unclear how agency engineers weighed these factors when ordering these new cars – New York City Transit did not have a formalized speed policy committee until 1989 – but the outcomes of their choices were dramatic. These changes invalidated the basis of all signal designs preceding the introduction of the R10s. So, while signaling from then onwards would be based on the R10’s performance characteristics, this change began to set the stage for a crisis: every foot of track governed by older signals became a source of risk.

From there on out, conditions only worsened. Through the 1950s and ‘60s, New York City Transit focused on maintaining and incrementally improving its subway system. In some respects, this was good for signaling: older, pre-R10 signal installations were replaced with ones whose design matched the capabilities of the era’s equipment. But in others, it fed the fire.

One of New York City Transit’s key initiatives through the postwar period was platform lengthening. Between the 1940s and early 1970s, the agency extended all stations on the numbered lines to handle 10 car trains (11 on the 7), and converted much of the former BMT to handle 10 cars as well. These projects increased capacity, but they sometimes added to the system’s signal problems. Most subway stations have signals right before and right beyond the platform, so that the platform is contained entirely within one (or is subdivided into multiple) blocks. Lest platform extensions reduce line capacity, New York City Transit’s engineers hoped to preserve that property of station-area signaling, so in areas where the signal system had not been provisioned for longer platforms, or where signal system rebuilds did not accompany platform work, they simply moved signals. The changes involved were small – about 50 feet or so – but they worked to compromise the signal system’s safety. If your blocks and trip arms are not where designers imagined they’d be, and if you don’t take extensive action to remediate that disjunct, your system is less safe.

The final piece of the subway signaling crisis would fall into place twenty-odd years after the last platform extension, at twelve minutes past midnight on August 28, 1991. That night, work crews were doing some track maintenance on the Lexington Avenue Line’s (4/5/6) southbound express track between 14th Street and Brooklyn Bridge, so express trains were crossing onto the local track at the switches just north of Union Square Station. Those crossovers were sharp, good for only about fifteen miles per hour. So, when the intoxicated train operator of a southbound 4 train plowed into them at 50, his train derailed, killing 5 people, and causing so much structural damage to the tunnel that Park Avenue above it sank by half an inch.

The derailment sparked dramatic changes in subway management. The agency began taking drug and alcohol testing very seriously, while beefing up its supervisory ranks. It also shocked the agency into altering its signaling assumptions. Previously, train operators were generally thought to be reliable enough that they would heed speed limits when they approached switches or other areas of sensitive trackwork (the assumption that trains would heed station stops disappeared around 1980 for unknown reasons). After that night, those assumptions were no more: signal designs from then on would assume reckless operation. Having been so violently burned by their past signaling approaches, they would leave less up to chance.

With this policy change, New York City Transit’s signaling transition was complete. By changing assumptions about acceleration, braking and operator awareness, and by altering infrastructure with little regard for signaling impacts, the agency had undermined the technological and institutional underpinnings of signal safety. Each resignaling project brought signals into closer alignment with these assumptions, but those efforts were slow. Transit’s engineers had created a gap between what they thought their signals ought to do, and what they actually did.

Breaking Braking

Lengthening, resignaling, equipment performance changes – these factors all combined to create a system that, by the agency’s standards, was less safe. But while the subway of the later twentieth century was far from a perfect place, the signal system’s problems had yet to become entirely clear. Collisions happened more frequently than they do today, but they generally did not kill. And while there exist news reports of incidents that sound suspiciously like signal system design issues – for example, a B train which rear-ended the system’s revenue collection train in Brooklyn in 1968 – these seem to have been rare. Most of the accidents whose stories made it into the press seemingly had little to do with signal system’s deficiencies.

That began to change in the 1980s, when a series of engineering decisions cast the signal system’s problems in ever-sharper relief. The 1960s and ‘70s had seen New York disinvest from its transit network as the costs of antiurban, racist, and increasingly austerity-minded American policy increasingly fell on New Yorkers’ backs. The resulting transit maintenance crisis defined civic narratives around the subway for years to come – and eventually forced elected leaders to change their approach to the system.

One central element of electeds’ renewal efforts was the creation of regular and fully-funded capital plans. These allowed agency leaders to renew worn infrastructure and equipment, and, most historians agree, helped the transit system on an upwards trajectory into the 21st century. Among the main initiatives funded by those capital plans was known as the General Overhaul Program, or GOH. The idea was simple: the agency would use its capital funds to rebuild its existing fleet of subway cars to address defects and install more reliable mechanical equipment. Among the many changes made as a part of this program was the replacement of brake shoes – the things which push against wheels to stop trains – across the subway fleet. Up until that point, most cars had had shoes made of cast iron, long the standard in the railroad industry. However, in the 1950s, material scientists perfected ceramic composition brake shoes, which allowed smoother, more consistent brake performance at lower brake cylinder pressures. Agency engineers hoped that installing these shoes would help save the agency money, and mitigate the subway’s persistent dust control and noise issues. This change from cast iron to composition shoes was a perfectly sound idea, but somewhere in that process, New York City Transit’s engineers lost the plot, because the changeover inadvertently compromised trains’ brakes.

It took the Union Square derailment for agency management to realize that there was a problem. Operators had been noticing degraded braking performance for years, but before that night, their complaints evidently failed to reach – or interest – agency management. However, only two days after the accident, a group of New York City Transit managers ordered a braking performance test with the same type of cars (R62s) that had been involved in the crash. When crews ran their tests, the cars failed spectacularly, exceeding the signal system’s assumed stopping distances by 100 feet or more. Over the coming months, the agency’s own board of inquiry worked with investigators at the New York State Public Transportation Safety Board to do more braking tests with R62s, all of which yielded the same outcomes. Evidently, in the changeover from cast iron to composition brake shoes, the agency’s car equipment engineers had compromised the efficacy of at least some trains’ brakes.

One might expect that such an alarming mismatch between design and actual performance would yield a concomitantly intense institutional reaction. But in an agency whose internal organization was described by a contemporary investigator as being a “confederation,” rapid action was not forthcoming. Despite the immense and recognized risks of low-performing brakes, the test schedule initially proposed by NYCT’s car equipment department foresaw braking analyses of the whole fleet taking about six years. Only after the intense protests of safety regulators did they agree to shorten their timeline – to four years. And only after another collision, and another spectacularly failed brake test, did the agency’s car equipment bureaucracy agree to quickly run some basic, diagnostic brake tests on the entire fleet. When these wrapped up in February 1993, agency leaders faced the undeniable reality that the subway’s signal system had been compromised: not a single car class performed within design specifications.

The agency’s response to these problems was initially weak. As they grew appreciative of the scale of the crisis, they adopted a much more lenient braking standard (giving 311 feet from 30 mph, versus 250) for the design of future signal systems, and funded a study of signal safety. What they did not do was implement braking system fixes. Their tests had showed that increased brake cylinder pressures would remedy cars’ problems, but through 1994 and 1995, the agency made little effort to implement that change – even on cars that were already cycling through the overhaul process. Later, agency officials would admit they had had “no concrete timetable” on which they planned to fix these safety problems.

Their apathy would prove disastrous.

Around 6 in the morning on June 5, 1995, the day’s morning rush hour was getting started. High above Kent Avenue on the Williamsburg Bridge, riders on an M train waited patiently for a red signal to clear, and for their journey into Manhattan to resume. But after about three minutes of sitting in place, there was a loud crash, and their train leapt forward thirty feet. They had been hit by the J train behind them; fifty-four were injured, and one was dead.

That crash should not have been possible. Over the coming months, investigators would not only find that the train’s acceleration performance was well in excess of what the 1915-vintage signals on the Williamsburg Bridge had been designed to handle, but also would pull back the curtain on the system’s braking problems, finding that the train’s brakes were wildly out of compliance with signal system assumptions. As this news broke into the press, and as investigators honed in on these braking and signaling issues as root causes of the Williamsburg collision, the agency was forced to take rapid, decisive action.

Timers, Everywhere

New York City Transit’s immediate response to the Williamsburg Bridge crash had three parts. The first two had to do with acceleration and braking performance. The agency did not slow its cars so much that they conformed with pre-1948 car specifications, but they did slow them enough to ensure that they perfectly obeyed the signal system’s designed acceleration curve; in the years between 1948 and 1995, cars had actually gotten even faster. They also, finally, restored brake performance to pre-GOH levels, allowing the agency to safely and honestly return to the 250 foot/30mph braking curve that had governed signal design until the 1993 braking tests.

The third and final component entailed modifying the subway’s existing signal system to improve its safety. The acceleration and braking modifications brought train performance back into closer alignment with signal design, but problems remained – around busy stations and complex junctions, in areas with pre-1948 signaling, and otherwise. To provide full protection against collisions at those locations under the agency’s 1990s assumptions, NYCT proposed to do one of three things: lengthen problematic signals’ train separation distances, add grade timers to limit train speeds, or – in an effort that had been underway since the Union Square accident – add specialized speed control signals that would control train speeds over switches.

Over the ensuing fifteen years, the agency would forge ahead with these modifications, identifying and deploying fixes for hundreds of problematic signals. With each of these minor changes, trains slowed down. Rides that were once fifteen minutes became sixteen, then seventeen; stations that once could pump trains out every 100 seconds after a disruption could now only do every 110, 120, or 130. Signal safety improvements eroded the subway system’s capacity, speed, and resiliency: slower trains running further apart do not make for a good trend line.

To some extent, these impacts were just the cost of a safer subway. But the story really is not that simple. After the fervor about the bridge crash died down, and as the agency’s signal improvement efforts turned towards the rollout of Communications Based Train Control on the L line, and other pressing improvement projects, the monies, staff capacity, and track access needed to actually deliver fixes became increasingly scarce. That disinterest had a cost: underfunded signal modification problems stretching over decades of work begat excessively impactful signal designs. ­Take, for example, a signal whose control line – enforced separation distance – is just a bit too short. Ideally, what you would do to remedy that issue is cut in an insulated joint to define a new block that would bring the control line up to standard. But adding new insulated joints takes money and track time, and if you’re trying to avoid using both of those resources, what you will do instead is simply extend your control line to the next insulated joint – which may be hundreds of feet away. In doing so, you fix the safety problem, but do so at a relatively significant cost to capacity.

But what cost riders the most in this era was New York City Transit’s approach to signal and train maintenance, and operator discipline. In much of the system, the “timer” part of “grade timer” referred to a literal timer relay in a signal cabinet. Electromechanical equipment might be reliable, but it also requires maintenance – to check its function, and also to ensure that track replacement projects and the like do not disrupt signal logic. In the years following the Williamsburg Bridge crash, the agency’s emphasis on speed conservatism permeated maintenance processes, leading to a slow reduction in allowed speeds. A timer that nominally enforced 20 miles per hour might end up clearing at 17, a change that not only aggravated the speed- and capacity-reducing impacts of signal modifications, but also eroded train crews’ trust in the signal system.

That issue of technical trust was only made worse by changes to train cars themselves. Before the Williamsburg Bridge crash, few subway trains had speedometers. The agency expected crews to know their route and their trains well enough that they could work without them. However, pressure from investigating bodies coupled with the conservative speed reduction turn in agency management drove an effort to equip the entire legacy car fleet with speed indicators. Unfortunately, thanks to a combination of poor design choices and poor upkeep, many of these speedometers have proven to be wildly unreliable. So, as train crews navigated the increasingly speed-controlled subway of the aughts, they not only had to build uncertainty around signal design, but equally about the accuracy of their on-train equipment.

What transformed these problems into a much larger crisis was the agency’s commitment to hard-line operational discipline. With their newly expanded supervisory ranks and redoubled focus on stopping red signal overruns, the agency spent the 2000s punishing increasingly large numbers of train operators for operational mistakes. Discouraging reckless operation is an eminently worthy goal, but this trend’s coincidence with declining signal speed reliability served to exponentiate the trust-eroding effects of the agency’s maintenance regime. Given that the agency tended to discount the relevance of mis-timed signals when considering a train crews’ fate following a red signal overrun, operators had every incentive to treat signals cautiously; one’s approach to timers became a litmus test for each train operators’ degree of risk tolerance, and knowledge of their railroad. The upshot of this for riders and planners alike was disastrous. A now-famous analysis of subway signals in 2014 found that the average train crew operated well below design speeds through timers – and it took little more than a bit of attentiveness on your daily travels to notice how widely that ad-hoc safety margin varied between crews.  

Hope

Whatever the problems of signal discourse in the 1990s and early 2000s might have been, we today live in the seeming beginnings of a better world. New York City Transit stands alone among North American transit agencies in having a formalized speed improvement program, one whose explicit purpose is to push back against the accreted momentum of the system’s history to raise speeds, restore operators’ trust in their signals and supervisors, and increase ridership. These efforts are not, generally, reverting signal modifications — they rather focus on signed-but-not-enforced ‘civil speed limits’ — but have had large impacts on the speed and reliability of service. For the long run, the agency is also investing billions in signal modernization, replacing the worn signals of our speed crisis with computerized systems – ones which are not only cheaper and more reliable, but also provide the agency with an opportunity to safely break out of the train performance straitjacket imposed by the current signal system. With each increased speed limit, each increment of change to agency discipline culture, each retrained service supervisor, and each new mile of CBTC signaling, the agency is – finally – healing.

But even as the horrors of the subway’s signaling crisis fade into the past, the basic questions which created it will not. A few months ago, the new CEO of Washington’s Metro pointed out the media and regulatory tendency to treat even non-fatal transit incidents as major news stories without batting an eye at the daily carnage on highways. As many pointed out in response to the tweet, there is hardly a one-to-one tradeoff between rail safety and mode share; some of the safest railroads in the globe are also among the most used. But Washington is not going to be Japan tomorrow; New York was not going to become Switzerland between June and December 1995. Obviously transit systems should strive for – and be held accountable to – the highest achievable levels of safety for their mode, but slow, infrequent, unreliable transit pushes people into cars, and into whole new worlds of risk. Alongside all the subway’s lessons about infrastructure’s capacity to encode social values, the way complex institutions negotiate danger, and the importance of treating workers fairly, the signal modification story inevitably begs the question of how infrastructure institutions should navigate known risks. I have no answers here, and I certainly do not think that the American legal or liability assignment system provide good incentives for any parties involved. Every time I ride a Manhattan-bound 5 train through the one-shot timers under Eastern Parkway, I turn the question over in my head. For I know I will have time to give it thought; it is a slow ride to Atlantic.

If you follow my Twitter account or have read other articles on my blog, you will have likely seen a chart that looks like the above. These are what are called stringline (or Marey, or time-distance) charts, a very common tool used in transit and rail planning. Usually displayed with distance on the vertical axis and time on the horizontal, each one of the lines on these charts represents an individual vehicle moving along a line. This sort of visualization is incredibly useful for understanding the function of transport networks. By synthesizing the flow of traffic across a whole route, they allow dispatchers to monitor their territory, and help planners write timetables and work plans that minimize conflicts between vehicles. But stringlines can be much more than just a planning tool: they can provide a basis for interrogating the values, constraints, and risks within our transit networks. In this post, I hope to both illustrate these ways of reading stringlines, and also (for those interested) give instructions on how to make your own stringlines using widely available transit data. 

Origins

Stringline charts are a product of the managerial revolutions of the nineteenth century. Sometime during the 1840s, French railway schedulers began using them to write timetables on the increasingly busy bits of railroad connecting Paris with other major French cities (for a more detailed account, one from which this paragraph draws extensively, see here). It is essential for the smooth flow of traffic that trains moving in opposing directions or at different speeds meet at locations with overtaking or passing tracks, and while it is possible to design timetables to achieve that without the use of graphical aids, plotting trains on a chart and then aligning their schedules to intersect at points with more than one track makes that process quicker. This is exactly what French railway schedulers did, drawing trains as lines (with each lines’ slope corresponding to the trains imagined average speed) on their charts and constructing their route timetable upwards from that graphical, and geographical, framework. 

Over the ensuing century and a half, graphical timetabling proliferated across the world, and has become a mainstay of both scheduling and operations management. Even in this era of computer-aided dispatching and timetable planning, these visual representations remain a powerful tool for succinctly understanding interactions between vehicles and constraints across lines. The usefulness of these charts for planners is one that mirrors. Plotting a transit service’s timetable with a stringline allows an interested observer to understand much, much more about the operation of a given bit of service than can be ever gleaned from a tabular timetable.

How to Read Stringlines, In Detail

Essentially all stringlines you will ever encounter work on the same basic principle: plotting distance on one axis, time on the other, and the paths of vehicles in between. The code I wrote — and which I will describe how to use below — is capable of implementing complex transformations of schedule data, so I want to offer a primer on how to read its outputs. 

Take this stringline as an example. It shows B62 bus service between Long Island City and Downtown Brooklyn. Complicated, right? Let’s break down what’s going on. 

The B62’s route is typical of an urban surface transit route in the Americas. It’s a mix of two way streets and paired running on one-way segments. The code reflects this by breaking up paired one-way running segments into separate sections in the plot, and showing the two-way sections as one. To show continuity of trips over sections of the plot devoted to buses travelling in the opposite direction, it uses thin lines; to show sections where buses in that direction are making stops, it uses thick lines, with dots to show where buses stop (big dots for labeled stops; little dots for non-labeled). 

If you’re plotting two routes simultaneously, the same principle applies. Here’s a plot showing the combined N and W timetable. The two routes run together from Astoria to 42nd Street, where the N becomes an express to Brooklyn (continuing to share platforms with Ws until 14th St), and the W stays local to Whitehall Street. The thin lines help show N service’s continuity over the W-only part of the plot. 

You can make things pretty complicated with this code! It will, for example, allow you to plot multiple bus lines with separate-direction segments, odd branching on commuter rail networks, and so on. But these basic principles of how to read the plots are pretty constant. More on how to make these charts later.

Budget-Driven Headways: MBTA’s 94 Bus

As a transit-oriented student at Harvard, I have always been fascinated by the odd headways used on some of the buses in the MBTA network. A good example is the 61, which run from Waltham into some office parks along Route 128. Its Saturday service runs on a 50 minute headway, putting easy-to-remember clockface timetables out of riders’ reach.

Plotting a stringline of the route revealed why. The 61’s timetable runs with just one bus, shuttling back and forth between Waltham and the office parks. With one-way running times of about 20 minutes and a good bit of time for layovers at each end, the route ends up with a 50 minute headway (and a 60 on weekdays, when there’s more traffic).

Creating Risk on the Weekend F Train

Timetables tell stories. Those of the weekend F and G trains reveal service planning’s intersection with repeated service failures on the New York City subway. The F and G trains, for those not familiar, are the two subway services which run on Brooklyn’s Culver Line. The F runs to Stillwell Avenue (with some weekday peak trains turning at Kings Highway), but thanks to the relatively low ridership of F stops below Church Avenue, the G turns there. 

As anyone who rides the lower part of the F can tell you, the G’s terminal at Church Avenue causes problems. Southbound G trains often hold up Fs as they’re cleared of passengers before heading into tracks beyond Church Avenue where they turn around. Perhaps more surprisingly, northbound Gs regularly cause delays in F service as they enter service at Church Avenue. One might think that G trains entering service on schedule, and F trains still relatively near their origin terminals would make for a problem-free interaction at Church Avenue’s northbound platform — especially on weekends, when frequencies are lower. The reason that’s not the case has everything to do with timetables.

Back in the day (i.e. in 2006) both F and G trains ran every eight minutes. As maintenance policy-driven weekend capacity reductions have taken their toll on the system’s function, service levels have fallen. So today, the F, hard-hit by construction impacts, runs every 12 minutes, and the G runs every 10. This asymmetric headway is at the heart of the Church Avenue problem, and is readily visible on a stringline. Running both these services at their average headways would lead to conflicts (eg. Fs at the :00, :12, :24, :36, :48 and Gs on :06, :16, :26, :36, :46, :56 would conflict at :36 past every hour), so planners shift conflicting Fs slightly later to accommodate Gs. On paper, the timetable ‘works’ but this solution is fragile. Even the slightest perturbation in F or G service can lead to conflicts and delays at Church Avenue — which is exactly what happens a dozen or so times a day, every weekend (and weekday; F/G frequencies aren’t matched then either).

Transit’s Priorities: The Pascack Valley Line

New Jersey Transit’s Pascack Valley Line is one of New York’s lesser known corridors. It has neither the ridership of the Long Island Rail Road’s Babylon Branch, the scenery of Metro North’s Hudson Line, or the complexity of NJT Morris & Essex Line. Its timetable is one of the clearest encapsulations of commuter rail’s bent towards peak service — and the political factors which keep it that way. 

The first thing you might notice about this stringline is its directional structure. In the morning all trains run to Hoboken; in the afternoon, you have a mix; and in the evening, all run from Hoboken. If you spend thirty seconds looking at a map of New Jersey, it’s not hard to tell why: Hoboken (and Secaucus, one stop to its north) are where riders can connect to trains and ferries into Manhattan’s Central Business District. This timetable is peak, white-collar oriented railroading in the flesh; the thousands traveling from Jersey City and Hoboken to jobs, friends, stores and appointments in Hackensack and beyond simply cannot use the line for their trips.

If you sit with this chart for longer, a deeper problem becomes clear. Especially when looking at railroad timetables, something always worth noting is where trains pass each other. On this stringline, those ‘meets’ take place in three places: between Pearl River and Nanuet, between Anderson Street and New Bridge Landing, and between Teterboro and Hoboken. That’s not an accident. While the Pascack was once double tracked as far as Oradell, it today is a single track line from Teterboro to Spring Valley, with passing sidings at Anderson Street and Pearl River. The fact that it takes a bit more than half an hour for a local train to travel between those two sidings is what transforms heavy, core-oriented peak commuting volumes into the absence of service for other types of travel. It is simply impossible to run the high-density peak direction service that decisionmakers want and run meaningful reverse-peak service. The off-peak portion of the stringline makes it easy to understand why. If an inbound and outbound train meet at New Bridge Landing, half an hour will elapse as the outbound train makes its way up to Nanuet and meets the next inbound. In turn, another half hour will go by as the inbound train makes its way down to New Bridge — add it together, and you have a minimum headway of an hour or more. 

Here’s the real kicker: New Jersey Transit tried to fix this. In 2003, the agency proposed adding several new passing sidings to the line to break up that distance and allow for reverse-peak service. One would think that the towns which would get much-improved train service from this new infrastructure would welcome it — but in reality, a coalition of six towns ended up suing New Jersey Transit alleging (just about baselessly) that the sidings were actually a ploy to introduce more freight service on the route. These NIMBYs won; only four of the planned six new sidings were built. The long gap survived efforts to remove it, and with it, poor service levels for non-peak travelers. A stringline cannot give you all this historical context, but in showing how service works it can help reveal salient operational problems and structures on a line, and provide a starting point for questioning the political histories of transit infrastructure.

What These Charts Hide

Stringlines are a powerful visualization tool, but they cannot tell you everything about a transit route. Like every chart, they are abstractions of reality. Some of those are visible on stringlines, but are unexplained; those are ones you can piece apart with “why” questions of the sort I explored above. Others are less obvious, and to be an erudite reader of strings, you must be attentive to things they fundamentally cannot display. 

When making stringlines with public agency schedule data, you are necessarily getting a limited version of reality. Trains and buses moving along a line don’t accelerate instantaneously and travel at constant speeds between stops. The motion of vehicles between two stops shown on a string chart is complex, and its complexity only increases as stops get further apart. So, for example, if you’re looking at a chart of the Metro North New Haven Line, you should keep in mind that the express trains that make no stops between 125 St and Stamford move in ways poorly reflected in the straight, constant-speed line that connects them on the plot.

Similarly, transit vehicles are not point masses. They take up physical space, and in railroad contexts, they occupy signal space as well. To give an extreme example, a stringline will represent the progression of an 8,000 foot freight train on a route with signals every two miles as a line like any other, but in reality, the train not only occupies 8,000 feet of space behind its line, but also causes signals stretching back 6 miles from its rear end to display worse-than-green aspects. If you have data on signal systems and train lengths, you can find ways of representing those realities — but public GTFS schedule feeds do not allow that. 

Lastly, it’s always important to remember that transit services are networked. Stringlines can help you understand the movement of vehicles on a route, but oftentimes, there are inter-route relationships that play a role in determining schedules. Planners might have optimized service on a bus route to make timed connections with a perpendicular route at an important intersection, or timed train service to simplify interline crewing. More so than issues of variable movement or signaling, these interconnections are understandable with research. However, that research must extend beyond the confines of just one line — though they are tools for interpreting one route, stringlines must be given network context. 

Making Your Own

Now, the fun part: making your own stringlines! I have written a code that should allow you to plot most agencies’ schedule data as string plots. Here’s how to use it. If this is hard to follow, I have configured the code to make a plot of 2 and 3 train service in New York, so you can start by running that and seeing how these steps impact it. 

Step 1: Install R and RStudio (both free).

A good tutorial on how to do this can be found here; if you want to familiarize yourself with the basics of R before beginning to use the code, try this tutorial.

Step 2: Download my code from here.

Step 3: Open the code in RStudio

Step 4: Install all the packages listed at the top of the code in the libraries() section.

R may prompt you to do this automatically; if not, type install.packages(“[name of package]”, “[name of next package]”) and so on in the command line, or use the packages tab on the right of the RStudio screen.

Step 5: Find your agency’s GTFS feed.

Transit.land and transitfeeds.com are my go-tos for schedule data, but if you can’t find what you want there you might also try searching [agency name] GTFS in your search engine — it has worked for me in the past. 

Step 6: Download the GTFS feed to your computer.

Step 7: Check to see whether your GTFS is suitable for this script.

Paste the path to the GTFS feed into the box on line 29 of the code. Then, select lines 1 through 69 and either click run or do command (mac)/control (PC) + enter. You can also paste the link to the download URL of a GTFS feed, as I have done in my example.

The only requirement of this code is that the GTFS must have a stop_sequence column in the stop_times object. Additionally, the route you are plotting must not contain internal loops or reversals.

You can figure these things out by typing View(dat$stop_times) into the command line of your RStudio window and by looking at a (geographical) route map of your line. But if you aren’t sure about any of these points, or are just lazy, I would recommend just running the code. Your computer won’t explode; you’ll just see a messy plot and/or a bunch of error messages.

Step 8: Figure out the date range of your GTFS.

If you’re downloading data from transit.land or transitfeeds, the download pages for the feed will tell you its start/end dates; if not, type View(dat$.$dates_services) into the command line and see the dates included in the feed. Choose an included date, and alter the date_tar argument to reflect that choice.

Step 9: Choose your routes.

Some GTFS feeds have intuitive route_ids (eg. the A train in NYC is “A” in the subway’s GTFS), and others do not. You can find route_ids in two ways. Within the code, you can type View(dat$routes) into the command line and find your route (usually the route_short_name and route_long_name fields have the common names for the services). You can also look the IDs up on transitfeeds or transit.land. 

Note: some routes will have multiple route_ids associated with them. Unless there’s a clear difference between them (eg. one corresponds to the night rail replacement shuttle, and the other rail service, or one is northbound service and the other is southbound), I would suggest putting one ID in the routes_tar field (line 37) and seeing what it produces, perhaps with the other route_ids in the routes_secondary field (line 45). 

If all of this seems a bit confusing, I have built a workaround into the code. If, when you type View(dat$routes) into R, you see a route_short_name column with values in it, you can use those values instead of dealing with route_ids. Simply change line 39 to TRUE, and you’re on your way. This was quite useful for me, to give one example, when trying to plot Berlin’s S7, which has 4 route_ids associated with it — it saved me having to figure out which one(s) were actually in use.

When entering routes into the routes_tar and routes_secondary fields, there are some important rules/distinctions you should know. 

The routes_tar field can accept up to two routes that:

• Overlap for at least two stops
• Have the same direction_ids at those stops

The code will plot the entirety of all routes in the routes_tar field (as shown in the N and W plot above, and the 2/3 example plot). If you’re not sure whether a combination will work, try it. This functionality can get messy so perfect results are not guaranteed, but it works for me about 90% of the time.

The routes_secondary field can take pretty much anything. It will plot the portions of the routes you put into it that overlap with the route(s) you entered in the routes_tar field. If direction_ids don’t match, things can get wonky — but if you plot both directions, problems usually disappear.

Step 10: Choose your directions.

Because the 1 and 0 values don’t always correspond intuitively with, well, anything, I would try running the whole code before choosing a direction to plot. Moreover, some GTFSes (eg. Trenord) have both directions of train service coded as one. The iterative method here can really save you some pain. 

Step 11: Choose methodological factors.

• Template_choice: if you want to view stringlines based on the most common route variant, use “Most” here; if you want to see them based on the longest, use “Longest.” 
• Mand_stop: if, when you plot your stringline, the code has chosen a route variant that does not include some branch or stop of interest to you, look up a stop on that branch. To make your life easier, run 1-245, then paste View(stop_times%>%inner_join(.,trips%>%select(trip_id,direction_id,route_id)%>%inner_join(.,stops%>%select(stop_id,stop_name)) into the command line. Find your stop of interest, and paste its stop_id(s) into this argument to force the code to choose variants which serve it. Be sure to include stop_ids for each direction of service!
• Name_elim: this is a purely aesthetic argument that removes some stops from the final plot to make the y axes less cluttered. 
• Time_start and time_end. These ones are pretty self-explanatory, but it’s probably worth a reminder that you need to give start/end times which will encompass some amount of service on the line you’re plotting. Asking this script to plot, say, overnight train service in Washington DC will lead to meaningless and error-ridden results. Trust me, I have made that mistake more times than I care to admit. 

Step 12: Run it!

If you have already downloaded your GTFS, select lines 30-1039 and do command or control + enter. Once the plot has been made, you can save it (which often makes it more legible, because you can adjust the size of the plot image output) using the final two lines of the script. 

Step 13: Troubleshooting

This code isn’t perfect, in part because I have zero (0) formal training in R, and in part because transit systems and their data feeds are legitimately complicated. When things don’t work, I usually try:

• Ensuring I haven’t made some configuration error (eg. choosing an invalid date, or put route_ids into the routes fields while leaving the use_rt_sht set to TRUE).
• Changing the template_choice method to get a different shape
• Changing route_ids, or changing to route_short_name filtering
• Looking at a map of the route to ensure it doesn’t have any internal loops or reversals

I am certain there are parts of this that will break even with these fixes, or potential errors I did not catch. My email is in the “about” page of this blog. No guarantees I’ll get to more complex problems quickly (I’m a 21 year old thesis-writing and job-hunting college senior, at the end of the day), but I really genuinely do want this to be as useful as a tool as possible for transit advocates/enthusiasts so do shoot me a message if needed. And of course, this code is a work in progress; check back here for updates.

Happy plotting!

Every now and then, someone on Twitter posts an image of a subway countdown clock showing an extraordinarily long wait for the next train. Sometimes, that headway is the hard reality of our transit system, beset as it is by crew shortages, long off-peak headways, and disruptive maintenance. But sometimes, that seemingly distant train is actually a lot closer than the clock says. Why, you ask? Because over the last four years, New York City Transit has been slowly and subtly changing the way it handles the impacts of maintenance on subway service. This nearly invisible process has not only degraded the accuracy of countdown clocks, but also has contributed to increasing operating costs, falling service reliability, and an erosion of scheduling’s power as an organizing tool for transit.

Backstory

As transit riders, we often come to think of schedules as being a guide to our local transit network, or perhaps as a yardstick by which to measure its performance. But schedules are much more than that. Fundamentally, they’re the organizational system of transit networks — they make sure that vehicles, crews, and support staff are where they need to be when they need to be there. The New York subway runs on two sets of schedules. One is the base schedule, which lays out how a normal day’s service should work, and which establishes regular jobs for the system’s crews. The other is what’s called a supplement schedule. Supplements modify the base schedule to account for transient changes to service. These can be anything from minor schedule adjustments for ‘leaf season‘ on the Q line to the complete shutdown of a Manhattan trunk over multiple weekends. Critically, supplements are not blank slate rewrites of the base schedule for their applicable time period. They must dovetail with base service at each end. They also need to respect the base’s crewing fabric. Though supplements can change the movement of trains at will, they must pay regularly assigned crews for the travel time between their normal and supplement-schedule start locations, and must pay overtime for any shift backwards from normal start times. Any additional crews required then have to be woven into this highly constrained fabric, adding an additional layer of complexity to the process. Supplements, in other words, are organizational capacity saps: not only do they require a considerable commitment of scheduling resources to write, but they also must converse with a much broader infrastructure responsible for coordinating track access and diversions to become useful planning implements.

Given how resource-intensive the creation and operation of supplement schedules tends to be, one might imagine that they are used sparingly, acting as an institutional incentive for diversion planners to minimize the number of outages, maximize repeatability, and strive for precise estimates of service impacts. In reality, this could not be further from the truth. Responding to the subway’s 2016-2018 service crisis, agency leaders and politicians developed a series of initiatives to re-focus the agency on basic operations (SPEED), intensify maintenance (Subway Action Plan), and accelerate system upgrades (Fast Forward). SPEED aside, these corrective actions were not free lunches. In the absence of efforts to remedy NYCT’s maintenance productivity and service management problems, more work meant more service disruptions (during all off-peak hours, but especially on weekends, which will be the focus of this post), and more disruptions meant more supplements and variability.

The resulting scheduling problems have two components: one of scheduling policy, and the other of schedule quality. Increasing work volumes led to changed supplement scheduling practices that have degraded the accuracy of the system’s schedules and contributed to rising operating costs. At the same time, the sheer volume of supplement schedule production demanded by the agency’s work regime has collided with the dire effects of an agency-wide hiring freeze and longstanding underinvestment in scheduling capacity to seriously erode the quality of the schedules written by the agency. The net result: far from being an efficient scheduled transit service, the subway’s weekend reality today is chaotic, as reasonable management of costs and rider experience has become increasingly impossible in a time of imprecise schedules.

Shooting Blind

As maintenance volumes escalated, especially following the July 2017 announcement of the Subway Action Plan, the system slowed down. After a series of worker fatalities in 2006, NYCT adopted a stringent set of roadway worker protection rules which mandate that all trains passing alongside or through work zones slow to 10 miles per hour. While lighter work volumes had previously meant that those delays could be handled within the basic running time adjustments programmed into diversion schedules (eg. the added time that comes with an express schedule revised to run local), heavier workloads created an operating environment in which weekend trains would often run very late. Poor schedule adherence meant that weekend trains would arrive at merge points well off of schedule, exponentiating delays from work projects. As these work and merge delays propagated across a line and through the day, slow service interfered with the overall fabric of the system’s timetable. Frequencies fell below scheduled levels as trains and crews failed to cycle quickly enough along lines, and as congestion around work zones reduced throughput. Between February 2017 and February 2018, the fraction of weekend trains actually run fell from 98 percent to 94 percent as diversion-driven cancellations cut into service levels on lines with heavy workloads. This was not a tenable situation, yet was one which seemed to becoming the norm for New York’s weekends.

The agency’s fix for this set of service problems was to adjust their supplement scheduling practices to accommodate the system’s changing realities. There were essentially two prongs of this effort. The first involved reducing scheduled frequencies to align train volumes with line capacities; the second involved adding extra running time to ensure on-time arrivals at termini. The shadow service cuts involved in weekend work commanded more initial attention (they have been baked into base timetables since then), but the crudely implemented spate of weekend running time policy adjustments have been more lasting and harmful. Since the fall of 2018, the agency’s supplement schedules have included 3-8 minutes of additional running time for each work zone or diversion through which a line passes. These values are not tailored to expected work intensities along the line, but rather are standardized for each type of diversion (eg. when Es, Fs, and Rs sharing a track along Queens Boulevard, their schedules get an extra 8 minutes).

To make matters worse, the insertion of this time into schedules is imprecise. Rather than extending schedules where trains will pass through work areas, running time additions are sometimes implemented as long scheduled stops at stations remote from the actual work zone (this method of addition is also what lies behind the countdown clock issue — the clocks think trains will actually obey those holds). This approach means, for example, that adjustments to F schedules for trackwork in Queens might show up as an 8 minute long scheduled stop in South Brooklyn. That is trivial for line-level on-time performance calculations, but it is in fact very important for the operational integrity of a service. With its compensatory running time additions near the end of the line, an F delayed by work in Queens will run late into its merge with the G at Bergen St, as well as any other merge created by service changes routing trains onto the F’s tracks in Manhattan.

These modifications consequently have a mixed effect on weekend service. While they solve equipment cycle time problems, merges remain poorly coordinated, and with standards-based (rather than condition-based) time additions to schedules, timetables continue to diverge from realities on the ground. To take two examples from the last weekend in August: service changes ran E trains over the F in Manhattan, for which the F received some running time adds in Brooklyn. Those, along with padding baked into the base schedule, provided the F with a cushion well in excess of what was actually required, so F trains stacked up around Coney Island running down the clock until their scheduled slot into the terminal.

Over on 4th Avenue, schedules erred in the opposite direction. D, N and R trains were all running express towards Manhattan to allow work on the local track. Supplements provided 5 minutes of additional running time, inserted as an extended stop at 36 St (a great example of a running time increase implemented at the work zone). These 5 minutes were far from enough; D trains left the far end of the service change running 10-15 minutes behind schedule. When those Ds reached Manhattan, they entered a second diversion, running alongside the A and C on Central Park West’s local track — where their unplanned arrivals interacted with similarly off-schedule A and C arrivals to create relatively frequent merge delays at 59 St. Each of these mismatches is costly. Overlong F schedules inflate operating costs and crew requirements in the face of a budget and crew headcount crisis; inadequately padded D schedules cause rippling delays at merge points and force shadow service cuts as train and crew cycle times shoot up.

At this point, you might be wondering why NYCT does not use a more precise methodology for timetabling. The short answer? Predictability. Within each weekend, running times are only a bit more inconsistent than they are on weekdays; when trains are slow, they’re consistently slow, and vice versa. However, between weekends, running times vary widely. To give one example: through the fall of 2019, NYCT frequently ran all service on the Queens Boulevard line local to accommodate signal modernization work. On some weekends, service ran essentially as it did on weekdays. On others, trains blew through their 8 minutes of padding as the Queens corridor turned to molasses. Moreover, the localization of these delays varied extensively between weekends. On one, delays were concentrated around 63rd Drive; on another, they fell near 36th Street and Queens Plaza. It is not enough, then, to simply know that there will be work on a line. Writing accurate schedules requires precise knowledge of what work is going in each diversion, and where that work will impact service.

That information is difficult to provide. Thanks to their complexity (and to staffing and infrastructure issues I’ll discuss in a moment), supplement schedules take time to write; the agency generally plans them on six-week time horizon. While the development process for diversions and their supplements obviously involves an understanding of what work needs to happen during outages, a six-week projection for capital or maintenance projects may not be able to accurately predict the exact locations and intensities of work on a given day of a diversion, rendering precise estimates of schedule impacts impossible. This is not to say that the agency’s current practices have no scope to change — there may be some potential to begin tuning time additions to the number of different projects in each work zone, or the like — but in a project management environment where delays and cost overruns are common, it is understandable that planners might take long-run fine-grained work forecasts with a grain of salt. Writing better supplements will either require a complete overhaul of the agency’s work planning and asset monitoring processes to allow high-precision planning, or shorter supplement schedule lead times. The reason NYCT have not acted on one of these fronts yet? Institutional capacity.

Mixed Signals

NYCT’s struggles with its scheduling capabilities is most visible in the quality of the timetables it produces. The reality of the system today is that even if maintenance predictability problems were to disappear, the schedules put out by the supplement scheduling process would still likely be imperfect. Especially since the beginning of the pandemic, the agency has been producing increasingly inoperable timetables. In these, trains on the same track might be scheduled to pass each other, while others run at impossible 0 second headways, while even others run down varying amounts of padding at the same platform, resulting in as many as 4 trains being scheduled to be in the same place at the same time. Perhaps most concerningly (as noted in the caption to in the F line chart above), some service changes are now being done entirely without supplements, relying on dispatchers and padding built into the base timetable to manage the impacts of the reroute. The impact of these scheduling failures on delivered service is attenuated by the fact that weekend trains tend to have poor schedule adherence in general, but nevertheless present a roadblock to improvement. They all but guarantee that trains running on time will become late, which is hardly a desirable outcome for a schedule — these are supposed to be planning instruments which make the system work.

Behind these scheduling failures are conjoint problems of technology, staffing, and workload. NYCT, like many other transit operators in the US, uses a software called HASTUS for its vehicle and crew scheduling. Originally developed for bus operations, NYCT customized the software to be a useful support for rail scheduling tasks as well. However, the agency’s HASTUS installation is out of date. It has now been eight years since the agency last upgraded HASTUS, which stands in contrast to the six year upgrade cycle seen at less operationally complex agencies such as Chicago’s CTA. Thankfully, there is now hope for a better future in this arena: in July 2022, the MTA board approved a procurement to obtain a newer and better-customized version of HASTUS in the coming months, hopefully paving the way for a more efficient and effective supplement scheduling process in the years to come.

Software is only one small part of the battle here, however. Few departments have been hit as hard by the MTA’s ill-advised and abortive ‘Transformation Plan‘ and hiring freeze as has Operations Planning, the division of the agency responsible for creating schedules. Between the 2016 and 2021, the department went from about 400 budgeted and 377 filled positions to 350 budgeted and 284 filled. While a small portion of the change in budgeted headcount is likely attributable to changing responsibilities (as you might have guessed from its headcount, Ops Planning does a lot more than just scheduling, and some of those other responsibilities have been shifted over the years), the steep drop in the size of the planning workforce in the face of fast-increasing planning and scheduling requirements is grounds for alarm. Scheduling (and transit planning more broadly) is complex and contextual; even when supported by up-to-date software, schedulers must have a deep understanding of the quirks of each agency’s rules and network to write timetables and use heavily customized software effectively. A wave of departures like this not only places impossible strain on the agency’s remaining schedulers, but also risks interrupting the smooth reproduction of this institutional knowledge. The end of the agency’s hiring freeze provides some hope for improvement, but with a fiscal crisis looming and staff departures continuing, the horizon is hardly clear of storms.

There is, of course, another basic reality driving the divergence between agency resources and the demands of its right-of-way work: NYCT’s patchwork maintenance model. With its multi-track main lines and flexible interlockings, New York’s subway is well designed for its current maintenance strategy, in which service is rerouted around work zones but continues to run through the line. While it provides continuous (albeit confusing and slow) service during disruptions, this maintenance paradigm requires an enormous amount of supporting infrastructure, whether that be customized supplement schedules, flaggers, or otherwise. Before the Subway Action Plan and accelerated system upgrade projects exponentiated disruption levels during off-peak hours, subway schedulers were writing about eighty individual supplement schedules each week. Today, despite staff departures, that value remains high. Service this past week (8/27-9/2) involved the use of about 70 different supplements — and by New York standards, this week was not that rich in service changes.

Concentrating work into periodic full line segment shutdowns (as is done in Paris and Madrid) might both provide more consistent and predictable service for riders at lower costs, and help reduce the scheduling pressure and unpredictability that comes with today’s constant slew of minor diversions. Doing these shutdowns well would require considerable investments in bus capacity, regional rail alternatives, maintenance equipment, and otherwise, yet they might provide a much improved service product and more sustainable set of institutions for our transit network. Make no mistake: the first step towards better and more manageable off-peak service is improving the productivity of work crews and existing track outages through coordination and process enhancement. Yet alongside those efforts, there needs to be a parallel conversation about how we structure track access, especially given the post-COVID environment of depressed weekday ridership and strong weekends.

Conclusion

New York City’s supplement scheduling problem is a problem with no simple answer. It is equal parts staffing, technology, process, and infrastructure, and is entangled with the agency’s struggles with maintenance productivity, off peak service, and staffing, making its resolution an all-or-nothing project of holistic and far-reaching reform. Yet ensuring that New York City’s transit network can operate with the reliability and efficiency the city deserves is a goal worthy of such extensive efforts. Especially as the peak — formerly the time when transit networks were strongest — wanes, reliable off peak and weekend service will be essential to growing transit ridership following the pandemic. However esoteric the subject might be, however complex and impervious the institutional dimensions of the problem might seem, it is a project of reform which simply must be done. New York’s future relies on it.

Since the passage of the Bipartisan Infrastructure Law, Amtrak expansion has become a hot-button topic. Articles in the press have examined proposed new services, discussed the future of high-speed rail, and highlighted ongoing infrastructure projects to expand the nation’s passenger network. Yet these stories are as much about great challenges as they are grand plans. As almost every story on Amtrak expansion will tell you, the company’s trains do not use publicly-owned tracks outside the Northeast and a few lines in Michigan and California, putting passenger trains on freight railroads’ tracks. This creates capacity and reliability problems. In the US, freight and passenger railroads’ operating paradigms are different, with freight carriers running loosely scheduled services at low speeds, and Amtrak seeking to run high(er) speed, high-precision timetables. To bridge this disjunct, freight carriers ask for significant (and perhaps sometimes excessive) infrastructure investment before the addition of new passenger trains. These investments expand capacity, but significantly increase the costs of new service, and often seem to fall short of creating a truly reliable railroad. Though there do exist some successful cases of high-quality shared corridor operations (whose lessons about the importance of scheduling and precision dispatching should be kept in mind for all future service plans), the ideal passenger line is one separated from freight traffic.

So why don’t we have more of those passenger corridors? Lagging intercity rail investment is assuredly part of the answer here, but so too is the fact that American rail policy has largely memory-holed the history of the nation’s one route that is separated from freight traffic: the Northeast Corridor. Often treated as an exceptional case of passenger-primary infrastructure in a region whose conduciveness to passenger rail is also supposedly unique, modern rail discourse has effectively transformed the Corridor’s advantages into an axiomatic truth divorced from history. But if you pull back that veil, you will find a history full of lessons for today’s rail planners. The Corridor has always been a busy passenger corridor, but it also was once the single busiest freight artery in the Northeast. Yet over a short period between 1976 and 1990, almost all freight traffic was moved off of the line, helping transform it into a (relatively) high-capacity and reliability passenger railroad. How did this happen, you ask? Planning. For a brief moment in the 1970s, the federal government dipped its toe into comprehensive rail network planning in an effort to save the Northeast’s railroads from financial ruin. This effort saved the region’s railroads — and gave us the NEC we know today.

The Broad Way

Through the first three-quarters of the twentieth century, all facets of Northeastern railroading were defined by the Pennsylvania Railroad’s main line. Stretching from Washington to New York, the carrier’s route carried a plurality of regional freight traffic, and was the unquestioned dominant force in passenger transport through the Northeastern megalopolis. Though the Pennsylvania had pondered ways to segregate passenger and freight traffic on this extraordinarily busy route, it largely shelved these plans, and operated its main as a shared corridor. Electrification and multi-track construction helped manage the resulting conflicts, and in that era of shorter, more nimble freight trains, and slightly slower passenger trains, the Corridor worked decently, especially as rail traffic declined following World War II.

The forces which finally disrupted the Corridor’s shared stasis were set in motion during the 1960s. In 1968, the Pennsylvania merged with its longtime competitor the New York Central and regional railroad New Haven to create the Penn Central. About two years after the merger, the woefully mismanaged and structurally disadvantaged company fell into bankruptcy. Railroads had long struggled in the face of subsidized competition from roads, and industry’s shift to suburbs, the South, and the West, but before the Penn Central failed, the plight of the rails had not been regarded as a national crisis. In 1970, it became just that. Through the years following the PC’s failure, Congress created Amtrak to relieve railroads of their passenger service deficits, and created a new arm of the Department of Transportation to develop a plan to restructure the Penn Central alongside five other bankrupt northeastern carriers. Despite a Republican president and an increasing turn towards austerity and decentralism across the political spectrum, there was relative accord that the situation demanded expansive, and potentially even expensive, action.

Ironically, it was just as Northeastern railroads reached their nadir that the public began to dream of rail-based futures again. When Japan Rail introduced its Shinkansen (arguably the world’s first true high speed rail service) in 1965, American newspapers extensively covered the event, jolting American politicians into action. Not wanting to spend on entirely new lines as JR had, President Johnson instead directed the DOT to fund an experimental high speed service on the Pennsylvania’s main line. This initiative gave America the Metroliners. Laden with untested space-age technology, and running on decayed infrastructure, these cars ended up being medium-speed mechanical lemons — but nevertheless were a massive hit with riders. Rail patronage on the corridor increased by 46 percent between 1969 and 1970, providing a then-rare moment of hope for passenger rail service. As many had hoped, success begat plans for more rail. In 1971, USDOT concluded a study of Northeastern passenger transport, coming to the conclusion that the region needed high speed rail. Though it offered no explicit conclusion on the need for dedicated passenger infrastructure, it set the stage for a major set of investments in Northeastern passenger capacity.

Making the Corridor

These two threads — the railroad crisis, and interest in high speed rail — collided in the planning processes meant to restructure the Northeastern network. Though the corridor had once handled more total traffic than it was in the 1970s, the high-frequency, high-speed passenger services envisioned for the Corridor promised to make it increasingly difficult to weave freight trains among passenger runs, as would the increasing density of suburban commuter service. In light of these constraints, planners proposed that the reorganized freight railroad network shift most traffic not serving local industries off the Corridor, and that its ownership be handed over to Amtrak. To achieve this, planners sought to leverage duplication in the Northeastern network. Much of the American rail system had been built with multiple competing lines between cities; this burdened railroads with costly excess capacity as traffic began to decline, but became a boon to passenger and freight separation planning in the Northeast. Rather than running with Amtrak trains on the Corridor, freight trains from Northern New Jersey and Philadelphia would be sent inland through the Lehigh Valley and Reading, or would be routed alongside the corridor using a combination of lines from the Northeast’s bankrupt railroads and the solvent Chessie System. When the reorganization planners made their plan public, their proposals for freight-passenger separation were met with widespread acclaim, even winning the support of freight railroaders. The stage was thus set for transformation.

On April 1, 1976, these restructuring plans congealed into a new, government-owned railroad named Conrail, which absorbed all of the bankrupt carriers into a regional monopoly. Conrail immediately began shifting traffic off of the Corridor. But there was a catch. Government planners and Conrail executives had failed to reach a track sharing agreement with the Chessie System, meaning that most Corridor freight south of Philadelphia had to remain on Amtrak rails. Driving this, in part, was a spat with Amtrak over track use charges on the Corridor. Following their takeover of the line, Amtrak had raised access charges to values well above the national norm, which had upset Conrail. The freight carrier wished to continue some through-freight operations on Amtrak’s rails, in no small part because they could use electric freight trains on the line, and they worried that completing an agreement with Chessie would strengthen Amtrak’s position in negotiations. Conrail eventually relented, and had shifted most freight off the corridor north of Philadelphia by the early 1980s, but they did not resume talks with Chessie. Trains consequently remained trapped on Amtrak rails between Baltimore and Washington. While this choice marked a significant departure from the earlier vision for a largely freight-free corridor, it was one which had Amtrak’s blessing. The railroad’s studies of the Corridor had concluded that upgrading the parallel Chessie route would require too much money and effort on the part of involved parties. Without the will or the money to change, Conrail and Amtrak remained locked in this uneasy holding pattern into the mid-1980s.

That stasis was broken on a tragic winter afternoon in 1987. In the small town of Chase, Maryland, a northbound Amtrak train from Washington to New York collided with a set of Conrail locomotives whose crew had failed to stop at a red signal. The resulting collision killed 16 and injured 164, and set off a wave of railroad safety and operations reform efforts. These included strengthening drug use rules (the Conrail crew had been under the influence of marijuana at the time of the crash), installing speed limiters on freight locomotives using the Corridor, and redoubling efforts to get freight trains off of Amtrak rails. From 1987 to 1989, Conrail worked to remove the remaining trains — and largely succeeded. Much of the freight that had run on Amtrak rails until 1987 had been destined for interchange with southern railroads in Alexandria, VA. This traffic had historically been difficult to shift, but a round of railroad consolidation in the Southern US and Conrail’s eventual success at negotiating trackage rights over the Chessie line allowed Conrail to move all of this traffic off of the Corridor. What little overhead freight remained (and remains to this day) served terminals in Baltimore and northern Delaware, where the necessary track connections for a shift did not exist. A mere 13 years after Conrail’s creation, one of the most significant reconfigurations of American transport geography was complete.

Winners, Losers, and the Route Forward

On some level, the Northeast Corridor’s lessons about the importance of planning are painted most starkly when considering its flaws. Rather than being problems inherent to the separation concept, they were mainly consequences of incomplete implementation, and a failure to institutionalize regional rail planning in the years following Conrail’s creation. The most obvious such shortcoming is the continued presence of freight trains on the line in Delaware and Maryland. While volumes today are comparatively light and manageable under a mixed operating concept, the failure to complete the corridor development plan means there would be few easy ways to manage any potential future traffic growth. The original plan for Corridor freight diversion included a new track connection between Conrail and Chessie rails in Perryville, MD, which would have allowed most of these trains to avoid Amtrak’s line. The connection was shelved following the abortive track use negotiations with Chessie and Amtrak’s corridor studies. Today, the necessary right-of-way is slowly being covered in exurban development.

More costly has been the failure to institutionalize rail planning after Conrail. The 1970s were an era of deregulation, in which rail regulatory bureaucracies rightfully came under fire for having contributed to the industry’s decline, but lost in that milieu was the value of a federal role in planning rail capacity. Today, places like Reading, Pennsylvania suffer because of it. Conrail once owned a high-quality freight bypass route which linked the Northeast Corridor to yards around Harrisburg with minimal passenger interaction. As a part of the effort to reduce Corridor freight, traffic from New Jersey was routed off this line in favor of routes via Allentown, while freight headed west from Philadelphia which once used part of the route shifted onto a line via Reading to avoid conflicting with SEPTA and Amtrak trains between Philadelphia and Thorndale. This arrangement has become problematic in its relation to Reading-Philadelphia passenger service. North of Norristown, passenger trains between Philadelphia and Reading ran on the line which became Conrail’s diversion route for Philadelphia-area freight, and would use that line again if service were to be restored (as has been proposed often since the last train ran in 1981). Unsurprisingly, vastly increased freight volumes on the line to Reading have complicated these passenger service plans, contributing to restorations’ multi-decade holding pattern.

It did not have to be this way. Some minor investments in new track connections and expanded clearances would have allowed Conrail freights to use a (passenger-free) segment of the Reading route in conjunction with the freight bypass route to eliminate the conflict. This would have created a high-quality passenger route from Philadelphia to Reading, and preserved Philadelphia’s important freight link to Harrisburg. The point is moot. Without any organized means of advocating for current or future public interests in the rail capacity plans of private freight carriers, there existed little potential for recourse or negotiation when Conrail asked for permission to rip up a key segment of the bypass route in 1989. They lifted the rails in 1990, and though restoring the line for this purpose has been studied, costs would be high. The Reading-Philadelphia passenger service plans in Amtrak’s current expansion plans entail a three-times-a-day train, considerably less service than the corridor had in 1882.

These stories of the Corridor’s flaws are really just manifestations of a much broader American transport planning deficit. Our semi-privatized system of goods and passenger movement is riven with split incentives, inefficiencies, and inconsistencies which collectively produce a transport system neither capable of adequately serving public need nor able to control its social impacts. From Los Angeles’ highways to Pennsylvania’s cities, our national aversion to treating our infrastructure, its operators, and its stakeholders as comprising a single integrated system is harming us. In this light, the Corridor’s history not only highlights a technical approach to rail capacity management which could allow significant passenger and freight rail improvement given this country’s extensive rail network duplication, but also a broader ethos of direct and comprehensive governmental engagement with infrastructure. For one brief moment in American rail history, the government asserted itself in planning the nation’s rail network, saving the Northeastern rail industry, and winning rail passengers and freight shippers alike high quality infrastructure.

Future advocates and planners would be well served to learn from this example. America’s passenger rail future is likely to lie as much in new high speed lines as it is reusing legacy infrastructure, but especially for regional services and in urban areas, accessing existing lines will remain important. Whether through stronger and better funded federal rail planning interventions (in both passenger an freight movement) in the capacity and planning decisions of private companies, expanded public control of rail infrastructure, or greater attention to coordination problems, maximizing the value of our rail investments will require a more cohesive approach to infrastructure. Recent years have brought hope that such novel approaches — separation and otherwise — might be catching on, but with linked crises of climate, housing, mobility, and equity threatening our national future, it is imperative that we move boldly to create the greener transport system this country so sorely needs.