What's Wrong with Seattle's Streetcars?

Ever since I read this Seattle Times article on the South Lake Union Streetcar, I haven’t been able to get the phrase “streetcar opponent Eugene Wasserman” out of my mind. Homer’s works use epithets (in the term’s original sense) to describe an existential characteristic of an entity. I’m sure this Eugene Wasserman is a multifaceted human being, but in this article’s story, he is a character in the same manner as “swift-footed Achilles.” So it’s not intended as a personal attack when I say that I find the idea of a “streetcar opponent” to be absurd, and emblematic of how stupid discourse around streetcars tends to be. When thinking about what differentiates the value of a transit route, the type of vehicle operating it doesn’t matter that much, compared to the route’s measurable properties. Inhabiting the role of “streetcar opponent” just doesn’t make sense.

That being said, as I’ve developed some access-based measurements of public transit routes, Seattle’s streetcar routes have stood apart. They are comfortably worse than all King County Metro bus routes that primarily serve Seattle, when considering two such measurements, the percent replaceable and journeys per in-service second. Given that, is being a “streetcar opponent” actually a sensible position?

I’m unwilling to attribute the failings of Seattle’s streetcar routes to their use of streetcars, just because it is the most visible characteristic that they share. A more sensible course of action is to develop and test some hypotheses concerning the underperformance, using measurable characteristics that apply to any public transit route.

The Streetcars’ Failings

Before introducing those hypotheses, I want to describe why I consider Seattle’s streetcar routes to be underperforming; it differs from most who take a similar stance. Light present-day ridership is most often cited as an indication of their failure, as it is in the aforementioned article. I’m largely indifferent toward that. While the purpose of public transit infrastructure is to move people around, the failure of specific routes to do so is a symptom of many diseases—some of which are unrelated to the route itself. Ridership may vary due to changing economic conditions, and measuring it is fraught with instrument and sampling error. I choose to use measurements of underperformance that provide a cleaner signal.

Aberrantly low values for intrinsic indicators of the route’s capability to get individuals from origins to destinations, quickly and readily, signal underperformance. The two access-based measurements mentioned before—journeys per in-service second and percent replaceable—are such indicators. The streetcar routes rank last in them, but that alone remains insufficient evidence of underperformance. Two routes are always going to be the worst and second worst, and that may not signal a problem at all if the measurements are tightly clustered. That is not the case with the streetcar routes, though.

Journeys per in-service second quantifies a route’s ability to serve as part of the fastest path between an origin and destination, on the basis of the level of investment in it. With a value of 362.405 the South Lake Union Streetcar is the worst among routes that primarily serve destinations within Seattle, when considering a 30-minute time budget. For reference, the worst of all non-streetcar routes in this category is King County Metro’s route 70 at 2,093.208. This makes the South Lake Union line nearly six times worse than route 70, and over 19 times worse than the journeys per in-service second of the entire network. For the same time budget, the First Hill Streetcar is around 2.5 and nine times worse, respectively. To give a sense of a practical upper bound for this value in Seattle, the 1 Line is the best performer at 16,221.920 journeys per in-service second. This is almost 45 times better than the South Lake Union Streetcar, and over 20 times better than the First Hill Streetcar.

Percent replaceable builds on journeys per in-service second by determining how often a route that makes up a fastest path is necessary for completing that journey within the time budget. The more replaceable a route is, the more other routes would pick up the slack if it did not exist. For a 30-minute time budget, the streetcars are worst in this measurement as well, with South Lake Union at 84.4% and First Hill at 76.7%. The next worst Seattle-centric route is Metro’s route 10 at 74.7%. To get a sense of the scale of this measurement, the 1 Line is again the best route at 33.1%.

These measurements paint a damning picture of Seattle’s streetcar routes. Outside of the First Hill Streetcar’s percent replaceable, they aren’t even in the same neighborhood as Metro’s bus routes. To decide the fate of the streetcar system, though, it’s inadequate just to know that its routes are underperforming. There are other ways of slicing the data that make up access calculations, which allow testing hypotheses as to why the routes underperform.

Routeshed Constraints

The streetcar routes could underperform if they suffer from a constrained routeshed. A routeshed is an alternate way of counting the journeys component of journeys per in-service second. Rather than counting every journey, the routeshed size is the number of unique origin sectors represented within the journeys. Therefore, it describes how large of a region in which the route has any relevance. It captures information about both the walkshed of the route’s stops and its connectivity with other routes.

I specifically thought the First Hill Streetcar could be constrained due to its meandering path. The stops between 12th and Jackson and Yesler and Broadway form a backward C. Three stops—and, importantly, the in-service time it takes to run the streetcar between them—are competing for an area inside the C that could be captured by a single stop on a more direct path.

If this path is harming access, the routeshed measurement doesn’t capture it. The First Hill Streetcar’s routeshed is less anomalous than its performance measurements. With a routeshed size of 15,307 sectors, it is just worse than route 65’s 15,427. Route 65 is a darling of both percent replaceable and journeys per in-service second, so it’s possible to be a productive route with a routeshed of the First Hill Streetcar’s size.

The South Lake Union Streetcar’s routeshed size falls amidst routes with minimal service in Seattle. Routeshed is not normalized for the route’s size or cost—be that the number of stops, distance, or service-hour investment. As short as the South Lake Union Streetcar is, it could also be described as having minimal service in Seattle, so its placement here is not surprising. Overall, routeshed is not a particularly elucidating measurement for diagnosing the issues of either of the streetcar lines.

Corridor Competition

Seattle’s streetcars overlap considerably with bus routes—the article notes this for the South Lake Union Streetcar, but it’s true of the First Hill Streetcar as well. The South Lake Union Streetcar covers ground that overlaps with route 40 and the C Line. The First Hill Streetcar’s segment on Jackson Street is shared with a litany of routes. The only significant addition of frequent, all-day service provided by the streetcars is the First Hill line’s corridor on Broadway. When routes serve the same corridor, they can compete for being the fastest path between the same origins and destinations. In-service time gets spent redundantly, since for a combination of origin, destination, and time of day there can only be one fastest path. Typically both journeys per in-service second and percent replaceable will be depressed compared to routes with less overlap.

To test whether this is depressing the access provided by the streetcar routes, I created a new network where I lopped off parts of any route that had a significant amount of overlap with streetcar service, and then recalculated the network-wide access. To be clear, I don’t think that this modification would benefit the overall transit network, no matter what it would do for the streetcar routes’ performance. This is merely to test a best-case scenario for the current streetcar system. If there no other routes directly competing for fastest paths along its corridors, can the streetcars produce access measurements at least in line with the worst bus routes?

They cannot, but there is an improvement. The First Hill Streetcar’s journeys per in-service second increases from 793.435 to 1,144.478, and its percent replaceable is down from 76.7% to 74.0%. The South Lake Union Streetcar is up to 538.6496 journeys per in-service second versus 362.405, and decreases its percent replaceable from 84.4% to 80.5%. These remain the worst scores of any Seattle-centric transit route. There’s something to this hypothesis, but the effect is not large enough to explain the streetcars’ performance. A bus network redesign alone would not save the streetcar routes, but the present configuration is holding them back.

Speed

If the First Hill and South Lake Union Streetcars fail to deliver the access of even the least productive bus routes, even when given dominion over their corridors, perhaps their access is suppressed by a different kind of competition. The slower a transit route is, the more likely walking will be competitive with using it. If a rider must wait at a stop before riding a vehicle that moves barely faster than a walking pace, then, aside from the most fortuitously-timed trips, it may make more sense to walk.

When factoring in the time that a vehicle must stay stationary, most ways of moving through a city are fairly slow. But even by the standard of the average speed of transit routes within Seattle, the streetcar routes do not acquit themselves well.

The slower routes are a school bus route that only makes two stops in Seattle with time to load students factored in, and a route that weaves through First Hill during the rush hours to pick up or drop off commuters (remember that these speeds reflect only the service that these routes provide within Seattle, and these routes run significant service elsewhere). Route 10 is the slowest Seattle-centric bus route, and it manages an average speed of 7.49 mph. To better understand the nature of the First Hill Streetcar’s low speed, I broke it down by the average speed of each segment between stops.

From Pioneer Square to Capitol Hill

From Capitol Hill to Pioneer Square

The streetcar maintains its highest speeds as it goes downhill on Broadway, but the absolute highest speed is achieved as it pulls out of its Pioneer Square terminal, which is flat to slightly uphill. Pulling into this terminal is the slowest segment. This contrast highlights that speeds are generally higher at the beginning of the route. Presumably the schedule is padded more as time goes on to allow operators to make up for falling behind. I therefore hesitate to draw any conclusions about vehicle limitations from the data.

Perhaps the route taken by the streetcar is inherently slow. Signal timings and traffic may depress the speeds of all vehicles. Tight stop spacing can also slow down a route. For the Jackson Street portion of the streetcar’s route, I compared it to some of the bus service that covers it as well.

Route 1, to 10th and Armour

Route 14, to Mt. Baker

Despite making more stops—which would increase the time that the vehicle is not in motion, and necessitate more acceleration and deceleration—the buses maintain higher speeds between 5th and 12th on Jackson. The speed difference is especially pronounced going from 12th to 5th. These routes are most often run with trolleybuses, which, like a streetcar, must maintain contact with overhead wires.

With abnormally low speed identified as a problem, a new set of hypotheses emerges in service of determining a root cause. Is the streetcar limited in its uphill speed? Do the streetcar routes factor in additional dwell time at stops due to larger passenger capacity? Is the schedule of the streetcar padded due to an inability to maneuver it around obstacles the way a bus could? Does King County Metro simply run a more aggressive schedule on its bus routes? These questions are critical to making intelligent and responsible conclusions about what the future of streetcars in Seattle should look like, but are not answerable from the public data sources that I have consulted.

What to Do About It

The article mentioned at the beginning of this post alludes to three possibilities for Seattle’s streetcar program. It could be continued as is, dismantled, or the two lines could be connected. I don’t have sufficient information to make a full-throated endorsement, but can suggest some outcomes, conditioned on deducing the reasons for the streetcar routes’ low speed.

No solution that maintains streetcar service, whether it involves connecting the lines or not, should be considered if the routes cannot be sped up. If the vehicles themselves are ill-suited for faster operation on the grade of their routes, then anything short of dismantling the system is purposeless. The status quo is abysmal, and bolting two slow lines together with a middle section—likely to be comparably slow—will not magically yield a high-access route.

But I have found nothing that definitely indicates this lack of suitability. Rather than looking at the streetcar routes as streetcar routes that necessitate an all-or-nothing solution, consider them as just bad transit routes. Making a slow route fast is an undertaking that has known, if sometimes divisive, solutions. The slowness could simply be a matter of overly conservative scheduling. Giving the streetcars signal priority or truly dedicated right of way along their route could allow more aggressive schedules to reliably be run. I don’t think that the city’s current administration has an appetite for this, though. For that reason, I do see merit in suspending, but not dismantling, the streetcar system until the roads it traverses can be modified to speed it up sufficiently to not be an access laggard. The infrastructure that allows high capacity electric vehicles that have a longer lifespan and smoother ride than a bus is not something that should be impulsively discarded, even if the current execution of the service on it is wanting.

A permanently streetcar-less Seattle would represent a missed opportunity, but one that is temporarily without would not be a tragedy. The low journeys per in-service second and high percent replaceable indicate that rider impact would be minimal if the streetcar service were suspended. Bus service should replace the Broadway section that would lose frequent, all-day coverage, though. One of route 7, 14, or 106 could be redirected from its path on Jackson Street west of 12th, which is already covered by many frequent routes. Existing overhead wire for trolley buses, even providing a terminal at an unused layover spot, allows this modification with no changes to infrastructure. This may even be in-service time neutral for King County Metro. Otherwise, the Seattle Department of Transportation funds currently allocated to the streetcars could fill the gap under an existing program allowing the department to purchase bus service from Metro.

In the future, I plan on modeling what an effective streetcar system in Seattle would need to look like. How fast would it have to be? Is the proposed connection between the two lines the best adaptation of the present infrastructure, or are there better avenues for expansion? What would a complementary bus network look like? The data from this exercise would quantify the upside of making the potentially divisive infrastructure changes that would enable an improved system. This is entirely academic, though, unless the way that the city’s decision making around streetcars changes.

To create useful public transit infrastructure, streetcar projects must no longer be described as having a special, separable purpose. The South Lake Union Streetcar was framed as adding vitality to a redeveloping business district. The First Hill Streetcar was pitched as consolation for First Hill residents after their proposed Link stop was deleted, as a way to connect them to other Link stations. The hypothetical connection between the streetcars is now being labeled as the “Culture Connector.” This is a complete misconception of the purpose of public transit routes, and a distraction from what should be measured when introducing them. Transit routes aren’t intended to play a thematic role and help a person do a single thing. Transit networks are intended to help many people get to many things. A route adds value to the network based on how much it makes possible what was once impossible. That’s measurable, it’s calculable in precise terms in the planning stages of a project, and it’s what access measures. Planning streetcar projects based on a poetic hook and highly speculative ridership projections is a disservice to future riders. The promised fringe benefits of launching streetcar service won’t materialize when streetcars that don’t help people reach their desired destinations are empty.

I don’t want to hang the blame for an ineffective streetcar system solely on the city’s past and present decision makers. The public can’t hold leaders accountable when they are being told the wrong story about Seattle’s streetcars. Look no further than the article that’s been referenced in this post. Informing the public about the system’s problems and potential isn’t accomplished by getting quotes from a city council member and someone who is unironically described as opposing a specific type of box on wheels. The ridership, projections, and cost figures that the article cites are the most superficial of measurements. Where’s the analysis? Why is the streetcar set apart—positioned as something that is to be supported or opposed in whole—rather than assessed in the context of the network of which it is a part? In a way, the article unintentionally reveals exactly what’s wrong with Seattle’s streetcars. Technically what’s wrong is a lack of speed. But more importantly, what’s lacking is the sort of analytical scrutiny that treats streetcars in the same manner as any other transit route.