San Francisco: Go Forward by Going Back?

We’re currently working with the San Francisco Municipal Transportation Agency (SFMTA), which manages the city’s transit network called Muni, on options for how to develop their network as the pandemic wanes.  This piece is cross-posted with the SFMTA Blog.

Source: Wikipedia user Pi.1415926535

When a transit agency comes back from the COVID-19 crisis, should it aim to put service back the way it was, or try to put back something better?  That’s the question that the SFMTA will be asking the public later this summer.  

Muni started out as a service that took people downtown, and even today, most of the service is oriented that way. Meanwhile the pandemic accelerated ongoing trends that have shifted travel patterns away from a single focus on downtown and towards many locations across the city.  So are we sure we want the network to be exactly as it was?

Later this summer, the SFMTA will be sharing three alternatives for how service might be restored in winter and inviting the public to provide feedback on those alternatives. The input received from the public will help the SFMTA Board determine the pattern of Muni service to be implemented in early 2022. The three scenarios the SFMTA will be laying out for the public to consider are: 

  1. Return the Familiar Network​
  2. Build a High-Access Network
  3. Develop a Hybrid Network, balancing the best features of the first two.

The Familiar Network alternative would put back the routes people are used to from prior to the pandemic. But the service that people are used to isn’t always the service that helps the most people get where they need to go. 

The High-Access approach would shift some patterns of service to expand people’s ability to get to more destinations sooner. (See here for a full explanation of how access works.)

When we plan for high access, we aren’t just thinking about trips people are making, or the trips they made before the pandemic. We’re also thinking about all the trips they could make. Better access can mean more opportunities in your life. Right now, many people’s lives are changing as they find new jobs, get their kids started at new schools and explore new types of recreation. A high-access network tries to give people as many options as possible. 

What does a high-access alternative mean in practice? Here’s an example from the Richmond District:  Once Line 31 Balboa comes back in August, the Richmond district will have frequent east-west lines spaced every quarter mile. But Muni’s 2 Clement runs just one-eighth mile (a long Richmond block) from the frequent lines on California and Geary.  

Pre-pandemic map of San Francisco’s Richmond district transit services.  Note the consistent spacing of east-west routes every 1/4 mile, but the exception is Line 2-Clement in the upper right.  Source: SFMTA

Closer look at Line 2 Clement, and the more frequent lines 1/8 mile away on either side.

To measure the total access for people in a particular place, we look at all the trips to all the places they might be going, and calculate how long those trips take on the network. This travel time includes walking time, waiting time and riding time. In other words, we measure travel time starting from when you want to go, not when the bus comes.  

When we calculate access from points along Clement, we find that the 2 Clement doesn’t add much, because the nearby service on Geary is so much faster and frequent.  Even if you walk (or roll) slowly at 2 miles per hour, it would take you 8 minutes to get from Clement to Geary.  But your wait would be 5 minutes shorter, on average, because the 38 Geary is so frequent. You may save even more time if you get a 38R Geary Rapid, which is faster. At most, the 2 Clement service only saves riders a minute or two. And if you walk at a more average pace, 3 miles per hour, it’s almost always faster to walk to Geary than wait for the bus on Clement.

Such close spacing of parallel routes is not something the SFMTA provides in most parts of the city, so does it make sense to dedicate Muni’s scarce resources to provide it here? Should those resources go where they can measurably expand access to opportunity, such as by moving toward five-minute frequency on many lines?

I’ve talked at length about this high-access approach because it’s less familiar and therefore requires more explanation, but that doesn’t mean the SFMTA has already decided to do it. The choices between familiar and high-access approaches is a genuine question, and we’ll want to know what you think.

Finally, all of these choices are harder because the SFMTA faces severe resource constraints. It still faces a labor shortage and has lost much of its income from fares and parking revenues, not to mention the structural deficit that existed even before the pandemic.  So the agency can’t afford to restore all of the service it ran before the pandemic.  Even if the labor shortage were resolved (and the SFMTA is working on it), restoring 100% of the previously scheduled service would run the risk that just a year or two later, when one-time federal funding runs out, drastic service cuts would be needed that could leave us with even less service than we have now. 

Instead, it makes sense to offer only a level of Muni service that the SFMTA is sure they can sustain, at least until they find new resources to replace funds that have eroded over the last decade and fallen dramatically during the COVID-19 pandemic.

Next, we and the SFMTA will lay out exact plans for each alternative, showing the exact routes and frequencies that each alternative would provide. We’ll then analyze how each alternative affects access to opportunity.  We’ll look at this for the whole population, but we’ll also calculate the benefits and impacts for specific neighborhoods, for people of color, for low-income people, and for people who walk or roll relatively slowly.  

The SFMTA will bring this information to the community, so that everyone can think about the choices and express their view.  This will help the SFMTA Board reach a decision that reflects the values of San Francisco.  

11 Responses to San Francisco: Go Forward by Going Back?

  1. david vartanoff July 15, 2021 at 12:28 pm #

    While the 2 Clement looks like duplicative service, it really should be considered as a “short turn” relieving crowding on the 38/38R, and A or B rush hour Expresses as well as the 1 California A and B Expresses. . Whether that is a good use of assets is not a simple question. For decades, these routes considered as a corridor were second only in ridership to the 1st/2nd Ave bus route in Manhattan. (this is the route NY MTA operates instead of building the long promised Second Avenue Subway) One would rationally catch a 2 Clement to get a seat because the 38s and 1s would already be SRO inbound. One should remember that the headways shown are a fiction.

    For many decades, Muni has tailored budgets to fit the insufficient funding available, but claiming to field the full published schedules. The MOU with the drivers union has a clause giving drivers a premium for operating directly after a missing run. When Muni allowed the internal daily reports of missed runs to be public, one could see which routes were cheated, and how badly. Between a higher than transit industry average absenteeism rate, and a chronic staff shortage, they could not run the full schedule even with overtime. Typically the OT budget ran out before the end of the fiscal year followed by a spare parts acquisition freeze which exacerbated equipment shortages.

    Decades ago, SF Muni promised SF citizens a new light rail or maybe subway on Geary. But, then the project was “value engineered” to be a BRT route. However, Muni decided to do the Van Ness corridor first. That has become the far too typical patronage distribution with a few crumbs of actual project going on for years. Worse yet, the Van Ness project opted for center platform (dedicated bus fleet) useless for Golden Gate Transit which also runs on the street.

    Fast forward. Courtesy of the covid-19 disruption, Muni has been able to do Red Painted lane BRT Lite w/out lengthy hearings, NIMBY pushback, and the 38s are able to run faster. If they link the current sections of Red Lanes, and install traffic signal priority they will have 90% of what they want to waste mega millions on, before they could even file all of the paperwork.

    I point at this because Muni has a horrible record of poorly implemented but vastly expensive projects. Or, simply put, very little credibility.

    It may well be that another thorough route rethink is in order. There was one in the late 70s, and de facto another early this century. That said, Muni’s history of not activating transit signal priority on streetcar routes has led to rail being slower than buses on the same streets. Far more than restructuring routes, the Augean Stables need cleansing.

    • Sean Gillis July 16, 2021 at 7:23 am #

      So it sounds like the 2 is needed to relieve crowding because of operational/ resource issues on Geary and California. Might it not be better to dedicate the resources from the 2 to getting more buses on the overcrowded trunk routes? I don’t know SF at all, just my impression. 200 metre spacing is tight.

      Poorly implemented and expensive projects describe many North American agencies. The cost explosions that Alon Levy writes about at Pedestrian Observations are pretty discouraging. Add on decades of tight funding for expansion.

      Seems every big city has corridors that have been tagged for metro lines/ extensions for decades:
      Montreal = Blue Line extension, east and west. Pie-IX Blvd. was tagged for metro but is being built as BRT.
      Toronto = Downtown relief line.
      Vancouver = Skytrain to UBC.
      NYC = 2nd Ave., Utica Ave.

    • chris jones July 21, 2021 at 7:02 pm #

      Van Ness is center-running, but not center-platformed. Muni (fortunately) saw the light on that and changed the design to side platforms so that it didn’t need to acquire a new fleet.

    • RossB July 21, 2021 at 9:46 pm #

      According to the map, the 1 runs every 8 minutes, and the 2 every 20. The buses can’t possibly be synchronized (the 1 and 2 don’t form a spine) nor does 8 minutes seem like the point at which you get nothing from running a bus more often. Now if the bus was running every 3 minutes it would be a different story. At that point schedules are useless, and you might as well as many buses along that corridor as possible (including shorter versions of the same thing).

  2. david vartanoff July 16, 2021 at 10:12 am #

    Yes, the 2 Clement could be replaced by more buses running faster on Geary. If the 2 were abolished, would Muni actually move those buses to the 38 and 1? Maybe. Remember, this is a transit agency that a year or so pre covid, posted schedules for the Metro showing headways that forced conflicts–trains from different routes arriving at the same times on a single track.

    About ballooning costs. I admire Alon Levy’s efforts. That said, I despair at the level of corruption. Muni is several years late and millions over budget on the terribly designed but politically mandated Central Subway. A while back it was discovered that all of the installed rails were of a lesser (undoubtedly cheaper) grade than spec’d in the plans. Speaking as an electrician, on any construction site someone is responsible for “receiving” materials. That rask includes checking for quantity AND the precise items being delivered. If Muni itself had no one there, they are incompetent, if they had someone who failed to physically check the rails, why is she/he still employed? No one “noticed” until all of the rails had been installed.

    • Jarrett July 27, 2021 at 1:32 pm #

      For the record, there is little or no overlap between people now in charge at SFMTA now and people responsible for the Central Subway design.

  3. Chris July 21, 2021 at 10:11 am #

    Have the bus lanes on Geary allowed for an improvement in operation sufficient to allow for more buses without increasing bus bunching? It’s equally plausible to me that the 2 Clement should run more often to relieve pressure on California and Geary. Picking on the 2 just because it is so close to other bus routes without taking into account the ridership on it seems to be a very coverage based approach and I thought this site was more patronage based.

  4. Chad July 21, 2021 at 1:27 pm #

    A longer walking time should be not be discounted, especially in elderly and disabled passengers, and especially due to the hills of San Francisco. Do all the bus stops on the 1 and the 38 have benches and shelters to accommodate those passengers? And based on what david vartanoff has said in the comments above, I wonder if the 2 is more on time and reliable than the 38.

  5. Sean Gillis July 22, 2021 at 4:29 pm #

    I wouldn’t say that removing a route within 200 metres of another route is a coverage approach. One could argue that the 2 is a coverage route (regardless of its ridership, which would be expected to be high in a dense, mixed use area). The 2 is trying hard to bring a route – coverage – close to the people on and near Clement. This despite much higher frequency routes being available within a reasonable walk.

    Trying to maximize ridership – with a set amount of resources – means removing duplication and putting it elsewhere.

  6. Ruediger Herold July 26, 2021 at 3:43 pm #

    I have never been to San Francisco.
    I made some assumptions, deductions and conclusions by only looking at those maps, reading the text and the comments and remembering other blog articles or what I heard about bus network redesign in my own city (Luebeck, Germany).
    But my assumptions and conclusions might be partly or entirely wrong…
    First, the big question is probably not if the 2 is useful to the residents of Clement Street (I assume that East-West streets are called streets and North-South streets are called avenues. IF up is North which I also do not know.) Rather, what is the 2 doing Northeast of the map I can(‘t) see? I imagine the 2 as a coverage route doing something useful there but justifying only lower frequency and probably done with shorter busses. I also imagine that it is hard or impossible for longer busses to make the turn from California Street into Arguello Avenue and from Arguello Avenue into Clement Street (or opposite). Plus, I imagine Clement Street being narrower. There you would not want two busses of the same line going in opposite directions face/meet each other. But because of the low frequency of the 2 this is never a problem on this 3/8(?) mile segment of Clement Street. I think the reason why the 2 runs along Clement Street is that it could not easily be turned around when running on presumably busy California or Geary Street. It’s hard to imagine a street sign like “No U-turns – except for Bus No. 2”.
    So line 2 probably starts/ends so to connect to two frequent East-West and one (even more?) frequent North-South corridor within normal walking distance where it can turn around while saving on labour cost.
    I have more thoughts but I do not like too long comments, so enough for now…

  7. Robert S. August 15, 2021 at 12:21 am #

    Something that the map doesn’t show is the coverage east of Van Ness Ave. The #2 handles the busy Post / Sutter couplet just up hill from the Geary / O’Farrell couplet. At one time that pair of streets had service from three lines : #2 was long haul into the outer avenues (30’s), #3 Jackson and #4 Sutter were short haul with limited schedules.

    Also the next corridor to the north is a cable car line on California running over the brutally steep Nob Hill (some of the blocks have stairs instead of smooth sidewalks). You have to go another block to the north to get to the westbound #1 on Sacramento. That hill / ridge was enough of a barrier that they punched a tunnel through it on Stockton for a streetcar which is now a couple of well used trolley bus lines (#30 + #45).

    Further changes to the western end of the #2 may be justified. But the eastern end may be something of a sacred cow in view of the geography. As a resident in S.F.’s SOMA I’ve found the revised #27 Bryant, a fusion of the former #12 Folsom and #27 Bryant, to be very usefull. So I hope other prudent tweaks make it through the sausage factory.