San Francisco: Sometimes Cuts Are an Improvement

Most North American transit agencies are cutting service this year, but there’s a huge difference in how they’re doing it.  My last post discussed the painful cuts happening at Tri-Met in Portland.  Here’s better news out of San Francisco, where service is being trimmed and shaped not just to save money, but to create a simpler, more frequent, and arguably fairer network.  The changes are informed by a long study and outreach effort called the Transit Effectiveness Project, which is finally bearing some fruit in this year’s harsh desert of funding.

President Obama’s adviser Rahm Emanuel likes to say: “Never let a crisis go to waste.”  Transit planners often say something similar: “Never let service cuts go to waste.”  Every transit system gradually acquires odd bits of route that really don’t make any objective sense.  They may have been added to take care of some noisy complaint, or they may just be obsolete services that have been superseded by something added more recently.  These oddities and complexities tend to hang on because if you try to cut them, the people who are used to them will complain.  There’s often just not that much to be gained — at least in the eyes of our political masters — from enduring these complaints.

So a lean budget season, now and then, can be a blessing, because it creates the political will to fix these things.

That’s the theme of the new round of changes in San Francisco.  The lines being eliminated all have some obvious built-in inefficiency or weakness.  Usually, the problem is that they overlap the market of other lines to too great an extent, so that the transit system is competing with itself.  Wherever you can replace two infrequent overlapping lines with one more frequent one, you’ve usually got a winner in total ridership terms.

The changes reduce the number of routes overall, thus making the system simpler.  (Whenver you hear an agency brag about how many routes it has, remember:  The number of routes is not a measure of the quality or quantity of your service; it’s a measure of it’s complexity.)

Even better, they’re re-investing some of the savings in higher frequencies on the routes that remain, so that overall mobility isn’t degraded much, and may even improve in some cases.

26-36A fine example is San Francisco’s ancient 26-Valencia bus line, the dashed black line in this SFMTA graphic.  Valencia Street in the Mission district is one block west of Mission Street.  Mission Street not only has two BART stations but also one of San Francisco’s most frequent bus corridors, composed of Lines 14, 14L, and 49.  When the 26 is running every 20 minutes, the Mission buses are coming every 3 minutes or so.  Most people figured out long ago that if you’re on Valencia and you don’t see the 26 coming, walk to Mission.  You’ll probably get where you’re going far sooner than if you wait for the next 26.

Given this, what, exactly, does the 26 accomplish?  Can we really justify such an overlap when so many other corridors don’t have enough frequency?  Many have asked.

San Francisco transit geeks have been questioning the 26 for decades, but it’s always been too painful to cut.  South of 24th Street, there’s a hospital right on Valencia, St. Luke’s, and hospitals tend to be a destination of mobility-impaired riders for whom a block’s walk is difficult.  This problem has now been sort of resolved by revising the less-frequent Route 36 to go by there, but it’s still a change for hospital patrons and doubtless a hardship for some of them.

Even on the core part of Valencia, people had mixed feelings.  Valencia is a happening street.  A few years ago it was restriped to reduce car lanes, create wide bike lanes, and generally slow down traffic to support a vibrant indigenous retail and cafe culture.  Some people felt the bus was an important resource for Valencia’s identity and vitality.

Inevitably, too, some people had chosen to locate themselves or their businesses because the bus service was there, and felt betrayed when it was removed.  This is one of the many reasons that transit agencies should publish Frequent Network maps highlighting their best services, as those are the most likely to be permanent, therefore a much better basis for decisions about where to locate if you want good service.

Finally, next month, the 26 will be gone.  The need to cut total service certainly helped push this recommendation over the top.  Also important is that they had good data on ridership, as part of the ongoing Transit Effectiveness Project (TEP), and that this change was done together with a range of similar ones all over the city. The TEP has proven to be a great forum for talking about all these odd scraps of not-very-useful route that were hanging around the system.  The result will be fewer routes, but in some cases, even more access.

14 Responses to San Francisco: Sometimes Cuts Are an Improvement

  1. Jarrett at HumanTransit.org November 6, 2009 at 4:32 pm #

    test

  2. Alon Levy November 6, 2009 at 10:47 pm #

    .tseT

  3. Jarrett at HumanTransit.org November 6, 2009 at 10:56 pm #

    test

  4. Jarrett at HumanTransit.org November 7, 2009 at 3:27 am #

    test

  5. Jarrett November 7, 2009 at 3:34 am #

    test

  6. Brent November 7, 2009 at 8:53 pm #

    Here is another question that is more of a challenge.
    Typically transit agencies have a walking distance standard saying something to the effect that a certain percentage (85%, 95%, 100%) of addresses (residents, etc.) must be within a certain distance of a bus stop (400 m, 500 m, 5-minute walk).
    What if, by restructuring the routes, you found that there were more areas of the city that exceeded this standard, but that it freed up buses that could then be applied to other routes to improve service and hopefully attract riders? Note that this is not necessarily a question to be asked as a budget-saving measure. This could be a case of a suburban area with lower-frequency service, and is it better to have east-west routes at 750-metre spacing running every 30 minutes or at 1.5-km spacing running every 15 minutes?
    Much would rely on the context, such as ridership on the various corridors, surrounding land uses and densities, the network design and available connections, etc., but I would tend to lean more toward accomplishing more with less.

  7. Jarrett at HumanTransit.org November 7, 2009 at 9:19 pm #


    Basic rule is:  People will walk further to better service.  (But if you propose to move their service further away due to this rule, they'll yell at you first.)

  8. EngineerScotty November 7, 2009 at 11:35 pm #

    A related question, thinking about some of the bus routes in Portland, and about the proposed serpentine LRT in Southern Maryland that you blog about in a separate post.
    What, in general, do you think about serpentine bus lines, which make numerous deviations from a straight course to provide at-the-door service to destinations which are along some corridor, but at a slight distance away from the corridor’s center? It seems to me if you make the route straight, you can improve service by two ways–a) the time to traverse the route is shorter, which enables b) headways to be improved for the same amount of money. (Or the operator to save money).

  9. Jarrett at HumanTransit.org November 8, 2009 at 1:03 am #


    Scotty, my point in the Maryland post is that once you've put all the destinations just a little bit off the corridor, you've made the corridor marginal for any kind of transit.  If you're going to do it all with one line, the deviation that such geography requires is just death on through-ridership.  Actually the deviation-penalty works much like the waiting penalty.  People are more content when they're moving toward their destination.  Both waiting and deviation defeat that desire, so people become more impatient, so they perceive the time as longer.

  10. Alon Levy November 8, 2009 at 3:56 pm #

    Brett, 1.5 km spacing is very wide for a bus. Most subways average 1.2 km, most bus systems much less.
    Besides, the only way to double frequency is to halve the run time. Doubling the station spacing isn’t enough for that. When you start with 750 meters, there may not be any way of halving the run time at all.

  11. Wad November 9, 2009 at 8:36 pm #

    Ah, a thread where I can cheaply plug an effort that I am doing on MetroRiderLA.
    I like to run a series called Open Source Transit, where a transit user would employ methods similar to planners to create services that would be better for the end user.
    My newest series is for LADOT.
    LADOT, which runs the Commuter Express and DASH buses in Los Angeles, is also restructuring services citywide as it is also facing revenue shortfalls and a $260 million operating deficit over 10 years.
    The solution I proposed, free to the city of Los Angeles, is for LADOT to refocus its DASH services to … as the kids today would say … pick up Metro’s sloppy seconds.
    LADOT has two problems: It is collecting too little of a fare (25 cents) and most of the DASH routes owe their productivity to overlapping portions of productive Metro services. This doesn’t really expand services or choice, and it splits a captive market.
    LADOT’s asset is that it can operate a bus for half the cost of Metro. LADOT, using private contractors (lower paid but union), operates a service for about $55/revenue hour. Metro’s cost is $110/hour.
    So I propose that LADOT take over unproductive Metro services (those typically running less than 30 minutes) within the city of L.A., and charge a fare that could be anywhere from 25 cents to $1, so riders would still pay less than Metro’s $1.25 cash fare.
    Metro would continue to focus on high-productivity lines, and LADOT would salvage services that may be on Metro’s kill list.
    The prologue is here, and has a poll for what you would think would be a good fare:
    http://metroriderla.com/2009/11/09/dash-restructuring-prologue/
    Jarrett, you’re a pro and you have some knowledge of the L.A. area. Stop by and look at the services I propose.

  12. Brent November 13, 2009 at 9:24 pm #

    Alon, to clarify, I mean the spacing between routes, not the stop (or station) spacing. That is, if you reduce the number of routes by reducing the route density, and then take the buses that would have been assigned to the eliminated route and use them to beef up the service on the remaining routes. The remaining service becomes more attractive for riders within its catchment area. For riders outside the “ideal” 400- or 500-metre catchment area you have a longer walk, offset by improved frequency.

  13. Bill Bryant November 19, 2009 at 11:42 pm #

    Hi Jarrett – Could not start Outlook email program, so will attempt to communicate this way. It’s been great to see your site, and would be fun to catch up sometime on Seattle events – drop me a line sometime:
    [email protected] (no longer at metro)
    [email protected]
    Cheers
    Bill

  14. Thewhitefolder.blogspot.com November 5, 2011 at 7:27 pm #

    I agree totally – sometimes a service, like a good pruning, is just what the doctor ordered. I had the privilege of interning at Atlanta’s MARTA as they prepared for their biggest service cut ever…
    Taking effect in September 2010, it ended up just being an 11% cut to bus revenue hours. However, it cut a system of 131 bus routes down to just 90. While there was a lot of fear before hand, (and I don’t want to downplay those who were inconvenienced,) I heard a lot of good feedback – people could still get where they were going about as easily.
    Overall, I feel like the MARTA service cuts made the bus system a lot leaner, and has allowed the agency to bounce back fairly well. They have added some service back and made some small changes… but it helps when you are tweaking on an overall system that is up-to-date.
    Interesting thing about that situation – they originally planned for a 34% cut to bus service, so they were really planning for a worst-case scenario, and then adding-back service. Of course, it helps when you have AVL-linked APC data, performance metrics, and regular performance reports… they were able to look at fairly precise ridership numbers by stop and time of day.