One of the very first questions I expected when I published the Frequent Network Plan about two weeks ago was “How much frequency would we get at night?”  And, indeed, reader lakecityrider brought the topic up in the second comment to my original post.  I wrote at the time that I needed to crunch the numbers.

21 bus on 3rd Ave at night
Metro at night: Skeletal, and raggedly efficient. Photo by zargoman.

Now I’ve done that.  And the results, summarized in this map, show just how badly Metro’s night network has suffered in recent years.  Night service has borne the brunt of all the cuts and efficiencies in the last decade.  As a result there are just not a lot of hours to put into core-route frequency.  The existing all-day network in the area covered by the FNP uses about 324 buses; the existing night network uses only about 196 buses during early “night” hours (about 7:30 to 10:30 p.m.), with the number rapidly diminishing as the night wears on.  Further, there are no peak trippers at night that may be made redundant by a superior all-day network, so there are no “extra” buses to add to the new night network, either to provide more frequency or to add recovery time.  Night service does run faster than day service, but not by enough to make a huge difference; there is no alternative but to cut frequency substantially from daytime levels, and to cut a small amount of service entirely.

Speaking in broad terms, most 10-minute routes in the FNP would have to become 15-minute routes in the early part of the night, except for two that become 20-minute routes.  Most 15-minute routes become 20-minute routes, although there are several that become 30-minute routes.   The 30-minute routes stay at the 30-minute level, but several suffer truncations of varying severity.  A couple of through-routes that would be impossible during the day would be used at night to save additional hours.  Further details after the jump.

I haven’t calculated what would happen later in the night, but it’s safe to assume that frequencies would be roughly halved after 10:30 without new resources, and service would end around 1:00 depending on the route (with a few key routes running later, as they do today).

What’s more, even these reduced frequency levels rely on assumptions about recovery time that are a bit more aggressive than those I used for the daytime network.  A few runs staying out past the early evening would require some relief in the form of longer recovery, not all of which is accounted for in these calculations.  In order to provide it, frequencies might have to start dropping earlier than 10:00 or 10:30 on some routes.  (The 31 and 120 in particular have  recovery short enough to need extra relief in this proposal, but I would probably drop frequency earlier on another route to provide good frequency for longer on the busy and critical 120.)

The dispiriting conclusion: it’s just not possible to provide proper frequencies for a nighttime transfer network at the current level of resources without cutting a lot of night coverage entirely.  The good news is that it is possible, even in this skeletal environment, to fix some glaring pain points in the current network, even if we can’t get to a truly effective transfer network.  To name a few of the most satisfying outcomes:

  • As with the daytime network, no routes within the area covered by the plan would run less often than every half hour, at least before roughly 10:30.
  • Routes 8 (currently 8N), 58 (currently 358), and 120 would receive the bump to 15 minutes at night that they richly deserve.
  • Routes 5 and 40, together providing downtown/Fremont core service, would both be able to run every 20 minutes at night, for combined service every 10 minutes between downtown and the Center of the Universe.  Connecting buses at Fremont, including the 13, 28, and 31, would run every 20 minutes as well for predictable transfers.
  • Route 16 would run every 20 minutes at night for improved service to Wallingford.
  • Much of the new crosstown service would run every 20 minutes; this includes routes 52/55, 50, 31, and 71.  Routes 44 and 48 would run every 15 minutes as they do today.  Riders from the UW campus to Link would have a bus every 15 minutes on the 67/73 common corridor.  I prioritized providing as much frequency as possible on crosstown service because it will have the highest proportion of transferring riders.
  • Riders in the Central District would see night frequency improvements on all of their east-west service.  Route 3 would improve to every 15 minutes, while routes 14 and 2 would improve to every 20 minutes.
  • Several farther-out areas would see upgrades from hourly service to 30-minute service, most notably Magnolia, a number of areas in far north King County, and the multiple south Seattle communities along always-underserved routes 131 and 132.

This nighttime plan requires some aggressive and creative through-routing, which is much more practical at night than during the day.  Route 1, and half of trips on route 7, would need to use hybrid equipment, because they would be through-routed with non-trolley routes (131 and 59, respectively).  Routes 12 and 47 are through-routed, which will keep the 47 out of Pioneer Square during the party hours when 1st Ave S is impassable, but will take away the 12’s front-door service to Colman Dock.  The current route 8 is recreated at night (when Denny Way is much more reliable) by through-routing the FNP 8 and 6, although only half of route 8 trips continue onto route 6.

Routes 59, 69, 78, and 81 would be truncated.  Route 59 would not serve Upper Rainier Beach, Arbor Heights, Gatewood, or Genesee Hill at night (none of these areas currently have night service). Route 69 would be truncated in Lake City, with timed transfers to ST route 522 for continuing service.  Route 78 would no longer serve East Green Lake, terminating at Roosevelt Station, and would no longer travel west of Jackson Park on its north end (an issue that would have to be fixed somehow or other before the Lynnwood Link opening).  Route 81 would be truncated at Kenmore, again with timed transfers to ST route 522 for continuing service to Bothell.

Of course, at this already marginal level of service, there is not a lot of room to absorb further cuts.  Cutting 17% of the hours in this night network would result in a network with frequencies similar to today’s, but requiring more transfers.  That’s not a formula for building ridership or support for transit.

81 Replies to “Your Bus At Night, Only A Little More Often.”

  1. Thanks for offering up this plan. It sure beats 2-ish-hour headway on night-owl-only milk runs from Heck, filled with people who couldn’t find shelter for the night.

    I do have to question the assumptions about the service model for when and by how much service levels should shrink and grow during any time of day. There is a reason we say “peak hour” (a cultural relic), when it lasts for much more than an hour. Commute patterns have dispersed to disparate times of day. Bank hours anchor the peak of peak, but banks operating non-traditional hours are even starting to rust that anchor.

    You’ve probably noticed how buses are fuller during afternoons than in mornings. If efficiency remains important, the post-peak morning frequencies may have to give up a little in order to provide sufficient capacity in the afternoon pre-peak period.

    Night service, though, is the difference between having service and not having service at all. Where I currently live is a direct result of lack of night service. I’m one of the many who can’t count on a 9-5 latte shift. For most who can’t count on night service to access their jobs, the solution is a car. That means more cars on the road during the day, including during peak. That’s why I prefer the night map to resemble the all-day map to the extent feasible. Lost night riders are lost day riders. Without that resemblance, what’s left once subtracting commuters is the rolling homeless shelters.

    So, in terms of priorities, I prefer keeping the same route paths overnight to keeping a guaranteed headway all day. If we have to give up a little frequency between 9 am and noon to get David’s night time FNP, I would gladly give up morning frequency on some of my favorite routes to gain nighttime mobility.

    1. For the sake of simplicity, I kept the allocation of hours between day and night the same as it is now. My view is that the ideal network would have a base frequency that would apply throughout the service day (except maybe for one or two “mop-up” trips at the very end of the night), with possible increases during the very busiest hours. That is so the network is reliable, predictable, and simple. Unfortunately, there are just not enough hours to accomplish that now.

      There is indeed a lull in ridership between 9:30 and 11:30 a.m (with the notable exception of U-District routes). It could be worth exploring moving a few hours from that time period to the early evening (7:30-9:30 p.m.), when ridership tends to be pretty high.

    2. David’s FNP night map looks pretty good, considering what he had to work with. I like that it goes to north/northeast Seattle with more apparent coverage later into the evening. However, I’m forced to agree with Brent here.

      Unfortunately, David’s reply post is kind of what I expected when I posted my referenced comment. We’re never going to get Metro to agree to “give up a little frequency between 9 am and noon” for expanded night service. Doing that would annoy more people than it would improve service, even though I’d love to be one of the people whose area and travel it would improve. That appears to go ultra-super-double for anything approaching added Owl service. All of the Owl routes, with the possible exception of 83, are on the chopping block for the Great Service Cut of Fall 2014. I think that’s very telling and quite frustrating since night people pay the same taxes (and somebody has to be up all night doing this stuff).

      1. There will be no more Metro service between approximately 1am and 5am when/if the 2014 cuts are put into place. ALL owl service will be cut, including on the A,C,D lines.

        Service after 10pm will be drastically curtailed, to be equivalent to what is on the streets at 12am-1am now.

  2. One of the big benefits of this plan is reliability, not just frequency. With shorter runs that avoid bottleneck spots, you have a very good chance of having the bus actually match the schedule. This is even more important given the long frequency of buses at night. For example, You might build your schedule around catching a 73 at 7:20 in the U-District, knowing that the next one appears at 7:45 (as it switches to the night schedule). So you show up five minutes early (or your other bus does). With this system, the bus should arrive around 7:20, or maybe 7:25. You get home at a decent hour and wonder why it took Metro so long to keep their buses on schedule.

    With the old route, the bus might be twenty minutes late. So, instead of waiting five minutes, you wait 25 minutes. You get home at a lousy hour, and wonder why Metro sucks so much. You start thinking about driving, and ask your co-workers what parking is like around the office.

    1. Unfortunately, to achieve reasonable frequency for the night network, I had to adopt some additional through-routes. So reliability may not be quite as good as hoped, although some of today’s worst offenders in terms of night reliability are restructured such that they should improve a bit.

  3. David: Great job again of crunching the numbers. I have to wonder how your results take into account that Link will be completed to Northgate in 2021, as shown on the map, and to what extent you rely on ‘re-deployed’ bus hours in addition to service hours newly deployed via rail. In other words, is this the 2021 network, with just Metro hours reshuffled, or do you take as a given that fewer hours are needed in the corridors for buses, as Link is now the prime mover?
    I ask that because the actual experience with Central Link provided NO economies of scale before and after Link was started. (ref: ST ‘Before and After Study’)
    In 2008 (B4 rail), Metro ran 652k bus hours of service in the Link corridor, and raised that to 678k annual hours, three years after Link startup, raising Metro’s cost to $90M/yr. Link added another $50M/yr in operating cost, but the corridor ridership grew by very little counting all trips on bus and rail combined (26M/yr to 30M/yr).
    One of the selling points of rail is a train can do it cheaper than a bus, and offer better service to boot. It didn’t work out so good for the south segment.
    If Link is going to provide the lions share of heavy lifting in the north segment, I would expect your network would bank on that, and save the hours, or spend as needed to enhance the connecting network to rail – both day and night.
    Are your calculations separate from any new service provided by ST and if yes, why?

    1. This is a restructure that takes University Link, and North Link as far as Northgate, into account. It uses the same (to a pretty good approximation) number of bus hours as the current early-night network. Metro has always said that it won’t contract its service following the opening of Link lines — instead, it will redeploy its existing number of service hours to accommodate the Link line.

      Link cannot replace local service along its corridors; the stop spacing is too wide. It can only replace express service. Accordingly, the only line fully replaced by Central Link was the northern half of the 194, and the only lines fully replaced by North Link to Northgate would be the southern half of the 41 and the southern half of the 71X/72X/73X. Just as the 8 continues to serve local riders along Central Link’s Rainier Valley alignment, the 67 in my plan would continue to serve local riders along North Link’s UW-Northgate alignment.

      1. Make no mistake: ST’s total lack of interest in building corridors you can use without a feeder or a local shadow — rail in the mode of, y’know, every successful urban transit system on the planet — is directly responsible for the failures of efficient service redistribution that Mic cites.

      2. I suspected as much, but wanted to hear it from you before I speculated further.
        I agree with your comment that ” It [Link] can only replace express service.” That’s the tragic result of following the peak hour mentality of BART all these years (freeway oriented, park and rides, miles between stations, and a few oasis’ of TOD scattered about for the press corp to faun over) that d.p. and others have cited. Sure, it provides a relief valve for our freeways during the crappiest commute times of the day, but doesn’t really provide much incentive to drive most of our cars of a cliff. Wouldn’t a stampede of iron relics going over the Magnolia cliffs be a sight to behold?

      3. Funny, the Seattle portion of Link (south of 130th) is not freeway-oriented. It has better evening frequency than any Metro route, and it provides the full-time express that Metro has never been able to achieve. Try going from downtown to Columbia City on the bus, or (in ST2) Capitol Hill to Columbia City, or the U-District to Columbia City, or the U-District to Northgate, or Columbia City to the airport, and then tell me how long it took on the bus and how many times you had to transfer. Link is by far the biggest step in transit improvement Seattle has had in my lifetime.

      4. I think d.p.’s biggest complaint is that Link doesn’t serve the most important areas first, and have a nice grid system for serving those areas, not that it doesn’t serve the areas that it serves well. Outside downtown, the airport is the most popular spot for Link. Next to that, you have Tuwkila. That is pathetic. It doesn’t even go to the airport very quickly. The Tukwila stop is mainly for people working there (not for folks commuting to downtown). In other words, Columbia City is an OK destination, but it is not Capitol Hill. Nor is it the V. A. (a spot that somehow got missed along the way).

        Speaking of which, you are absolutely right about the next couple of stops being added. We should have started with U-Link, and grown from there. Go south (as we did) but include the V. A. Replace the 8 with a train, since it is not only the most congested route, but the most urban route (with the possible exception of U-Link). Add another line from the U-District to Ballard (via Fremont). Oh, and make every stop be within a 1/2 mile of each other. That makes a lot more sense to people like d.p. and me than building rail to serve the airport, Tukwila and Lynnwood.

      5. Back to David’s point, I think it is essential that we push for these types of changes when North Link is complete, even if we aren’t successful in changing everything. Changing the 7X and 41 routes is OK, but will only get a little bit of extra service. It also won’t make that much difference for people who live a ways away from Link.

        For example, let’s say you lived (or worked) close to Eckstein Middle School. How do you get downtown, once the beautiful, expensive new rail system is built. There is a station only about a mile away. Obviously, you can just walk to the station, but it is fairly hilly in there, and you will have to wait for a few lights, meaning it will probably take you more than twenty minutes. You can walk to the 64, which will get you right to the nearest station, but that only runs in the morning. You can walk the other direction, but none of the buses go to the Roosevelt station. Some of them go to the University, but not that close to a station. So, basically, unless you are willing to walk a mile or so, your life doesn’t change much at all once North Link is complete. I’m sure there are lots of folks who feel this way (and these are folks who are relatively close to the line).

        Now, compare that to David’s plan. You don’t even have to walk to catch a bus (showing that in some cases, you are not adding to the distance people walk, but subtracting from it). That bus takes you right to the nearest station. Not only is this very fast to downtown, the U-District or Capitol Hill, but really fast if you are going the other direction (to Northgate).

      6. I just don’t see that happening organically, or on the basis of subtle sideline persuasion.

        I’m sorry. I’ve been holding back my pessimism because I really don’t want to undermine the tremendous work David has done in thinking through this.

        But while David was back East for a few years, I moved to Seattle just as it was beginning to hit its mid-2000s growth spurt. I watched as the buses became busier, the streets became bogged-down, the legacy network ground to a dysfunctional halt.

        And what streamlining did Metro do? For years, none at all. Ballot initiatives promised slightly more trips on the same clusterfucked routes. Supposed flagship services were misaligned and watered down. An attempted restructure was backed off, the result a fence built entirely of weak links. For years, Metro insisted one-door only was working, even as you stopped on every block and missed every single light trying to get anywhere.

        This agency won’t significantly restructure unless forced. Forced by politics, forced by a binding referendum to prioritize ridership, travel time, and stewardship of tax dollars, or simply forced by looming cuts.

        There needs to be a version of this plan that still operates well, day and night, with 17% less money. Even if coverage gets a bit worse to keep frequency and transfers usable. That’s the only way we’ll see David’s network when the economy and political environment improve.

        If we keep plodding along, waiting for Metro to change, we’ll still have our crappy current network next year, and in 2021, and in 2030, and in 2075.

      7. There needs to be a version of this plan that still operates well, day and night, with 17% less money. Even if coverage gets a bit worse to keep frequency and transfers usable.

        If you think the political resistance to implementing a plan like this with current resources would be bad (and make no mistake, the problem is political resistance and the fear of it), wait until you see what would happen politically to a version of my plan subject to a 17% cut but that retained “[usable] frequency and transfers” on all remaining services. Whole parts of the city (Magnolia, Meadowbrook, west Queen Anne, Pigeon Point, Sunset Hill, 8th NW, various West Seattle neighborhoods) would likely lose all all-day coverage. Corridors would be consolidated to the point of requiring very long walks for lots of people. Frequency on extremely high-ridership routes would be low enough to jam buses all day. You’d have a plan that connected only dense centers anytime outside of rush hour. Maybe that is what you think represents the best start for rebuilding the sort of network you would like, but it’s sure not something that Metro is going to produce in response to a 17% cut, because it would be far more politically difficult than just nibbling at the current network yet again.

        Edit: In shorter terms: there’s no reason a 17% cut will make anything that could be built into my network more likely than it is anyway.

      8. d.p.,

        When you say “There needs to be a version of this plan that still operates well, day and night, with 17% less money,” that seems like the same faith-only thinking you rightly deplore elsewhere. David’s said that his entire network breaks down under the 17% cuts, and that makes enough sense to me that I believe him. (Of course, it’d probably work if we had your ideal Link with half-mile stop spacing, but we don’t.) So, could you give us the specifics? Which routes are you going to cut to make up the 17%?

        On the larger point, though, I agree that it’s very unlikely we’ll see this under Metro’s current political climate. So again, as a total novice in King County politics, I’ll ask: What measures can we take to change things, rather than just grumbling until we get fed up and move to New York or Boston?

      9. Well, since William brought up Boston and New York, there’s a pertinent question I’ve been meaning to ask David: why the clock-face adherence?

        Once you get to usable frequencies, it ceases to matter to the user whether the bus arrives at 9 or 10 or 12-minute intervals, as long as the frequency meets demand well enough to keep the system moving smoothly and with stable headways.

        Boston has no qualms about running 7-minute headways where needed in the daytime, while dropping to a worst-case 13 minutes at night to keep things spontaneity-enabling in a way that even 15 is not. Non-spontaneous-frequency routes are similarly arranged to meet demand: 23 minutes is still preferable to 30, and these are the only routed for which anyone checks a schedule, anyway.

        How many hours does your current plan waste by insisting on round numbers, yielding overlong layovers at both ends, or perhaps needlessly tight at one?

        There’s no way for a 17% tightened system not to be sub-optimal. But routes need not degrade by 33% or 50%. Ditch the round numbers, and the service prognosis doesn’t look nearly as dire.

        The implementation prognosis, on the other hand, improves significantly, because you’ve liberated Metro from its asinine promise of “no changes, ever”.

      10. why the clock-face adherence?

        For headways of 15 minutes or greater (what I see as the threshold for spontaneous travel), I think there is more value for riders in having a fairly predictable, easy-to-learn schedule than there is in wringing every last minute out of the system. I hated the T’s 23- and 28-minute headways, and I hate Metro’s 35- and 40-minute headways. Personally, I’d rather have 30-minute headways than 23-minute ones, and I’d rather have 20-minute headways than 18-minute ones. I could save a bit by using odd headways for a few awkwardly scheduled routes, but it wouldn’t add up to more than a bus or two spread over the entire 337-bus daytime plan.

        For headways of less than 15 minutes, you don’t need clock-face adherence, but there is much less likelihood of wasting too much time laying over, because adding or removing buses has a smaller effect on recovery. The knotty layover issues in my plan tend to be along 30-minute routes, not 10-minute ones. And round headways are a good marketing tool, so I use them whenever possible.

        Also, for what it’s worth: the night network here represents a 41% cut in hours compared with my daytime network (although it looks more like a 37% or so cut because of faster night running times). That should give you some perspective on what both a 17% cut to the daytime network and a further 17% cut to the night network would look like.

      11. I agree with your spontaneity cut-off at just under 15 minutes, but I thoroughly disagree with your stated preference for clock-face above that threshold.

        The vast majority of MBTA rides begin on a high-frequency service, and riders are usually unable to precisely time their high-frequency first leg to meet a low-frequency second leg. Running at 30 minutes rather than 23 would simply extend the average waiting time at the transfer point by 3.5 minutes, unnecessarily. For those whose first leg begins on a low-frequency service (mostly those coming from the periphery of the service area), a schedule is an easy thing to check from home or work or the plethora of smartphone apps that now exist. 23-minute headways will best 30 minutes there too.

        It’s always more productive to have buses rolling than sitting (as long as they aren’t rolling on deviations and inefficient routes, of course). I find it nearly impossible to believe that moving off of round-number adherence would only save you a bus or two system-wide.

      12. The vast majority of MBTA rides begin on a high-frequency service, and riders are usually unable to precisely time their high-frequency first leg to meet a low-frequency second leg.

        Only in one direction. In the other direction, those same riders are beginning with the low-frequency service. It’s nice to have your schedule memorized rather than checking an app before every trip. And, even when you’re starting with the high-frequency service, if you know the schedule of your connecting route, you will know when to catch the high-frequency service. It’s a tradeoff: improved legibility and improved convenience for regular riders vs. improved convenience for occasional riders.

        I find it nearly impossible to believe that moving off of round-number adherence would only save you a bus or two system-wide.

        On the daytime network, there are only a few routes with headways above 15 minutes, and most of those are pretty well optimized as is. I could save one bus on the 131, and maybe one bus from the 1 if I did some creative through-routing. That’s it. I could move the 71 to 12 minutes and save one bus that way, but the 10-minute headway there is better for Link connections.

        On the nighttime network, it’s a similar story… a lot of the 30-minute routes are pretty well optimized. I might be able to extend a couple of routes that got truncated a bit further, and that would be it. Looking through the schedule I can’t see a single route where I could save a whole bus by going to an odd headway. I could potentially take a couple of 20-minute routes to 18 minutes.

        It would take a more comprehensive restructure of my restructure to derive significant benefits from odd headways.

      13. d.p., I’m still waiting to hear (a) how you want to get “a version of this plan that still operates well, day and night, with 17% less money,” and (b) what you advocate doing instead of “keep plodding along, waiting for Metro to change.”

      14. The vast majority of MBTA trips period begin on high-frequency routes. Those that use infrequent services at all are in the minority. And because routes to places like Belmont and Dedham are far more likely to cross the threshold to infrequency in the evening, I feel comfortable in saying that these routes are very rarely the first leg of any journey.

        Clock-face has a place in the world, and that place is on commuter rails and commuter buses. Especially if passengers are driving or getting dropped at a P&R, it is very useful to be able to memorize the time that they need to be there (or to be at the commuter terminal when headed outbound).

        But the insistence on clock-face for local service reeks of vestigial one-seat thinking and biases developed before everyone had widespread access to the internet for at least their first leg. If you’re building a transfer-based network with spontaneous-level cores, clock-face is no longer useful to those making those connections. If it routinely causes longer waits, then it is doing measurable harm.

        William… I don’t have access to David’s spreadsheets, but again, the burden of proof is not on me here. Metro has a history of responding to great challenges and great opportunities alike by doing absolutely nothing. Route 2 buses still fight highway traffic on Spring every day. Front-door exit only disappeared because Metro was forced.

        The burden is on those who propose a sea-change restructure, while simultaneously supporting Metro’s campaign of intransigence, to address that cognitive dissonance.

      15. Actually, David posted a lot of his spreadsheets. I’m going to take his statement that nothing like the FNP is possible with 17% cuts at face value, meaning that you have the burden of proof when you say that it is.

        So if by “supporting Metro’s campaign of intransigence” you mean “opposing 17% cuts,” that’s why. The 17% cuts make it geometrically impossible to get anything like this. Without them, it’s at least theoretically possible, even if we’d need to get past tremendous political pressure.

        Again, what do you advocate? Let the 17% cuts go through, hope that Metro kills all Magnolia service to take us down to the skeletal dense-center-only graph David mentioned several comments ago, and hope that despite transit getting exponentially worse people will still vote for more money? If that happens, I think it’s at least almost as likely that Metro will “share the pain” and turn everything into a maze of half-hourly or hourly one-seat rides. Of course, if the cuts happen, I’ll advocate for David’s skeleton. But even that would be hugely worse than the current system and would therefore hugely diminish support for transit.

      16. Let’s talk about 15 minutes.

        Frankly, I don’t believe that 15 minutes is “so frequent you don’t need a schedule”. I don’t think anyone else believes this, either. Metro tried not providing schedules for every-15-minutes RapidRide, and it was a complete disaster.

        When a vehicle comes every 10 minutes or less, then I think it’s plausible. That means your average wait time is 5 minutes. Most people are willing to wait 5 minutes for almost anything. I certainly am.

        Outside of North America, my understanding is that 10 minutes is the most common baseline for “frequent”, too.

        This leads to an interesting observation. Certainly, 15-minute headways are better than 20-minute headways. But I would argue that 10-minute headways are *much* better than 15-minute headways.

        So, how do we handle 17% cuts? And how do we preserve a usable night network? Well, first, identify the truly frequent core, i.e. the routes which should come every 10 minutes for as long as Link comes every 10 minutes. This includes common corridors (e.g. 34/35). For now, we’ll say that all of them have 10-minute all-day service, with the exception of the few corridors that are truly capacity-constrained (like the 3 through Harborview).

        Next, the rest of the buses — including most of the buses that have 15-minute headways — get demoted to 20 minutes. (So some of the 7.5-minute corridors become 10-minute corridors.)

        This represents a 25% service cut for a large portion of the network. That’s probably enough to pay for the 17% cut right there. Of all the routes in David’s network, only 9 would be spared this reduction.

        At night, the priorities change. For the 10-minute routes, frequency is maintained until 10 pm (same as Link), at which point it drops to 15-minute headways until Link stops running. For the 20-minute routes, either frequency drops to 30 minutes pretty early, or routes stop running entirely.

        What this gives you is a truly frequent core — a set of trains and buses that come every 10 minutes all day every day, where “all day” has the same meaning for all services. The cuts are borne by the rest of the network.

        Just to be clear, I’m not arguing in favor of cuts. I’m just saying that I think priority #1 should be building a core network of routes with Link’s frequency, and maintaining those routes at all costs.

      17. “15-minute headways are better than 20-minute headways. But I would argue that 10-minute headways are *much* better than 15-minute headways.”

        That’s right. 15 minutes is the point where you start to feel like you can go to a bus stop any old time, 10 minutes more so, and with 5 minutes even transfers are painless. Each level up convinces more skeptics to take transit and support transit.

        But I also see a point in Jarrett Walker’s argument that agencies need to set some definition of frequent, even if it’s 15 or 30 minutes, so that they can make frequent-network maps with more than one or two lines on them. That both shows that there is a transit network that’s useful for something for more than just coverage, and acts as a baseline or starting point to improve further.

      18. Jarrett’s aim-low definitional threshold really only applies when he speaks of smaller cities, Mike. It helps a Bellingham or a Missoula turn a system so inconvenient that only the most desperate could possibly be bothered to parse it into a legible system that anyone with a little flexibility and a desire to leave the car at home might be willing to give a chance. It helps even more that only 3 or 4 routes in cities of that size could possibly qualify for such low-bar frequency. That makes the maps exceedingly simple.

        In big cities, when your aim is real modeshare and a full-time non-auto option for your citizens, declarations of “frequency” that fail to pass the spontaneity threshold* are substandard enough to not even be worth doing.

        Consider me another vote in favor of the Aleks plan. Pick the 9 or 10 truly vital corridors, and get them unerringly right. This will get you all around the city, reliably, at any hour, for the first time in 75 years. Service levels on the remainder of routes would be elastic as funding permits.

        *(For the record, Aleks, studies have suggested that the real spontaneity threshold is between 11- and 13-minute headways, with expectations varying based on time of day, on the speed and reliability of the awaited service, and on factors like climate and whether the system’s major transfer points are weather-protected. This is why I find it unacceptable to round down to a 15-minute baseline in situations when 11 minutes would be warranted and a bare-minimum 13 minutes could be achieved.)

      19. d.p., I see the logic of the Aleks plan, and how it could form the foundation of a better network. I just think that if my plan is the political equivalent of bench-pressing 600 pounds, the Aleks plan is the equivalent of shot-putting a minivan.

        Playing along, though, my critical corridors would be these, in roughly clockwise order:

        – 44
        – 15
        – 13
        – 5 (to 85th only)
        – 40 (downtown-Ballard segment only)
        – 58
        – 35 (north of BHS only)
        – 48
        – 8
        – 12 (to 23rd only)
        – 3
        – 14 (to MLK/Jackson only)
        – 7
        – 124
        – 120
        – 54 (to Morgan Junction only)

        All of these are essential to travel to or between either high-density, transit-friendly nodes or places with heavy concentrations of low-income riders. I’m fairly close to optimizing them during the day. At night, if I optimized all of them, there would be essentially no hours left for anything else.

      20. With regards to following clock-face versus non-clock-face (e. g. 15 minutes versus 13 minutes) I agree with d.p., I think we should maximize frequency, as long as we can guarantee reliability. This new system relies on transfers. The more frequent, the better for transfers. But reliability is just as important for transferring. If you plan out a route and you expect to wait three minutes, it is really aggravating if the bus you are on runs five minutes late. It is much better to simply have that bus run later, because at least then you can plan accordingly. At the same time, having all buses run at 15 minute intervals would make transfers really annoying. There are plenty of bus transfers that almost work. For example, the second bus leaves two minutes before the first bus arrives. If those buses work on the exact same frequency (e. g. 15 minutes) than you can’t adapt by leaving earlier or later. On the other hand, if the first bus runs every 13 minutes and the second bus runs every 15 (or vice-versa) you can do a little work (or let a computer do it) and plan accordingly. It doesn’t take a smartphone app to do this work, you can figure it out using the old paper schedules if you want.

        However, I disagree with d.p.’s overall pessimism with regards to this plan. I understand his belief that Metro (or the region) is slow to change (it seems like the only thing we want to build around here are stadiums). However, I agree with David, that the time to try and implement this is when North Link is complete (which represents a major milestone). This is when you can propose the change and see more people benefit from the change. But from a political standpoint, there is an even better time: when the buses get kicked out of the tunnel. It is one thing that to tell people that there one trip ride to downtown is over because we want them to transfer to Link instead. It is another thing to say that your one trip ride is over because the bus simply can’t move through downtown that fast anymore.

        Oh, and I think the shock therapy approach to these changes (making them via a cutback) would never fly (for the reasons mentioned). On the other hand, I think the opposite could make this system really popular. With, say, an extra 25% funding, you could extend the frequency well into the evening. You could also address whatever complaints come along. If people want coverage, give them a little (infrequent) coverage. If you want a one stop ride, give them a few slow, infrequent, unreliable rides. After a while, people will just ignore those types of buses. An audit will show that those buses don’t make sense, and we can cut those, and put the extra money into increasing frequency, especially along corridors which are likely to be very busy. I know everyone assumes that we will have cutbacks (and we probably will) but that is legal problem, not a political one. Despite all its warts (and there are some big warts) transit remains popular in the county and extremely popular in the city. We just need to figure out how to change the law so that we can tax ourselves to pay for the things we want.

      21. Listen, obviously I’m not anti-funding, or anti-new funding. I’m certainly pro-stable funding, and I’m definitely pro-funding-with-progressive-mechanisms. Because duh.

        But if it takes a bit of 17% Shock Doctrine to shake Metro from its straitjacket of accumulated bad routes and bad habits, then I can’t feel too badly about going down that path. Perhaps that means only 8 of the 16 corridors David listed above get the needed 100%-frequent-network treatment immediately, but even that lays the groundwork for getting all 16 implemented the moment funding recovers. By 2021, those 16 would be the established network core, and we could be increasing frequencies on secondary corridors.

        Metro has made a conscious choice to pursue a “SAVE MY BUS!!!!!!!!!!!!!!” strategy for fighting the 17% cuts. Give them that money, and David’s restructure proposal is DOA. Forever. I am unwilling to abide that strategy.

      22. d.p., the difference between your view and mine is that you think the Rodney Tom Chop would “shake Metro from its straitjacket,” while I think it would tighten the straitjacket.

        I see no alternative to a long process of grinding away at political opposition route by route to get small restructures through… just like Metro has been doing for the last 20 or so years.

      23. Yes, and my view is based on Metro’s actual rhetorical strategy, including explicit pledges to avoid “big bang” restructures in the future.

        GM Desmond has followed through on that pledge, “finding” money — money we obviously can’t actually spare — to avoid doing a restructure around RapidRide E, to the detriment of everyone who has to use our crappy North Central route spaghetti.

        Your view is based on what? The existence of RapidRide? That not-actually-frequent sub-faux-BRT that only exists because Metro saw it as a way to scam the feds for bus replacement dollars?

        There’s no there there. Metro won’t change except by force. They’ve said so themselves.

      24. Kevin Desmond’s statement shouldn’t be taken too seriously; it was in the midst of an overheated public response to a poorly-executed restructure. It also doesn’t exclude small restructures, which taken together have done much, much more than you give them credit for. Here are my best examples over the years:

        – RR B. It’s not BRT, but it replaced a spaghetti network of routes traveling between west and east Bellevue, and sped up service to a certain degree. What was there before? Two half-hourly routes and two peak-only routes, all headed in different directions from Crossroads. What was there before that? Two hourly routes and five peak-only routes.

        – The 358. It’s a frequent, relatively reliable core service. What was there before? A half-hourly local route, a half-hourly semi-express route, and an hourly all-day full-express route that no one rode.

        – The 345/346/347/348. These replaced useless hourly one-seat rides to downtown with two new crosstown connections and 15-minute service along two corridors close to Northgate.

        – The Shoreline 5. What was there before? Two routes with a single confusing number, which were theoretically spaced evenly but always ran together in both directions because of poor routing and scheduling.

        – The 120. What was there before? A half-hourly 20, which covered half the corridor with two nasty deviations, and a half-hourly 135, which snaked through the irrelevant Pigeon Point and SSCC on its slow way in from Ambaum. What was there before that? Hourly service on the 135.

        – The 140. Yes, it still needs its routing streamlined badly. But it replaced a hodgepodge of half-hourly and hourly routes, including an hourly one that ran straight through from Burien to Aurora Village via Bellevue and Bothell.

        – The 164/168 corridor. What was there before? An hourly route that served only the destination people didn’t actually need to reach. Now there is 15-minute service up and down the hill, and a packed half-hourly bus to GRCC, where people actually want to go.

        – The 150. A horribly unreliable half-hourly or hourly route all the way from Auburn to downtown turned into a reliable 15- or 30-minute route along a shorter corridor. (The same applies to RR A vs. the 174.)

        – The 40. Even you will concede this one was a big improvement over what came before.

        – The 31/32 corridor. Although I’d do it a bit differently, this created a crosstown corridor where there was none before. What was there before? Two routes that ran in common only between Fremont and the west U-District, on an irregular schedule. What was there before that? A half-hourly route. And before that? An hourly route that got stuck in traffic on 45th.

        All of these (and quite a few more) are proof that Metro will adapt when it’s under enough pressure. The pace of change is slow and you are right to be frustrated about it, but your conclusion goes way too far.

      25. I’m still smarting from the lies contained in TransitNow, whose dollars were — and still are — siphoned to reinforce the status quo.

        Then there was the two-year CRC, which came with the gift of the end of pay-as-you-leave*, but which was otherwise used to reinforce the status quo.

        And now there’s the 17% threat. That crisis will likely be avoiding with another contrained, regressive funding patch, which will be used for what? By pledge, to reinforce the status quo!

        You say that Metro will adapt under pressure. But then you insist that there should be inadequate pressure.

        *(no thanks to those within Metro, or on this blog, who still thought that clusterfuck was a cool-beans policy)

      26. only exists because Metro saw it as a way to scam the feds for bus replacement dollars

        Metro as an organization should not be faulted for “scamming”. The way federal subsides promote waste is a different discussion. The local transit agencies have no choice but to play by the rules they are given. Of course their stupidity constrains aren’t just foisted on them by the feds. There’s plenty of locally enforced stupidity. Nobody would rise to a position of leadership in Metro if their ambition was to run a customer oriented efficient organization. It’s just not structured that way.

      27. Metro will change because of several important (and not so important) changes that will occur when this proposal is slated to be implemented:

        New 520
        New 99 (Tunnel)
        North Link
        Buses out of the tunnel

        These changes are listed in order of least to most important. But all of them are likely to require (or at least encourage) change. The last one is a huge one. I just don’t see how you are going to move all of those buses onto the city streets without a major delay. At the same time, folks (like us) will be clamoring for better service, and David already has a model ready to be implemented. Again, does a guy really want to keep his one-bus ride if it only occurs once every forty five minutes?

        The pressure comes from citizens demanding change from our representatives. It does not come from people asking for cutbacks and hoping those cutbacks will result in more efficient service. I’ve seen that argument before, expressed towards the schools. Since we don’t like the way the schools are run, we will vote against the levy. No, sorry, it doesn’t work that way. If you don’t like the way the schools are run, talk to the school board. If they don’t do a good job, elect a new board. Meanwhile, give them plenty of money, because once the money starts running low, you can’t expect any innovation or change.

      28. “I’m still smarting from the lies contained in TransitNow, whose dollars were — and still are — siphoned to reinforce the status quo.”

        Transit Now couldn’t predict the recession of 2002, the runaway oil prices in early 2008, or the recession in late 2008. Metro has to keep the buses running. Expecting Metro to cut regular service to pay for Transit Now additions doesn’t sound very realistic.

        “the two-year CRC, which … was otherwise used to reinforce the status quo.”

        That was its explicit goal.

        “I just don’t see how you are going to move all of those buses onto the city streets without a major delay.”

        Those buses were on the streets before the tunnel, and when the tunnel was being renovated. It wasn’t gridlock, but it did take half an hour to get through downtown.

      29. Transit Now couldn’t predict the recession…

        You mean the recession that represents yet another example of Metro failing to pursue efficiencies and seek ways to provide better service on limited funds in a time of need.

        Metro could have reorganized around a network of core services in 2009. In fact, such a network existed in the plain text of the initiative — Metro could have used that as political cover.

        Instead, the agency didn’t even try to dispute 40/40/20, and blew 80% of the only new money spent from the initiative on exurban coverage services. The frequent-network boosts never happened, and every penny of additional sales tax we’ve paid on every purchase since gone into the black hole of Metro’s crappy legacy network.

        “the two-year CRC, which … was otherwise used to reinforce the status quo.”

        That was its explicit goal.

        And that is the explicit goal of our next funding patch. And the inevitable funding patch after that. And the one after that. And so on.

        This is precisely why a Shock Doctrine Reorg might be needed.

    2. This is clearly OT, but did you use 2009 operating costs for Link (when it didn’t start service until July) to derive your figures?

      1. Brent: You have proven to be a knowledgeable critical thinker over the years here on STB. We disagree at times, but I respect your opinion immensely.
        It is clearly NOT OT in my opinion. We can’t be designing transit networks in a vacuum. Bus, rail or both, they are just vehicles to move bodies. They have to complement each other to be efficient, which is why I tend to lump them together in my comments. Yes, different agencies, different funding sources, different oversight and different voter mandates, but the same goal – better public transportation in the end for all citizens.

  4. This great post points out how essential it is for us to get more funding for Metro. Once Link gets to Northgate, it will basically cover the bulk of the city. A large number of people will want to take a bus to Link, as opposed to those who just walk to it. One of the great advantages of Link is that it can run all night without costing very much (most of the cost was in building the thing). Metro has been focused on routes that they assume will be the most popular (e. g. commuter runs from the suburbs to downtown). They do well with these routes. But it is easy to forget that there are lots of workers who go a different direction. So far, Link serves these people really well.

    For example, more people commute to Tukwila, then from it. I doubt many experts predicted this. But it makes sense given the good suburb-to-downtown bus service that existed before Link. Basically, not that many have transferred over to Link for their daily commute into the city. On the other hand, the hundreds of people who work in Tukwila (cleaning rooms, waiting tables, etc.) now have a decent way to get there.

    The same is true for the rest of Link. It is easy to look at the Husky Stadium station and assume it is primarily for students commuting in for the day or maybe for doctors and nurses doing the same. But there will be a lot of nurses and aids working the night shift who will benefit even more. The same is true for a number of stations that are close to nursing homes and hospitals. Of course, most people don’t live right next to a Link station, nor are all the employment centers right next to Link, so it is essential that the buses interact well with them.

    1. The first sentence of the second paragraph should say “For example, more people commute to Tukwila via Link, then from it.”

      Obviously, more people commute from Tukwila to the city then vice-versa. It’s just that most of the folks are driving or riding a bus, not taking the train.

  5. David, did you intend for the two links in your second paragraph to point to the same map? And may I ask, at what time of day are those frequencies in effect?

    1. Yes, those should point to the same place.

      These frequencies are in effect from roughly 7:30 – 10:30 p.m. Before about 7:30 there are enough buses out to maintain the daytime frequencies. After about 10:30 frequencies would start dropping pretty severely. Like the current network, you would have half-hourly service on the core routes until 1 a.m. and probably hourly service on most of the others.

      Off the top of my head, there would probably be 2:15 a.m. trips on the 15, 58, 70, 124, 120, and 54, as well as versions of the 35 and 7 modified to allow for downtown transfers. Only the 58, 70, 124 and modified 7 would likely have 3:30 trips.

  6. I think that, fundamentally, this restructure doesn’t go as far as it could. A lot of area is covered by overlapping 20 or 30 minute routes that aren’t particularly coordinated. This is evident in west seattle, cap hill, and even the area north of UW where there are both north-south and east-west routes.

    I would prefer to see a starting point where most points in Seattle have no more than 1 bus route within .5 miles. If we can have sustainable 15-minute service on such routes, then I think the system will be much more usable/reliable than the current system. In your proposed network the ballard/phinney ridge areas are great examples of what I would like to see elsewhere, no overlapping north-south routes, and the crosstown routes are spaced much wider (44 and 71) to avoid duplicate service.

    1. It’s a valid criticism. The night network has to be fairly strongly related to the daytime network to be legible. If I were designing a network to stand alone at the level of service we have at night, there would be fewer, and different, corridors. The proposal here is an addendum to my daytime proposal, and uses almost the same network.

      I think as an ideal matter 1/2 mile is a bit too far to stretch the walk circle, especially since most people can’t walk in a perfectly straight line. My ideal network has a grid of frequent routes spaced 1/2 to 3/4 mile apart.

      1. I understand that this is based on the daytime proposal, but I’m wondering if a better approach to the daytime would be to build out an efficient skeleton that works at both day and night, and then add service into the gaps for the daytime.

        Regarding the route spacing, what mostly concerns me is areas where route spacing is less than 1/2mile. I agree that the ideal network is a grid at 1/2 or 3/4 mile spacing **on one axis** — but not both north south. If we have 1/2 mile spacing n-s, then you only need an e-w route once every ~2 miles. For example, if you look at cap/first hill, you have a way denser grid on both axis, which means you have a lot of very infrequent routes, the same is true in ravenna.

      2. I disagree that good spacing is only needed on one axis. It’s needed on both.

        The Ravenna grid is a bit closer together than I’d like, but the reality of the arterial street layout and the area’s topography preclude other options.

        Most of the routes in the Cap Hill grid are separated by steep hills — plus, the area is quite dense, and justifies slightly closer route spacing.

        You also mentioned the Phinney/north Ballard area. For the record, I’d add a crosstown route on 65th to the grid I have now, if only 65th were passable for buses. Such a route, and the major improvements to 65th that would be necessary to run it, will be in any “ideal” proposal I come up with.

        In southeast Seattle, mobility would also be seriously improved by a Graham Link stop and the Graham crosstown bus route to go with it.

      3. While it’s possible during the day that may be the case, but the fact of the matter is when we have limited service hours to play with, there’s a very real tradeoff. Instead of having corridors that are at 15 minute frequency, we have corridors that don’t even add to the walkshed of the network, at 20 or 30 minute frequency. If you can bring the headways down, then you don’t need many perpendicular overlapping routes because transferring becomes more tenable. Given that it’s likely that riders will have to transfer anyways, we’re just punishing users by giving them 2 routes every 30 minutes, rather than 1 every 15.

  7. Does the night plan assume Link will be shut down overnight? If so, does it consider how to replace Link destination pairings?

    1. I haven’t really thought that through, as late-night service is its own little weird world, but most of the Link pairings would be replaceable by operating the following bus routes:

      1. Modified 35 (routed similarly to current 49 north of Pine, and then back over First Hill on the south end of downtown), through-routed with route 67 at U-District
      2. Modified 7 (routed similarly to current 7)
      3. 124

      1. Any chance of fixing things so that a one-seat ride between the airport and downtown is still possible after Link stops running? We used to have this with the old 174. Asking people to transfer at TIBS at 3:30 in the morning is just crazy. Especially when downtown->airport is about the only origin->destination pair along the 124/A-line corridor that would get any ridership at all during that time.

      2. I’ve long thought that the 124 trips that run after Link is done for the night should be extended to the airport. The challenge is that doing so would (it looks like from the runcuts) require an extra bus.

      3. Does anybody on the 124 (or former 174) actually use the intermediate stops between the airport and downtown (Georgetown/Industrial District) late at night? If not, there’s you’re extra bus – simply replace late-night 124 trips with 194 trips (downtown->airport only) and you’re done. If we did it this way, the end-to-end running time would actually decrease.

      4. There are two major destinations along the 124 that wouldn’t be covered by a 194 but require night service: 1) Georgetown proper, which is still home to an extraordinary number of low-income people, and 2) the area of Pac Hwy around and immediately north of S 144 St. Those are destinations that have had well-used night service for many years.

        I think the service (both to the airport and to Georgetown and west Tukwila) is important enough that we should find the extra bus somewhere. I see the 124 as one of the most essential core routes in the system.

  8. There are also the big events that will blow the capacity on this plan. Take, for instance, this Saturday, when the Mariners play at 6:00, and the Sounders play at 7:00. Do the math on when each game is letting out.

    No Sounder service.

    If you are planning on going to either, be prepared to walk all the way home.

    1. There would be some improvement in the most in-demand corridors. Changing night frequency on the 5 and 40 from 30/60 to 20/30 would help a lot all by itself, as would shoving 71/72/73 passengers onto Link.

      1. Exactly. Those big events blow the capacity for every transportation system we have in the area (except maybe biking and walking). The difference is that with this design, it doesn’t take the entire system with it. If you think it is aggravating to wait a long time to get home from the game, imagine waiting an extra half hour for your bus when you didn’t go to a game (and just want to get home to feed your kids). Since your commute is no where near a sporting event, you might wonder what the heck is going on, until you find out later that it was due to an early game, then your reaction is, Fu** the Mariners, Fu** the Sounders, Fu** the Seahawks and good riddance to the Sonics. (By the way, I don’t feel that way, but I can sympathize with folks that do).

      2. There would be some relief just by the fact that Link could better distribute people to north of downtown to catch crosstown routes at other Link stations. There would presumably be 4 car trains running at such times.

      3. For what it’s worth, I’m still highly skeptical that you can provide any meaningful level of coordinated service to Fremont via 5-Dexter and 40-Westlake. It seems like there’s just too much risk of the buses drifting out of sync, and like the 43/8/48, the effective frequency will be much worse than the number of buses would suggest.

      4. The best you can do is route and schedule them for maximum reliability. This is the same situation you have now on the 26/28/131/132, FWIW, which are hardly models of reliability during the day but do better at night.

        Northbound, the 40 starts at Central Base and the 5 comes from the 21. The 21 is a fairly reliable route any time of day provided that it doesn’t get stuck at the Lander grade crossing. So northbound I think you can get fairly close to the schedule for most trips. Southbound is a bit more challenging on both routes. A 5 stop diet will help on the 5 side, as the problem on the 5 is passenger volumes and frequent stops more than traffic issues. The 40 is the bigger challenge, and you may be right that southbound reliability could be an issue.

        With some extra hours, I could do the 15-to-Northgate plan we discussed on the other thread, having the 40 originate at Crown Hill; that would solve the problem.

        I’m also still kicking around ways to extend the 28 all the way downtown to increase Fremont service, but haven’t found the magic pot of hours to do that either.

      5. With some extra hours, I could do the 15-to-Northgate plan we discussed on the other thread, having the 40 originate at Crown Hill; that would solve the problem.

        I’m also still kicking around ways to extend the 28 all the way downtown to increase Fremont service, but haven’t found the magic pot of hours to do that either.

        Well, let’s see.

        You get a few hours from not running redundant service between 85th/24th and Holman/3rd.

        You get a few hours from truncating the 13 at 36th and Phinney.

        You also get a few (very few) extra hours from truncating the 12 at 23rd, and reassigning the rest of Madison service to the 8.

        I’m still curious how expensive it is to deviate the 50 to SODO. There are other connections to Link, and other connections to downtown.

        Finally, it’s not clear if you considered the 101 and 150’s freeway hours to be “up for grabs”. If not, I wonder if the “Henderson restructure” that we’ve talked about before would save a few hours. I know it’s kind of cheating to reassign those hours to Seattle, but the freeway segments that would be eliminated are located within city limits… ;)

      6. Aleks, those sources of hours aren’t even close to enough. I mentioned to Bruce in the last thread that truncating the 13 at lower Fremont doesn’t get me even one bus. Changing Madison Park to the 8 would be a wash during the day and would actually cost quite a few hours at night, because the 8 runs more often than the 12 and it would blow up the nighttime 8/6 through-route. I might get one bus from truncating the 40 at 15th/85th rather than the Holman Road QFC, and maybe one more (although my 50 is currently pretty tight) from eliminating the Sodo deviation if I were willing to do that. But I need a lot of buses (I’m thinking at least six, probably more like eight) to break the 15/54 through-route and to extend the ten-minute 15 to Northgate.

        The 101 and 150 freeway hours need to be redirected to the south end. If we are going to suggest taking away Renton’s and Southcenter’s one-seat rides to downtown, we better have something to offer those areas in return. (And, for what it’s worth, the map I’m playing with right now cuts the one-seat ride on the 101 but not the one on the much-higher-ridership 150.)

  9. When we frame the conversation around not enough revenue we miss the big picture. Transit is a collective resource and we should fund it to address the need, at night, for frequent all day routes and for commuter service to our employment hubs. There is an enormous amount of potential revenue in this state. Our economy is thriving and growing and we have some of the wealthiest people in the nation who live here. The issue is not how much do we need to prevent cuts or restructure service at the same levels – the policies we should be considering is how do we take the money from those that are not contributing their fare share and use it to fund the programs and policies we need!

    1. We’re pursuing that too. The bottleneck is the state legislature, which hands out small drops of transit funding authority at a time, and each one of those takes a year of heavy pressure to get, or it’s tied to a big highway wishlist. The most effective thing here would be to convince the legislators to allow counties and cities to raise more than a token amount of money for transit, or to replace legislators with those who will. This means convincing legislators in rural/exurban districts, who view Seattle as a socialist menace that needs to be starved. If it weren’t for this, we wouldn’t be talking about a 17% Metro cut, or the fact that Community Transit lost its Sunday service five or so years ago.

  10. David, great work. Even though I don’t live in Seattle, the method you analysed King County Metro is one that could be successfully replicated in other cities.

    Have you considered the possibility of additional farebox revenue generated from the new customers attracted to improved service? For instance, the TTC in Toronto, which operates an incredibly frequent grid system with convenient bus+rail transfers, generates between 60-80% of its revenue from fares (depending on the source). As tax revenue is effectively “fixed” at a certain amount, any additional farebox revenue is like free money.

    How would the system look the services were privately contracted, like Las Vegas or Phoenix? Would the savings from privately contracting routes be enough to offset potential cuts or even add additional frequency?

    1. Additionally, the TTC operates what they call the “Blue Night Network” overnight. The guiding principle of the Blue Night Network is to have a bus within far walking distance – 3/4 mile – of as many people as possible. Each route operates every 30 minutes (the two routes that duplicate the subway run every 10-15 minutes or better), and there appears to be a limited effort to incorporate timed transfers where possible.

      With that in mind I quickly sketched out how a Seattle overnight network planned on Toronto principles might work.

      As Seattle is very narrow yet long, it only takes about 3-4 lines to sufficiently cover most of the North-South demand, with only 3 crosstown bus routes. Furthermore, as the main overnight ridership in residential areas tends to be people who have been drinking, and thus aren’t too time sensitive, large one way loops can be used to bring service within closer reach of more people without a large time penalty.

      1. The N48 shows the weakness in a pure gridded layout. Most of the other routes go downtown where people can tansfer to several other routes. The N40 doesn’t, but it works because a lot of people are traveling in both directions between the bars of Ballard, Fremont, and the U-District to their homes. But the N48 has really only one anchor, the U-District. Its other centers Lake City, Rainier Valley, and Renton have little nightlife, and in between them are just miles of single-family houses. So the overwhelming majority of ridership would be from the U-District to the ends. That raises the question of, How is somebody at the ends supposed to get downtown to transfer to a night job? The only transfer points are in low-density areas where you may be waiting 30 minutes, and it would be a three-seat ride to get to West Seattle, Queen Anne, or the airport.

        I guess I would address this on the north end by extending the N70 east to meet the N48 at U-Village. And on the south end, adding a downtown-Beacon Hill route to meet the N48 in Columbia City.

        And if they’re traveling to night jobs,

        The N40 works because there are a lot of young people traveling both directions between the bars of Ballard, Fremont, and the U-District to their homes. But the N48 really has only one anchor, the U-District. Lake City, Rainier Valley, and Renton have little nightlife

        But Lake City, Rainier Valley, and Renton have only a small amount of nightlife, and most of the areas between them are single-family, so the travel would mainly be from the U-District to the ends. But that raises the question, “How do I get from Lake City, Rainier Valley, or Renton to downtown so that I can transfer to somewhere else?” It would be a three-seat ride to West Seattle or Queen Anne, for instance. You can say, “But Chicago and Toronto have grids and they work fine”,

      2. Argh, I keep forgetting about the abandoned paragraphs at the bottom. So I guess I should finish the point about the difference between Seattle and fully-gridded cities like Chicago and Toronto. In those cities you can have a true grid where all routes go straight and you can get anywhere with a two-seat ride. Those long straight routes also go through several commercial districts that are larger than Seattle’s and generate more two-way demand. And even the parts in between those nodes are largely multifamily or higher single-family density than Seattle. That geography has influenced the businesses themselves over the decades, convincing them to locate all along the grid, and fostering a greater tradition of two-way transit use.

        Seattle, in contrast, has generally smaller centers, lower density in-between places, and barriers that force routes close together or make some grid lines impossible. These can be overcome in a day network with 5-15 minute frequency. But at night with a 30-minute frequency, all these problems come to the surface. People don’t want to transfer in the middle of nowhere, and they want their bus to go somewhere that’s a likely destination. They especially don’t want it to pass so-near-and-yet-so-far from a center or another route, when the immediate vicinity around the route is just houses, and only a small number of people live in those houses. So that’s why I think we need to warp the grid heavier here than in those other cities, and ensure that routes go to nearby centers even if they have to zigzag, and ensure that you can get to any other part with a two-seat ride.

      3. If this is an overnight network (2am through morning), where you’re trying to provide coverage, drop the N58 off of Aurora at least as far north as the Linden Deviation. The line on the map is deceptive — it looks like it provides coverage but of the handful of stops it has between Mercer and 65th, most are pretty inaccessible (there’s a reason the 358 doesn’t stop between 46th and the deviation, and it’s because you’re literally fenced onto the highway through Woodland Park). I mean, if you’re running the Ballard route up the counterbalance, clearly speed isn’t the primary consideration…

      4. Zmapper, thanks for sketching that out.

        I agree with Mike that for a truly late-night or all-night network in Seattle you need a common transfer point, and that point has to be downtown. I haven’t really thought the late-night network through, but my instinct is that it could mostly use routes from my daytime network. The exceptions would be for north-south travel in the east. I think I’d have to have modified 7 and 35 variants that would actually travel downtown, and I don’t think there would be any service along the 48 corridor.

      1. Why would he not get re-elected?… he pulled essentially the same s*it in the previous session and was re-elected. I think it’s reasonable to beieve that thsi is what the good people of Medina want.

  11. Great job David – while not ideal, the night network you describe is certainly a lot better than we have now. I appreciate the hard work planning this. My comments:

    1) Brent is absolutely right that good night service is important to get ridership during the day. People who can’t trust the bus to get them home will feel no choice but to drive all the way, no matter how bad traffic is during the peak-period segment of the trip or how good the bus service is during that period. I agree that transferring some amount of service hours from the daytime network to the evening is probably in order.

    2) The amount of nightlife is not uniform every day of the week. There is a lot more stuff going on on Friday and Saturday evenings than weeknights, especially Sunday and Monday. If we could give some routes a frequency boost just on Friday and Saturday nights, even if we couldn’t afford it every night, that would make a big difference. (Note that this is contrary to transit agencies’ instincts that Monday-Friday get a higher level of service than Saturday).

    3) What are frequencies expected to be on weekends during the day? Would it be the same as the evening headways described in this post? Or would it be closer to weekdays? What about the difference between Saturday and Sunday?

    4) Should some of the evening service hours be budgeted for special event overflow trips? I don’t think the provided frequencies on the 15 and 58 are going to have enough capacity to get everyone home from a Mariners’ game unless nearly everyone is afraid of getting stranded and chooses to drive. Especially since the games typically end right around the cusp of when service levels start to dramatically reduce.

    1. It should also noted that special event overflow trips could save both service time and passengers’ time by skipping stops downtown other than the stadium itself. For instance, overflow 58 trips could be in the form of an express variant that would get on the deep bore tunnel in SODO and continue all the way through downtown and Queen Anne without any stops. The first stop would be the critical transfer point of Aurora and 46th, followed by regular 58 stops from there until the end of the route. If we’re going to spend $4 billion on a car 99 tunnel, we may as well use it for transit the one time it actually makes sense.

    2. asdf, if it were up to me, I would definitely transfer a bit of service from the day network to the night one, and I would also increase both frequency and span on Friday and Saturday night. This is more about reimagining the network than about those other goals, though.

      This network would have the same frequencies as my daytime map on Saturday and most likely the same frequencies as my nighttime map on Sunday. The problem with Sunday is a union work rule problem. All shifts on Sunday have to be full-length “runs,” which makes efficient scheduling essentially impossible. It’s easy to see why the union demands it, though, when you look at the Saturday schedule — which has all but the most senior drivers working “combos,” i.e., split shifts.

      The real solution, which is mulishly resisted by the old guard at the union but which I think will happen when the next generation takes over, is to abolish the limitations on when part-time drivers can be scheduled, while keeping the absolute limit on the number of part-time drivers so that the full-time positions are protected. This would work out well for everyone. Part-timers would have more schedule flexibility. Full-timers would have fewer combos and more slots for weekend days off. Schedulers would be able to craft Sunday and weekday night schedules that made more sense and used labor more efficiently.

      Yes, we should have a budget for special-event frequency. No, I don’t want to take it out of the regular network. I’m not interested in prioritizing event attendees over people working night shifts.

      1. Events are something I can see both ways. On the one hand, if transit is about carrying people and relieve congestion, special event service is an integral part of what transit is about. There is also a part of me that feels squeamish about agencies making value judgements about the purposes of trips (i.e. that entertainment trips are less deserving of transit support because they are not as essential as work trips).

        On the other hand, depending on an agency’s policies, large events either:
        A) Impose a significant unfunded mandate to operate lots of extra buses around the event, most of which, would be useless to anyone not attending the event
        B) Impose a significant burden on the traveling public, by all modes, due to congestion resulting from the fact that the lack of transit service means nearly everyone is driving. If the event is in the evening, this congestion falls on top of regular afternoon rush hour, even if transit service during rush is plentiful. This is because for a round trip, transit service is only as good as the weakest link – if it can’t get people home, it’s not much good for getting them there, and if people need a car at the stadium to get home, there’s not much choice but to drive it there.

        Ultimately, I think the real solution is that anytime you have an event where tens of thousands of people are all arriving and departing around the same time, whoever is organizing the event should be required to pay for extra transit service to accommodate the crowds, just as they are required to pay for extra traffic cops to accommodate the crowds. The cost (minus whatever can be collected at the farebox) would simply be passed on to people attending the event through higher ticket prices.

        It may also make sense to impose ceilings on the number of parking spaces allowed at event venues, based on how many cars the surrounding roads are able to carry at once without causing gridlock. If the venue is to hold more people than what the roads and parking lots can handle, the event sponsors should pay for extra buses and trains to pick up the difference.

      2. @asdf: more importantly, access to a good variety of big events is one of the things that makes city living desirable/palatable.

Comments are closed.