From Human Transit:

“I think frequency is an overrated thing. Let’s say there’s a 20-minute [wait].  You can look on your phone, wait inside and have a beer.” — Portland Streetcar Citizens Committee member Peter Finley Fry, justifying the 18-minute frequency of the Portland Streetcar’s new Eastside loop, quoted last August in Willamette Week.

I don’t mean to pick on Mr Fry — I’m sure he is a person who sincerely wants to make transit in Portland better, and thinks he is doing so — but this quote is perhaps the crowning example of an incredibly misguided, but surprisingly prevalent strain of thought among the political leaders, advocates and managers of transit systems in the northwest; one which, until we slay it, guarantees we will flail ineffectually (and at potentially great cost) in our efforts to provide an alternative to near-universal car ownership by working-age adults.

The highway engineers, social engineers, and car manufacturers of the 1950s, who overthrew the entrenched dominance of public transit virtually everywhere in the United States, used many different tactics and appeals to do so, but one thing they certainly didn’t do was tell people they would have to wait 15 or 20 minutes before they could start their journey, so they should just cool their heels and read the newspaper for a bit. Quite the opposite: they promised freedom to travel where and when you wanted.

If we wish to emulate their feat, and install transit as the (vehicular) mode of choice in the dense parts of our cities, we need to internalize their language, their promise (go where you want when you want), and a proper understanding of how frequency affects travel time for spontaneous trips within a city. On transit, if you wish to travel spontaneously, or arrive at a particular place at a particular time, the average delay is half the headway. At 18-minute headways, that’s nine minutes of expected delay.

If you are travelling within a small city, and have to wait much more than about ten minutes for your transit vehicle to show up, taking transit has probably lost out to driving no matter how fast the vehicle goes, and can now only compete on the convenience and savings of not having to park, or saving you the cost of a cab if you are unable to drive. If, as is the case with the streetcar, you’re travelling in congested, city-center traffic and for a fairly short distance, the streetcar will need very high frequencies to make using it worth your while, compared to walking. If the goal was to make it easier to get around Portland, the Eastside streetcar wasn’t worth building until enough vehicles and operating funds were available to make it run a whole lot more frequently than every 18 minutes.

We can apply this understanding of frequency and travel time to find some quick, easy and impactful ways to improve transit service in Seattle. As an example, most of the city is consigned to 30-minute headways in the evenings and on Sundays, giving an average wait time of 15 minutes. Improving those routes to 15-minute headways would provide a typical travel-time savings of 7.5 minutes, a speedup which would be hard to match by making the vehicle faster, short of building a subway at tremendous expense. Improving RapidRide C/D to five minute headways would save five minutes per trip, which would almost make it worthy of its name.

In general, increased frequency, rather than increased vehicle speed, is going to have a greater impact on your ability to get around until headways get quite short, and headways in Seattle are mostly so long that the smart way to improve mobility is to run more service, and focus capital money on improving reliability so vehicles don’t bunch together. Conversely, understanding frequency can also tell us when it makes sense to make capital improvements rather than throw more vehicles at the problem.

As an example of that case: throughout the weekday, Routes 3 and 4 together provide 5-7.5 minute service from downtown to Harborview via James St (although they tend to bunch together somewhat). With an average wait of around three minutes, throwing more trolleybuses in there isn’t going to do much to reduce travel times, so it makes sense to look for ways to make those trolleybuses go faster. The problematic part of James St is basically an unfixable car sewer, but nearby Yesler provides a very fast, reliable path to Harborview, saving about three to seven minutes per trip, at the cost of slightly reduced access to First Hill. Building trolleybus wire on Yesler is thus likely to be a very good strategy to improve that corridor, rather than improving frequency.

Not understanding (or not caring) about the fundamental relationships between frequency and vehicle speed and travel time is the equivalent of being a cartographer who believes the world is flat (or that the shape of the world doesn’t matter). If a person don’t understand it, and why it matters, they will not be able to reason usefully about transit; they will spend money to fix problems that don’t exist, while missing problems that do; the people who they wish to serve will suffer for their efforts. For as long as our agencies are lead, advised, or influenced by people like Mr Fry, who don’t get this, we’re going nowhere fast.

141 Replies to ““I Think Frequency is an Overrated Thing””

  1. I agree with your points on frequency, and would add some practical restrictions to make that happen.
    1. Metro is broke, so any gain in frequency is at the expense of someone else currently riding the bus. We talk about cutting unproductive routes, and putting the hours into others, but in reality, we take productive routes and ‘dumb them down’ with half the efficiency they had before (eg. MT174 v. RR-A).
    2. Running wire is fine up Yesler until you hit the new FHSC folly at Broadway. That’s where it will end because of the conflict of operating two different systems under the same overhead. Pick one and do it well, but that horse is out of the barn now – so no wire on Yesler.

    1. As written previously, there are solutions to differing catenary designs that can accommodate LRT and trolley bus, especially if they operate at similar voltage/amperage levels. If you stagger heights, ensuring the positive trolley wire is kept at an altitude higher than the negative trolley/lrt wire, the two modes can operate interchangeably.

      1. I didn’t mean to imply it can’t be done. Mixing the two is an engineers worst nightmare. That’s why FHSC takes a circuitous route to get from Broadway to S.Jackson so as to avoid the special work along the way and why it runs on battery power to avoid other conflicts, such as 5th/Jackson.
        From a practical point, you’ll never convince Metro to run wire up Yesler now.

    2. Bruce is proposing running the 3/4 up Yesler “to Harborview”, not Broadway. I expect he would have the route follow 8th and then 9th to Jefferson.

      1. I’m wrong; apparently he’s proposing to bypass Harborview. In the original post he has the phrase “crush loads from downtown to Harborview in the AM peak”, so I don’t know how he can propose such a thing. There are hundreds of transit-dependent folks who use that stop every day.

        What is wrong with using 8th and 9th to get between 9th and Jefferson and Yesler? Grant, it’s a pretty steep hill, but nowhere near as vertical as James and Queen Anne. Hills are where trolleys shine.

      2. You were right the first time. Yesler->8th->9th->Jefferson is the proposal. Front door service to Harborview; avoids the I-5 queuing mess.

      3. OK. After having traced links back to a CD News story it’s become clear to me that the 8th/9th routing between Yesler and Jefferson is what we’re talking about. So mic’s concern about the streetcar conflict is a non issue.

  2. Those who persist in the delusion that Portlandia(tm) is “THE” Transit Utopia are in dire need of corrective measures to resolve said delusion.

    On a serious note, I believe that all fixed-guideway transit should operate at 15 minute or better frequency during the daytime.

    1. 15 minutes is way too low of a threshold. All fixed guideway transit should operate at 10 minutes headways max for 16 hours a day. If you’re spending tens to hundreds of millions a mile you better operate at a level worthy of that investment.

    2. Unfortunately, MAX, BART, and some other systems run at 15-minutes per line, and depend on overlap for greater frequency. This means you get great service only where two or more lines overlap, which is a disincentive to live in areas with only one line.

    3. Oh, San Jose’s VTA does too. And evenings and weekends it runs 30 minutes on the Mountain View – Winchester line. Of course, that line is also probably the worst in the country for housing/retail near stations, so that explains its infrequency (and why its peak hours look like Link in the evening).

  3. I don’t doubt the STB consensus on the importance of frequency is based on sound research about the habits and choices of choice transit riders (while I’m not familiar with it myself). But here’s my question for those who remain deeply committed to this position: does the arrival and increasing improvement of real time information change the equation for you somewhat?

    I know it’s a mistake to assume my own preferences are widely shared (the “Bailo problem”) but my own preferences have changed (and are changing) considerably as this changes, and I don’t think I’m alone. Example: when I’m in Seattle I live between two lines that usually have 15 minute frequency (10 minute walk to one, 12 to the other) with N/S service. I also live a block away from another N/S route, with 30 minute frequencies. This is fitting the density of the corridors, which is lighter on my immediate one, but denser on the ones that require a modest walk. The more frequent buses are also slightly faster (although are no more reliable) than my local 30 minute bus for downtown travel. Given the general view that Metro erroneously prioritizes coverage over frequency/reliable on major corridors, I can imagine there would by some sympathy in these parts for cancelling my 30 minute local, and using the service hours to increase the frequency on the denser corridors from 15 to 12 minute headways. (The devil is in the details, of course, but my sense is that the fine people here are at least on board with such a change in principle).

    10 or even 5 years ago, I would gladly have made that change. A 30 minute bus with unreliable timeliness is incredibly frustrating; I’d happily take the short walk in exchange for less standing around at the bus stop once I got there, and greater reliability. Now, though, I wouldn’t support such a service change at all. The arrival of real time information has transformed the 30 minutes bus’ usefulness for me. I start keeping an eye on my phone 15 minutes before it’s due, and if it’s late, I use the time to do a little cleaning, keep reading my book, or whatever. I almost never wait more than 2 minutes at the stop. OBA is less useful for calibrating when to leave the house for a 10+ minute walk, as buses fall further behind, etc.

    I don’t think this is just me. I have many friends who are users of the old 15/D line, and while I was away when it launched, I was in regular touch with them. Like many here, they were…underwhelmed by this new service, the largest complaint was the lack of realiable real time information. All of them valued that real time info FAR more than they valued the move from 20–>15 minute headways. The key for them was not being able to get a bus as close to when they want it as possible, but being able to use technology to minimize time spent waiting at the bus stop. Getting stuck standing the rain for 15+ minutes multiple times a week rather than a few times a year is a much greater negative than having the next bus come 3-5 minutes sooner on average is a positive for them, and I have to say I agree.

    I agree more frequency is better, all things being equal. But frequency trades off with other valuable things, and it seems undeniable that reliable and accessible real time info makes frequency slightly less valuable relative to other potential priorities. I am open to the possibility my own preferences are idiosyncratic such that I exaggerate the significance of this, but I wonder I would like to hear about how–if at all–this technological change has impacted your thinking about frequency.

    1. I agree with you. Since OBA I find that I’m almost never waiting more than a few minutes for a bus. Typically I’m either at home or at my office and can leave for the bus when I know it’s coming.

      I support increased frequency as well but I find that I don’t mind less frequent service as long as when I know when the bus is coming and when given the choice between a less frequent bus directly to my house versus having to trust Metro not to screw up a transfer I’d rather stick with the current scenario.

      1. “Since OBA I find that I’m almost never waiting more than a few minutes for a bus.”

        Let me guess: no transfers on your commute, right? I thought so. See my response below.

      2. I don’t have any transfers currently and will be fighting like hell to keep it that way.

        Sorry to hear about your experience on Saturday – isn’t that a time when the RRD is supposed to be at 15 minute frequency?

      3. “I don’t have any transfers currently and will be fighting like hell to keep it that way.”

        So you see the problem with ‘OBA solves everything’ right?

      4. Interestingly I often have a couple of transfer choices and leverage OBA to inform those choices. For example, if I’m on the 124 I can opportunistically transfer to the 60 to get to Beacon Hill. But if the next 60 is more than 6-7 min away, it doesn’t make sense and I’ll ride the 124 to Holgate and walk home.

        But let us be clear: OBA is merely a mitigation for the limitations of 20-30 minute headways, spaghetti routes and lack of reliability. It certainly doesn’t solve any of these problems. Also, OBA is only as good as the data provided by Metro, and that data is far from perfect. It really depends on the route as to how accurate it is.

      5. Oh, I don’t mean to suggest OBA fixes everything, or close to it. But I do think, at a minimum, it marginally increases the utility of low to medium-frequency (~30 min +/-10) routes with less than ideal reliability.

        Not having to think about a schedule and having waits of 10 minutes or less enhances freedom. But being able to time one’s travel to avoid ever waiting more than 3 minutes, with a bit less flexibility about when you leave the house, enhances freedom too. Which one enhances it more isn’t an empirical question; it depends on time constraints of the traveller as well as the subjective way in which different minor inconveniences are experienced.

      6. “So you see the problem with ‘OBA solves everything’ right?”

        No as neither I nor anyone else said that. Kind of a troll move to put something in quotes that no one has actually said.

      7. djw – yes, tools like OBA are awesome to help make informed decisions about your trip, particularly a spontaneous one. As others have pointed out, the mere availability of real time arrival information can reduce the perceived wait time and rider anxiety. The definitely enhance our transit system, whether for hourly routes or routes that run every 5 minutes.

      8. You promised to “fight like hell” to maintain your one-seat ride to your preferred destination.

        This is not only inherently inefficient, but it reveals your lack of experience with being able to travel many places freely, rather than being limited to a single smartphone-aided travel destination.

        Matthew was correct to suggest that poor systemic choices result from your stated bias.

      9. “This is not only inherently inefficient, but it reveals your lack of experience with being able to travel many places freely, rather than being limited to a single smartphone-aided travel destination.”

        No, it doesn’t. It does reveal that, like many other folks who use Metro, my primary transit use is between home and work – both of which are fixed destinations. Most of the places I go on a daily basis are within a quick walk of either location. I really don’t see how the ability to “travel many places freely” gets me to my office any quicker.

      10. OneBusAway is useful, but there are limitations. For instance, my typical commute involves a 15 minute ride on a kick scooter, followed by a bus ride. Checking OneBusAway to time when to leave home would be pointless because it would not account for delays the bus would encounter after I leave home, which can be several minutes. So instead, I leave home without even looking at OBA until I actually reach the bus stop, instead taking advantage of the very good frequency (a bus every 5-10 minutes) to know I won’t have to wait long.

        Also, for any trip that involves a transfer, OBA is a very poor mitigation for long waits caused by long headways. Maybe you can use it to decide to visit a coffee shop between buses, but it won’t decrease your total trip time by one second.

      11. I find OBA is highly unreliable for commuter routes — buses often arrive much later than the predicted times, I think because the prediction is based on when the bus left the last stop, so backups on the freeway cause problems. On the other hand, buses sometimes make up time, and given that Metro drivers will leave up to 5 minutes ahead of schedule time, trying to get there right as the bus arrives can be a big mistake. I find my average morning weight for the 197 has to be around 15 minutes if I want to be sure of catching it. I’ve also seen it run over half an hour late.

    2. In addition to the points about transfers, this only really works if you have either a smartphone or computer access at whatever place you’re leaving. I’m very grateful for OBA’s SMS functionality, but it depends on knowing the stop number. Unless the relevant stop is one that I use all the time and therefore have bookmarked/memorized, I need to be physically at the stop to figure out the number so I can text OBA. This is still useful, but doesn’t let me time my departure to wait inside in comfort. If you don’t have a cell phone at all, or forgot it that day, you can’t access the realtime info unless you’re at one of the major downtown stops with an OBA display.

      Don’t get me wrong, realtime info is very important. But it’s no substitute for frequency, and it’s less useful for anyone without a smartphone (i.e. poorer and more transit-dependent parts of the population, mostly).

      1. Even if you have a smartphone, you might sometimes accidentally leave it at home. Or take it out to consult OBA and realize you have a dead battery. I’ve made that mistake several times.

        Also, if it’s cold outside, taking my gloves off in order to use my smartphone is something I would really rather not do if I don’t have to.

      2. FWIW, OBA has a rather extensive touchtone interface which includes stop lookup, so you don’t need to be at the stop or have it bookmarked/memorized in order to figure out the number and get arrival information. There are several big drawbacks—it can be a hassle to find stops deep into the route; there’s no way to look up non-numbered routes (i.e. RapidRide)—but it’s pretty good.

    3. It’s not just the wait at the stop itself, but the waste of time in total (even at home/work). Rather than looking at how long you have until you can leave, consider the problem from the other side: At what times can I arrive at my destination? With 30 minute frequency you don’t have a lot of options and can be forced to be 15 minutes early (or late) even if your bus is on time. Having real-time info does very little to help with that.

      And, as others have said, transfer penalties are a huge turn-off for many potential riders.

      1. Exactly. I would have several hours per month more if one part of my commute were 15 minutes instead of 30 minutes. The other part is frequent but it’s prone to overcrowding and traffic. So only some of its runs match with the other route, which makes my entire trip subject to 30-minute intervals. I’d have several hours more per month if I could go when I was ready and not wait more than 15 minutes. It’s even worse when the first bus is so overcrowded I can’t get on it and have to wait for the second or third bus, by which time I might miss the transfer and have to wait another 25 minutes.

    4. I see your point but I think in general OBA simply make the system more usable in general but doesn’t negate the importance of frequency because ultimately OBA doesn’t shorten your overall travel time compared to driving.

      1. But it does, it two ways:

        1) It shortens the time I waste at my initial bus stop, if that bus is running late.

        2) It empowers me to choose the most optimal transfer strategy, where before I’d just be guessing.

        OBA technology came online when I was doing my Greenwood–>SU commute, and it definitely decreased my average door-to-door travel time.

  4. During a summer sojourn in Berlin some years ago, I commuted on a clockface route that ran every 20 minutes. Sure, 10 would have been preferable, but a very reliable 20 wasn’t bad at all.

  5. No, no, no. Frequency is not overrated. Frequency is *freedom*. Jarrett Walker could tell you that. During its operating hours, I should never, ever have to wait 18 minutes for my fixed-guideway transit spine service to show up. I shouldn’t have to spend money on a coffee/beer or a tablet/smartphone (not to mention the data plan) just to feel as though I’m not wasting the next 18 minutes of my life. Spend 30 minutes at the Federal Way Transit Center after dark and tell me that frequency is overrated.

    Anecdote time: I just moved to the University District so I could walk to work. I used to live near a stop on the 44. But my anxiety about bus commuting (unreliability, time consumed, etc.) became so dreadful that I had to make a change. My living + commuting situation is now better than I’d ever hoped for.

  6. The whole point of frequency — five minute head times — is that you can throw out your timetables. The natural variation in arrivals due to boarding times alone (or the accuracy of One Bus Away is less than the head time, so whatever the schedule says, the correct answer to the question “when should I leave for the stop?” is always “now”, because the answer to the question “when is the next bus/train/tram due?” is always “now”. Waiting is a waste of time, and a psychological transit killer. Especially since 18-minute head times, in real world practice, might mean 36 minutes, or more, if you miss one by five seconds.

    1. Minor, but it also saves on waiting for runners if people know that the next one is a manageably short time away. On holiday in Copenhagen, slightly disoriented, blood sugar exceedingly low, I needed to get up Nørrebrogade from Nørreport station, and AGH THE 5 was leaving, oh no oh no — until the station guide person kindly pointed out that the next one was a block away and I need not panic.

      Compare this to missing the 10:38-ish 545 to Seattle, where your sweaty hapless 10:38:30 arrival means you get to text /all/ your friends and read your kindle.

      1. The 41 gets down to a 4 minute headway. People still run for a 41 and the driver reopens the door for them. People work too hard to catch that bus.

      2. Occasionally, if OBA says another bus is coming really soon, I will let a crowded bus go by in order to catch an identical bus a couple minutes later with empty seats.

      3. I had a 158 driver close the door on me as I was running toward it, once. I was so startled I chased the bus for half a block before it sunk in what had happened.

  7. Interesting perspective! One thing the “real time information” response is missing is the experience of those who must transfer between routes to complete their trip. OBA is little comfort when I arrive downtown and have to wait 28 minutes for the next bus on my itinerary, which, incidentally, is exactly what happened to me when I was transfering to the D line Saturday afternoon. That leads to my second point: without reliability, the promise of frequency is nothing but a chimera.

    1. And we’ll never improve reliability as long as buses are sitting in auto traffic. Dedicated lanes, traffic signal priority, off board payment and more will all be needed.

    2. Interesting perspective! One thing the “real time information” response is missing is the experience of those who must transfer between routes to complete their trip.

      It depends! For several years I commuted from North Seattle (see previous post, my local 30 minute headway bus is the 28, between the D and the 5) to Seattle U. So I’d use one bus away to time my initial departure, described above. When I started doing it, I’d essentially guess, or try to use (largely useless) timetables, to figure out which stop to change buses at, and whether I should try for a 2 or 12 or a 3/4 (or just give up and walk up the hill). When I got a smartphone, that process became much more efficient; I could check the real time status of different possible transfers as the bus approached 3rd st, and choose my transfer point/route with more useful information.

      To your larger point, yes, that many trips rely on transfers is absolutely an important reason for greater frequency. I didn’t mean to suggest that frequency isn’t still really important, or that the guy quoted isn’t out to lunch about the value of frequency. But my own orientation towards frequency’s relative value has changed due to technology, and I’m definitely not alone. I know a lot of Ballard riders, and I don’t know anyone who found more value in increased frequency of the D outweighed the disutility of no real time info.

      1. Going from 20 minute to 15 minute headways only reduces your average wait time by 2.5 minutes. I’d be surprised if many people noticed an improvement.

      2. Worst-case wait time also makes a difference, because if you use a route often you’ll be subject to it sometimes. 15-minute frequency guarantees you won’t ever have to wait 19 minutes again. Of course, this assumes the route is actually punctual, which I hear doesn’t seem to be working out well with the D.

  8. What I take from Mr. Fry’s comment is that people in Portland aimlessly wander around, with nowhere important to go, so that waiting for an 18+ minute headway streetcar is not a big deal.

    I might add that Portland has very good beer and anybody that appreciates good beer knows that the wait time for an 18+ minute headway is an offensively short amount of time to finish a good beer.

    1. I might add that Portland has very good beer and anybody that appreciates good beer knows that the wait time for an 18+ minute headway is an offensively short amount of time to finish a good beer.

      Amen to that. Back on the old schedule, I’d sometimes stop in Ballard before catching the (30 minute headway) 75 home. If I just missed a 75, I’d go for a beer. But I didn’t want to rush it to get the next bus, so I’d finish with 20+ minutes, have another beer, and so on and so forth.

    2. And that Mr. Fry hasn’t considered that people who might try to use the streetcar to do business or anything that has a set start time.

  9. As usual, there are shades of gray here depending on the particular kind of transit one is talking about.

    For short-distance trips within the dense part of the city, I agree with Bruce’s argument, up to a point. But if speed is so poor that walking is not significantly slower, then resources need to be poured into speed over frequency, even if frequency is not optimal. For anyone who has lived in Boston, this is the B-Line Principle.

    For long-distance intercity trips, the picture is a bit different, because poor speed over such long distances can have a dramatic effect on travel time. I’ll use the example from my own commute — Lake City to downtown — to demonstrate. Before Sound Transit, this corridor was served by a half-hourly 307, which used the current 41 route between Lake City and downtown. The restructure for ST 522 canceled the 307, and split its hours into 15-minute service on the newly extended 41 and half-hourly service on the 522. In theory, you could gain a significant frequency improvement between Lake City and downtown by making the 522 like the old 307, and running service on that corridor every 10 minutes rather than 15 or 30. But I’d rather have the half-hourly 522, because it saves more than 10 minutes of travel time, more than making up for the increased expected wait.

    1. One compromise I wish would happen on the 522 would be to add 2 stops on weekends between Lake City and downtown. The first would be at Lake City Way and 15th Ave to serve the Roosevelt area. The second would be at the I-5/45th St. freeway station. These two stops would make it possible to travel between Kenmore/Bothell and many parts of north Seattle without the choice of either multiple transfers or a long backtrack into downtown. (On weekdays, the 372 addresses this, but the 372 does not run on weekends). At the same time, they would also provide much faster service between Roosevelt and downtown than what currently exists today.

      1. I’m totally on board with adding a Lake City Way/15th stop, not just weekends, but at all times.

        The issue with the I-5/45th stop is that it forces you to use the regular lanes of I-5 at all times. Currently, the buses use the express lanes whenever available, which are much faster as a general rule. Forcing the 522 into the regular lanes would eliminate some of its speed advantage over the hypothetical 307-redux during times when the express lanes are open.

        The real solution to this is to restructure, using hours currently used on the 68 and 72 to instead run the 372 on weekends. There is more need for the 372 on weekends than either the 68 or the 72.

      2. To clarify, I mean only the Roosevelt – Lake City portion of the 72. Obviously there is a shortage of service (especially on weekends) on the UW-downtown trunk.

      3. During peak commute hours in the peak direction, I totally agree that the 522 should bypass I-5/45th St. so it can use the express lanes. But off-peak, the express lanes make very little difference and often, the bus can’t use them anyway because they are pointed in the wrong direction.

        Ultimately, I agree that serving such a stop on weekends is really just a stopgap for lack of 372 service, but it’s at least something Sound Transit can do by themselves without depending on KC Metro.

        There is one issue with running the 372 on weekends, however, which is that the 372 would be mostly empty north of Lake City and the Woodinville tail would be overserved. So a weekend 372 would probably need to be a turnback route. Perhaps Saturday service alternating every other trip between turning back at Kenmore and Bothell, with all Sunday trips turning back at Kenmore.

      4. I strongly disagree that the regular lanes are an adequate substitute on weekends, as someone who uses that stretch of I-5 regularly by both bus and car. The express lanes are almost always free-flowing. The regular lanes are frequently slow leading to the Ship Canal Bridge in both directions, particularly northbound in the afternoon and evening when the Express Lanes are available. I suppose northbound the bus could do the 42nd Street jog like the 64 and 76 do on weekdays, but that frequently adds more than one light cycle to get through the 45th/7th light.

        If we put the 522 in the regular lanes at all times, we might as well put the 307 in place instead, which would allow for a single transfer to either the 66 or 72 to reach the UW area while remaining in the express lanes whenever possible and enabling 10-minute frequency (with the same slower speed) between Lake City and downtown.

      5. Forgot the last part of my comment: I would have the weekend 372 be Kenmore-only, timed for convenient transfers to the 522 (rather than the ineffectual effort to provide 15-minute service on a short corridor that we see on weekdays).

      6. The suggestion is basically that the 522 do what the 510 and 511 do — that what’s good for the goose is good for the gander.

        The real solution is all-day bi-directional transit lanes on every inch of I-5 and facilities for direct service to major cross-streets from them.

      7. How long has the 372 not run on weekends? I lived in Lake City in the late 90’s, and that would have been a major inconvenience, had it been the case at the time.

      8. To my knowledge, the 372 has never run on weekends. It used to be a peak-only route; then it went to an extended peak schedule with some midday trips; then it went to an all-day schedule on weekdays.

        I expect you’re thinking of the 72, which has run (infrequently) on weekends since the beginning of time and continues to run (infrequently) on weekends.

      9. Huh. Memory may be failing me, then. But the 72 used to go up to 145th, right? Because I was definitely able to get a bus all the to my apartment, which was next to the 145th/Lake City intersection.

      10. Does the 372 really need to go past Lake City? Especially if it becomes full-time. I assume the 522 is too full peak hours to absorb the 372, but at other times it should be able to.

      11. Yes, the 72 did go to 145th. It had a layover in the bushes at 145th and 39th. You can still see faded Metro paint on the curb at that location.

      12. I thought it still did go to 145th. :) I used to have friends who lived right between where the 72 and 73 terminated.

  10. Portland doesn’t have enough streetcars available to operate all the scheduled trips either. Portland ordered 5 new streetcars for the eastside line and so far, only one has been delivered and it is full of problems. The lack of operational streetcars necessitates cancelling some of the eastside runs, meaning that several times a day the gap between cars can be up to 36 minutes on the eastside. Service on the central spine between PSU and the Pearl District where both trains operate and most of the riders concentrate is usually about 10 minutes.

    But Portland has learned a few lessons that were applied to the eastside line. Stop spacing is better and the stations aren’t placed right at the intersections, which makes running times faster. The original line is dreadfully slow, but when it was free it was always packed. Now that a fare is required ($1.00) the cars aren’t as full.

  11. wait much more than about ten minutes for your transit vehicle to show up, taking transit has probably lost out to driving no matter how fast the vehicle goes, and can now only compete on the convenience and savings of not having to park

    Throw in a transfer, even with 15 minute frequency and you’re all but guaranteed ten minutes wait time. Without unlimited funding you can’t have universal 10 minute frequency or a complete system of one seat rides. What would help is Metro posting accurate schedules instead of fantasy schedules. After a routes been in existence for one shake up all the data is there to know when the bus actually arrives. If the bus starts out on time the first run of the day but is 20 minutes behind by the end of the peak publish the new schedule to reflect that.

    1. Adding 5 or even 10 minutes of schedule padding along the route, at key transfer points would help make those schedules reliable too. Make schedules reliable and then even 30 minute service with a transfer becomes somewhat reasonable. That said, I prefer frequency where I can get it (at a reasonable cost)

      1. Artificially enforced slowness is never reasonable. I can’t begin to tell you the number transfers I’ve missed because my bus was waiting with thumbs a-twiddling at a time point, only to hit a bridge/wheelchair/bike-loader/cashpayer delay between that timepoint and my transfer point.

        When your second-leg transfer has shit frequency, a bad experience is guaranteed to result much of the time. Slowing down routes by design cannot solve that, but it can do plenty of additional harm!

      2. Artificially enforced slowness is never reasonable.

        Agreed. Between predictable but artifically slow service, and less predictable service that will get me where I’m going as fast as is possible given current traffic, I’ll choose the latter. Artificial slowness is almost as bad as sitting at a bus stop, even when you don’t miss a transfer. When you do, it’s rage inducing.

        We need to accept that service won’t always conform to the schedule, because traffic in Seattle is too great and insufficiently predictable. That can be ameliorated by smart routing choices, signal priority, and other good policies, but unless there are some major changes to the region, it can’t be eliminated.

      3. Buses shouldn’t leave timepoints early. That too causes missed connections. And if a bus can’t reliably get to its timepoint on schedule, the schedule needs to change.

      4. Schedules should be set for a bus in average traffic, with a moderate passenger load, when driven by a competent and assertive driver. They should not be degraded by an expectation of perpetual delay. It should NEVER take 31 minutes to get from Ballard Ave to Pine, but a few times of a day a bus that does so will be “early”.

        Metro’s increasing inclination to set schedules to the Slowest Common Denominator is a real problem.

      5. I agree that buses shouldn’t leave the terminal at the beginning of the route early, but you can’t have buses stop every mile or two and wait several minutes just because the schedule says it’s a time point.

        Buses should never stop in the middle of route and wait except at major transfer points.

      6. Artificially enforced slowness is never reasonable.

        I just can’t agree with this. The only thing more demoralizing than sitting and waiting for a bus that never comes is to arrive at the bus stop a few minutes early and discover that the bus has already come and gone. It’s infuriating. It makes me think of the bus as an unpleasant last resort, something that might work sometimes, but not something I can rely on to get around the city.

      7. Re: buses leaving early. That’s why frequency matters. Run service every 10 minutes or less and missing the bus isn’t a big deal*.

        * assuming those buses are reliably on schedule and don’t bunch up

      8. Exactly, Oran. Yet another reason frequency matters. And better frequency actually helps to even out demand spikes, all other things being equal.

        That bus shouldn’t be early, Mars, because the schedule shouldn’t be so padded as to assume perpetual slowness.

        The whole “please arrive at the bus stop a few minutes early” advice Metro gives is insane, as it continues to presume you’re never leaving anywhere but work/home/a bar, that you’re never at an event that ends at an unpredictable time, that you never have an unknown distance to walk to reach the bus, and that you’re never relying on a transfer you can’t control.

        Seeing that advice in Metro literature is offensive, frankly.

      9. you’re never relying on a transfer you can’t control.

        That’s the big one. If there’s a transfer involved you can’t even decide what route to take. If you could set your watch by transit it would go a long way toward battling people’s zealotry for one seat rides without any other change. Conversely, it doesn’t matter how much you speed up the buses or try to increase frequency it won’t change people’s conviction that the only service of value to them is the one seat ride.

      10. I thought the drivers get reprimanded for early stops but not as much for late stops, because passengers are more irate if the bus comes early than if it comes late.

      11. Mike, you’re correct.

        You are likely to get written up for arriving at a timepoint more than 120 seconds early, repeatedly arriving at a timepoint more than 0 seconds early, or leaving a timepoint more than 0 seconds early.

        You won’t ever get written up for leaving a timepoint late unless you’re deliberately delaying service (usually by leaving the terminal late when you were there in plenty of time to set up the bus, use the restroom, and leave on time).

        The problem with scheduling as d.p. advocates, which would likely translate to a schedule that is within a few minutes of reality about 10%-20% of the time, is that there is no hope of understanding whether any given transfer is realistic. If we had super-high frequency on every line or even on most lines, that might not matter — but today, it does. People will get very confused if they find they are usually or regularly missing a transfer which they have, say, 9 minutes to make on paper.

        Scheduling should reflect an analysis of actual travel times which excludes outlier trips, but allows a normal driver to make the schedule under traffic conditions that are standard for the area and time of day. If few passengers are expected to board past a certain point, timepoints should be estimated after that point.

        It’s simply impossible to hire only drivers who have the unique talent — damn near impossible to evaluate before they get a few years of experience — for moving quickly and safely through traffic, and all it will do is compromise safety if you try to pressure all the drivers to move faster. If schedules are too slow or too unpredictable, the solution is to put speed and reliability improvements in place, not to hope that by putting drivers under more stress you will magically get faster results.

    2. You don’t need “unlimited” funding, you just need enough funding for a good level of transit frequency, as many cities have.

    1. Bruce does not delete topical comments.

      You may be confusing him with a different, more censorious author.

  12. I agree. Conceiving of the “wait” as an experience, a beneficial one, has been put into practice at many airports now where they have turned the terminals into shopping malls. Most people go into shopping malls and spend hours browsing, eating, walking around…what better time than when waiting an hour or two for a plane. Likewise, it comes down to the experience we build at our stops and stations. For example, the tunnel, for all its abstract beauty is pitiful even compared to a decrepit NYC substation which at least has kiosks for getting candy and coffee…a knish. Kent station is a better model, but on the station itself, there is no espresso stand. The reluctance to keep transit a kind of cathedral of pure transport has benefits (such as lack of trash) but at the same time maybe we need to make it more of a bazaar…with the types of shops and experiences that make a mall entertaining.

  13. If you miss a once-per-20-minutes bus, and as a result, you miss an hourly transfer, then you lose over an hour of time. It gets worse if you have multiple transfers, or if miss your transfer’s span of service.

    But it’s easy to say you only lose 20 minutes if you don’t ride the bus.

    1. Restrooms at transit centers would help a lot. There is nothing worse than being stuck with a 20 minute wait between buses with a bursting bladder and no place to relieve oneself. Coffee shops that are only open Monday-Friday during the peak commuter periods are not sufficient. We need restrooms available the full span of transit operation, especially during the evening and weekend periods when the waits are likely to be significantly longer.

      1. We need people to not trash and vandalize the restrooms when they do exist, then there would be more of them.

      2. What Mike said.

        There are lots of T-key (driver-accessible) restrooms at TCs, P&Rs, and various other points along Metro routes. They aren’t open to the public because the maintenance hassles and expense would be prohibitive.

      3. “We need people to not trash and vandalize the restrooms when they do exist, then there would be more of them.”

        That’s what happens when you give away things for less than their market value; people see no value in them and treat them with little respect.

      4. Normally, people think of restrooms as something that they have to be able to just walk in and use without any kind of accountability, which leads to tragedy-of-the-commons problems where people have no incentive to take care of it.

        Zipcar and Car2Go deter people from trashing their cars by holding each user personally accountable for any damage they cause. Restrooms could potentially do the same thing. Yes, I realize it would seem a little unusual to have to pre-register with a major credit card and have each visit you make to the bathroom be logged somewhere. But if that’s what it takes to get a clean bathroom to use while I change buses, I would say it’s well worth it. You could even throw in a charge of, say, $1 per use, to pay for routine cleaning, and I would still say it’s well worth it.

  14. Well, Seattle has managed to create buses that are both slow AND always late. So my wife now enjoys her new car and her 20 minute door to door drive from Ballard to her office. She’s the one who wanted to be green and take the bus so imagine my pleasure watching her give it all up for the nice new Santa Fe we bought. [Ad hom, ot]

    1. and causes one to become obsessed about transit

      This is true. When transit is grossly sub-par, and means to improve it within our financial means repeatedly botched or stymied, obsessive thought on the subject can become an addiction. Long-term effects include chronic frustration, acute malaise, mood swings, Head-To-Wall Contusion Syndrome, Repetitive Smartphone Refresh Disorder, Twiddler’s Thumb, social rigor mortis, and total loss of hope.

      And yes, it is also true that infrequent, unreliable, laborious transit is unsexy. No matter how many beers you drink while waiting.

  15. For years I’ve recommended a means to increase ETB (electric trolleybus) frequency that effectively matches supply to demand in circulator routes downtown with the least number of ETBs.

    Arrange ETB routes that remain downtown east/west between Broadway-12th and 1st Ave, and a separate north/south ETB circulator route between Jackson and Queen Anne on 1st & 3rd Aves. Include Lake Union in this compound circulator arrangement. Many routes allocate one direction of trolleybus on “paired streets” to keep overhead wire clutter to a minimum, gain service coverage and reduce the number of complex turns. This system can reach the waterfront near Coleman Dock (very important) and Sculpchewer Park. Outside this downtown circulator system, ETB routes run at less frequent intervals where transit demand is lower. For this I’m branded a troll by Seattle transit wonks and wannabees.

    1. This trolleybus reconfiguration aims for 5-minute frequency in both directions on 1st & 3rd Aves between Mercer and Jackson; 5-10 minute frequency between Capital Hill & Pike Place & through Lake Union/Queen Anne & Sculptchewer Pahk; 5-10 minute frequency between First Hill and either Western or Railway Ave near Coleman Dock. Because these lines are short, they require the least number of trolleybuses. Oh well. Seattlers know better than to listen to a Portlander who has advocated for light rail and streetcar there and around the country since 1992. Portland is just so much better than Seattle, it makes Seattlers jealous.

  16. I think there should be two kinds of bus service in this city. One: there should be one-seat commuter service to our major employment centers–downtown, possibly the U District–that costs MORE than your typical bus route because it’s a premium service (and besides, many employer-provided bus passes already cover more than the cost of a standard trip on Metro, so the pain to the commuter will be not that bad). These routes can be on a 30-minute headway because (a) no transfers will be needed because these are one-seat-rides, and (b) people tend to arrive at work at the same time every day, so they can pick the bus that arrives when they need it to for their work schedule and plan accordingly. These routes should not stop at every stop all along the route from their starting point to their destination, because the point is not to provide comprehensive transit service along the route, the point is to move people from their neighborhood to their place of employment. Such a route gets people out of their cars because it’s almost as fast as driving and avoids all the costs/hassles that come with driving (i.e., finding and paying for parking).

    Two: we should have a frequent all-day system for getting around the city in general, rather than for commuting. It should serve the needs of people who don’t want to/can’t afford to/otherwise don’t use a car, but still want to have a life. These routes can provide comprehensive coverage along a route and avoid the problem of people having to wait 20 minutes and buy a beer. This is what Rapid Ride should be but isn’t. This is what our streetcar lines should be but aren’t. This is what a subway is in other cities.

    I feel like, our problem as a city is that we want to do both with the same bus. We want to provide commuter service, but then we say “well, we’re going past that freeway station at 145th on our way from downtown to Shoreline, I guess we might as well stop on the way.” Or, we try to get people to commute on a bus that stops every 3 blocks for 5 miles and takes 3 times as long as driving, and wonder why they choose their car instead.

    Also, as a woman, the idea that a 20-minute wait for a bus is OK shows a complete disregard for the safety issues involved. I catch the bus on Aurora near several extremely sketchy motels. It’s scary enough being out there waiting for a bus for 2 minutes with crackwhores and their very mean and scary pimps. It sure as heck isn’t a place I want to hang out for 20 minutes. And frankly, there is no coffee shop, let alone a bar, available to wait in at 5:30 in the morning where I’m catching my bus. Not all of us are catching the bus in front of a cozy coffee shop on Queen Anne or in Ballard.

    1. This is what Rapid Ride should be but isn’t. This is what our streetcar lines should be but aren’t. This is what a subway is in other cities.


      And since those systems in other cities are of high enough quality to work even for commuting, those who still prefer a one-seat express bus with a guaranteed place to park their tochas recognize that they are choosing a premium service and pay a premium fare for it.

    2. All we’d need to do is set the farebox-recovery ratio for peak-express routes different than for other routes. I wonder whether $4 would translate into a 50% or 75% recovery.

      1. The peak express routes already have a higher fare recovery. If it’s financial parity you’re after it’s the local routes with near empty buses that would have to charge a premium. Of course you’d never be able to approach even the system average fare recovery by raising rates because it’s more expensive to operate than taxis.

      2. Question, which do people think would generate more push back; charging for parking at P&R lots or eliminating any transfers benefit associated with Express routes?

      3. The peak express routes already have a higher fare recovery.

        Nope. Untrue. Every one-way peak express route with a two-way core-service analogue experiences a farebox recovery rate that is 10%-20% lower than the two-way core route over the same period of time.

        But it’s not even about farebox-recovery parity. It’s about those who demand a premium service, at the expense of all other possible uses for and users of the transit system, actually paying a premium for that premium service!

      4. The in city Express routes have lower rides per platform hour. Depending on how transfers are allotted that probably means they have lower fare recovery. But if they are added to handle peak demand, making them all slow would be even worse. Of course a higher peak fare for all buses helps push demand out into the shoulders of the peak. When you get farther out the trend reverses. The 358EX tromps the 316 in the latest | 2010 :-( | performance report. The 193EX beats the 190 and 192 locals from Star Lake although is not apples to apples since the Express goes to First Hill instead of the Seattle CBD (although you’d expect demand to the CBD to be higher). To dredge up more examples you have to start looking at ST vs. local Metro service which again isn’t apples to apples. But ST “rides per service hour” is a tick over 30 which exceeds the vast majority of South and Eastside routes that serve the Seattle core.

      5. Of course a higher peak fare for all buses helps push demand out into the shoulders of the peak.

        Except not really, because the peak period is already so loosely defined as to ding you for most shoulder trips just the same. There’s nothing more offensive than riding the same old crappy infrequent bus at 3:15 pm, or a counter-commute bus so overwhelmed that you won’t get downtown until well after 6:00, and paying the same premium fare as the White Collar Overservice Shuttles. (I’ve been on plenty of express buses with seats available that left downtown at 5:01 PM.)

        It’s a premium service. Charge more for it.

      6. I would be curious to see what the ridership was like on the 316 before it began its routing past Green Lake. When I lived in Northgate in 2000-2002, it got on the freeway at Northgate and went straight downtown from there. Now that it meanders instead of being a true commuter route, I bet a lot more people drive to the P&R and catch the 41. Because, it’s FASTER. I mean, why get on a slow meandering bus? That’s not a commuter bus, just because someone put X on it and it only runs at peak. A commuter bus goes from a neighborhood, to an employment center. It doesn’t go from one neighborhood to the next to the next to the next and then for the last 2 miles gets on the freeway. That is not a commuter bus. That’s a local bus. Design commuter routes for commuters, design local routes for local service, stop making one route try to do both. Because, it will do both badly.

        Also, we have a model already for commuter routes to downtown costing more than local routes–it’s called Snohomish County. They pay $4, or $5.25 if they’re north of Everett.

        And you know what? It’s faster to take the 402 from Lynnwood Transit Center to downtown than it is to ride the 316 from 145th & Meridian to downtown. Let me say that again: your commute is faster living by Lynnwood Transit Center (29 minutes at 6AM) than it is living at the north edge of Seattle (36 minutes at 6AM). That says to me that we are doing it wrong. And if we are going to make it a premium service and charge more for it, we have to make it better than the local routes.

        I don’t care about financial parity. I’m saying, from a consumer standpoint, I would pay more money for better service. My employer sponsored bus pass already covers a higher fare than I am currently using. So, give me better commuter service, AND charge me more for it, I am totally fine with that.

      7. I would agree with all of that.

        I don’t object to commuter expresses existing. I object to the fraudulent suggestion that they are all transit can or should do in Seattle, and that all of our service presumptions should be built around them, at the literal expense of everything and everyone else.

        We are, of course, doubling down on your cited Lynnwood mistake. Soon enough, Lynnwood will get a train running constantly and providing faster access to central Seattle than Ballard or even the Central District have. If current fare relationships remain constant, this will cost Lynnwood-ites barely 25 cents more than someone crawling on a 2 bus during the arbitrary “peak” period, and less than a poor sap who makes the mistake of crossing the fare boundary on the 358.

      8. People would definitely scream more about paid parking.

        I expect you are right. Certainly there would be a few that would scream the loudest ;-) While I feel charging to park is the right thing to do if it’s politically possible to eliminate transfers on commuter express runs maybe that’s one way to charge a premium. The result would likely be more people buying monthly passes which might also help Orca adoption.

      9. I would happily pay a couple bucks for parking if it meant more reliable service, preferably service that ran later into the day. My schedule means the most reasonable bus for me to catch is the 7:59 am 197, but it’s the last one of the day, so if I arrive a little late I’m SOL and might as well drive to work. (The 158 to downtown, with a transfer to the 71/72/73/74 is also an option, but it’s pretty erratic as to whether it arrives before or after the 197, I find.)

      10. “Lynnwood will get a train running constantly and providing faster access to central Seattle than Ballard or even the Central District have. If current fare relationships remain constant, this will cost Lynnwood-ites barely 25 cents more than someone crawling on a 2 bus during the arbitrary “peak” period, and less than a poor sap who makes the mistake of crossing the fare boundary on the 358.”

        The solution is to improve Ballard and Central District service, not take it out on Lynnwood. One thing Link will do is dramatically improve the connections between north Seattle and Snohomish County. It’s ridiculous that from downtown there are express buses to Snohomish County but if you’re at Northgate or 85th your only choice is the 358+Swift, 16+346+Swift, or 347 (at which point you can transfer to the 511 at a lonely freeway station or Mountlake Terrace TC). I used to work in one of those office buildings on Meridian north of NSCC, and at least two colleagues commuted from Snohomish County. I couldn’t very well tell them to take transit to Northgate because it would have taken an hour or two on local buses and there was no express alternative.

      11. On paying for parking: a lot of downtown workers have bus passes paid for by their employers. (Not all, I know I’ve been generalizing here, but a lot of them do.) Unless their bus pass covered the parking fees at a P&R, I think people would complain. And start parking on the street up the road from the P&R instead of at the P&R itself, which makes the neighbors angry. Whereas, if you raise the fares, the employer just pays for a more expensive bus pass, so the commuter doesn’t alter their behavior or complain about the higher fare.

  17. To add some actual research to this discussion:

    I find the emphasis on frequency to be a bit of red herring. Research backs this up. The paper above surveyed riders for what was important to them about their transit experience. What came after “safety” ?

    “My bus/train was on time.”

    The popularity of real-time tracking also serves to echo this sentiment.

    Once frequency is high enough (and I’d suspect 30 minutes would be fine, humans being what we are), able to handle the capacity, then what really matters is that you can arrive at your stop with a minute to spare, and reliably arrive at your destination in time for your next engagement.

    Consider: Would you want 3 trains an hour, but 20% of the time they are 15 minutes late, or would you want one every 30 minutes, but they are virtually never late.

    Dedicated right of ways, light priority, efficient boarding — these things drive on-schedule performance and hence should be valued highly.

    1. Not necessarily. Another thing that really matters is that, when I have an appointment right at 5:45, I can get there after work without having to leave work early, get there twenty minutes early, and kill time because the next bus after that would’ve gotten me there ten minutes late.

    2. Pretty much any core Los Angeles service corridor is running minimum 10-15 minutes in the daytime. Some run as often as every 3-5 minutes at peak. Many if not most corridors run 15 minutes into the very late evening. While 30-minute services exist, that is a such a bare-minimum level of to be barely worthy of mention, much something you should “suspect is fine.”

      It is doubtful that survey respondents at any of the stops chosen for the above study even conceived of 30-minute service as an option to be balanced with other priorities.

      What the “my bus/train was on time” response tells us is that the expectation of reliability and ease of use is paramount. Frequency helps to deliver these things; infrequency does not.

      The survey responses also teach us that large plazas and pretty landscaping and other accouterments that tend to distract transit architects are decidedly of little interest to the transit user, who just wants to be able to get where she’s going!

      1. …much less something you should “suspect is fine”…

        Anyway, there as here, a commuter traveling 30 miles on a train from Orange County can afford to conform to a less-frequent schedule, as it still saves them sitting in terrible highway traffic over a very long trip.

        Someone traveling 7 miles from Hollywood to Culver City should not and will not wait 30 minutes for any leg of the journey (never mind both), because that’s stupid. You can do it in 15 in a car, even with traffic.

    3. I find pitting service frequency against on-time performance to be a red herring. No one here is saying or suggesting that dedicated ROW, priority, and efficient boarding are not valued highly (see the case of RapidRide as the counterpoint to your assertion).

      Both influence the decision to take transit or not. Like I said in another comment, frequency isn’t very useful when the buses are bunching up, increasing effective headway. But also, all other things equal, a service that runs every 30 minutes is less attractive than one than runs every 15 minutes.

      To cite some research I found in the course of writing my masters project: statistical analysis in this paper finds that the most significant variable in determining passenger wait time is service headway.

      Also, a survey of Berlin transit riders after a major restructuring of the network into high-frequency corridors shows that “a majority of respondents placed the highest priority on short intervals between vehicles. … The lines with the highest levels of customer satisfaction are those that run at intervals so short that customers do not even need to look at the timetable.”

      On-time performance isn’t even a factor if you look at the schedule of a potential transit service and decide: “the times are not convenient for me” or “I have to leave much earlier than driving” or “I can’t get there and back in a reasonable time” because the frequency and span are inadequate. But once you make the decision to take transit and the bus is late, then you start reconsidering your choice.

      1. But isn’t it just expected that transit in Germany runs on a schedule you can set your watch by?

    4. The thing that most irks passengers is waiting at bus stops. Real-time displays help, and punctuality helps too, but the biggest thing that turns off riders is “I have to wait 20 minutes for my transfer” or “I have to time my trip to match the bus’s schedule.” San Francisco, Chicago, and Vancouver have universal high frequency in areas equivalent to Greenwood-to-Rainier Beach. That means every 5-20 minutes until 10pm every day, 30 minutes thereafter, and half-hourly night owls to all parts of the city. (Although Vancouver’s night owls have some gaps after 3am.) North of 85th may be too spotty for blanket full-time frequent service, but at least Aurora, Northgate, and Lake City should have it. (Aurora’s is coming with RR E, Northgate with Link, Lake City is still waiting.)

      Metro and the county have always danced around universal frequency because “75% of a bus’s cost is for the driver”, so it has gone with things like articulated buses instead. But that doesn’t address passengers’ primary concern, which is frequency. So just do it. Not “now” when Metro has no money of course, but as soon as possible.

  18. I’m afraid the data do not support your assertion concerning increased frequency being inherently superior to increased on-time performance.

    The study I quoted said as such, and it’s not alone… another study (included below) referenced by the firs say that making things more on time is of approximately equal value to the same change in the time-between-trains.

    And further, it is clearly the case that while frequency often results in the average wait til for SOME train to arrive reducing, it is clearly not the ONLY way. (right of ways, etc)

    ** You need to make an argument that the cost of more-frequent-trains is less than making-existing-trains-come-on-time. (or show me studies which indicate that my assertions about the value of on-time to riders is wrong.)

    An interesting new factor is real-time-tracking… this may influence the outcome either way, as (on one hand) it means you’ll always know when the next train will be, even when late, and on average you’l wait less. But on the other hand, there’s a lot of value in knowing when to start walking to the station and having the train always be there… having trains closer to on-time means your schedule is less disturbed by problems. [Aside: I got involved in this conversation because I learned long ago of a Swiss study that showed their riders preferred “more on time” to “more trains”. I couldn’t find that study, so I poked around to find another one.]

    All these things are measurable. We don’t need to assert. My “suspect is fine” is due the fact Switzerland structured their trains with 30 minute headways as a standard. (I do not know if this still is the case).

    (As for your Hollywood->CulverCity example: why would someone wait 30 minutes when they (in my proposed scenario) KNOW the trains will leave at (say) :08 and :38. I think the number of people making truly spontaneous “gosh I think I’ll go to Culver City RIGHT NOW” decisions is small relative to people who are either making regular, planned trips or who have a few minutes between decision and needed departure. (Also measurable.)

    Most people I know who travel often by transit know which train/bus they are going to take, and arrive at the stop appropriately so as to minimize their wasted time.

    FYI: I’m seriously pro-transit. I’m also an engineer. Measure things. Insist on it. Reason about the data. Act accordingly.

    Travel time uncertainty is likely perceived as a significant burden by most travelers. Atkins and Polak (1997) show that the relative weight values of mean and
    one-standard deviation of wait times are 2.6 and 2.5, respectively, which suggests
    that reducing arrival time uncertainty (or increase in waiting time reliability) has
    about the same effect on generalized costs of transit trip as a corresponding reduction in headways.

    1. Re: Switzerland, you must be referring to their INTERCITY passenger rail network which is scheduled as a pulse system. [map] [paper].

      A train every 30 minutes is pretty frequent for intercity service but woefully inadequate for urban transit, especially when the competition (driving, biking, or walking) does not restrict you to a schedule and allows much more flexibility in departure time. Returning to the case of the Portland Streetcar, it’s so slow and infrequent you might as well walk or bike.

      1. Please confine your opinions about Portland as conjecture. Your experience in Seattle is detrimental to the formulation of an objective viewpoint. What is it about Portland’s successful streetcar system that causes Seattlers to regard success with suspicion? Could it be that the poor performing Lake Union Streetcar is such an embarrassment, Seattlers have developed an inferiority complex?

      2. Wells, it’s not my opinion. It’s a fact.

        Urban myth: You can walk faster than the Portland Streetcar to your destination.

        Finding: Confirmed. It’s no myth.

        Results may vary depending on the time of day, congestion, gait, distance and weather.

        I didn’t even mention the SLU streetcar, which suffers from the same flaw.

      3. And Wells (and anyone who doesn’t read Oran’s link in detail),

        The author beats the streetcar to his destination despite having waited out its awful headway unnecessarily before beginning his walk.

        Never has there been a more damning rebuttal to the “Portland Rides Well” delusion.

      4. Portland’s streetcars are much more patronized than the Lake Union line. In other words, streetcar speed is sufficient to serve the transit need if not the unreasonable wants of purist transit ideologues and the blantantly anti-transit sabateurs at Metro, Sound Transit, SDOT and Wsdot.

      5. Wow. The Oregonian article Oran and dp quote, only an anti-transit demogogue would consider honest. The article is about a race; a 1.7 mile direct walk versus a 2.7 mile roundabout streetcar trip with 18 stops. This is fact? This proves walking is faster? Baloney. Portland’s new streetcar Central Loop travels quite a bit faster on the Eastside than Westside. Fact: Even on the Westside, the streetcar is faster than walking.

      6. The article is pretty clear about that, quote:

        “The route you traveled on foot is less than half as long as the route the streetcar took,” said PBOT spokesman Dan Anderson.

        Really? That’s all they’ve got? Of course a pedestrian is going to take the most direct route rather than following the streetcar tracks through town. Oh, and again, I was competing against a machine designed to move people.

        Ultimately, however, it’s clear that the streetcar’s frequency is a greater hindrance than its speed.

        “With frequency, transit is always competing with a personal vehicle, even if it’s just walking,” Walker said.

        Substitute cyclist or driver for pedestrian and that statement is still true. With the frequency getting worse, those other options become more attractive.

      7. “With the frequency getting worse (less frequent), the other (travel) options become more attractive.”

        Whatever. Arranging and implementing streetcar lines requires a more comprehensive understanding than any debate about their relative speed can offer. Oran, you should’ve recognized the article was more political spectacle than informative.

        My support for the Eastside Streetcar line included a how it could act as a traffic calming devise. I’ve noticed that when the streetcar is absent, cars bunch together at higher speeds as they always have. But when the streetcar is present, even barrelling along at 30mph there, traffic is more dispersed and slower. Comprehensive. Try being that sometime.

      8. Wells, that’s ridiculous.

        A newspaper article isn’t inherently anti-transit because it happens to wound your civic pride by pointing out that rail in Portland is not inherently awesome and used by all.

        The unfortunate fact is that Portland has built four MAX lines and two zig-zagging streetcar braches, and yet transit commute mode-share in the city proper hovers between 10% and 15%, and much less for non-commute trips. At the rate things are going, biking will beat it out in a few years, which makes a lot of sense because you don’t have to wait up to 18 minutes to get on your bike.

        As Oran rightly points out, your criticism of the walker for not following the streetcar’s circuitous loop is absurd. The streetcar is scheduled to average a mere 6 mph downtown. That’s 2x walking speed. Perhaps it runs closer to 3x walking speed on the Eastside loop. But waiting for any length of time for it firmly negates that speed. Going far out of your way once on it negates the rest. As the Oregonian‘s test demonstrated.

        Slow + infrequent + far out of the way is the trifecta of poor transit service, and the Eastside streetcar does them all!

        the Eastside Streetcar line …as a traffic calming devise.

        You can’t be serious! There are dozens of ways of calming traffic, all cheaper than and likely more effective than a slowpoking train (especially one that only calms three times an hour). Transit is about moving people. What about that do you not get?

        You keep citing the equally ridiculous SLUT to prove we’re “jealous”. We’re not jealous. Frankly, we’re pissed that you sold our politicians a ridiculous bill of goods (with the TriMet fare chart still on it, stickered over after delivery) that doesn’t actually do anything useful, and we’re pissed that our politicians swallowed your silly sales pitch.

        TL;DR: The Portland Streetcar is stupid, and no, it is not successful. The only reason it ever “looks full” is that it’s a really small train that barely ever comes!

    2. Barry,

      Given your claim to be someone with a personal and professional imperative to “reason about the data”, I find it disturbing that you seem to have such a poor acquaintance with the on-the-ground facts of transit geometry and travel behavior, and such a disinterest in learning about the topic at hand before boasting of your superior conclusions.

      Your repeated references to “the trains that leave at [clockface]” further suggest, as Oran observed, an inability to distinguish intercity and outer-suburban railroad scheduling from the more complex systems required to make an urbanized area run smoothly. Frankly, your language suggests a preference for Railroad Tycoon/Sim City-level closed-system modeling over anything that could be instructive in the real world.

      You insist on repeatedly citing a Los Angeles-based study to suggest that frequency is unimportant, though even a cursory reading of the study reveals that you couldn’t misinterpret it more egregiously if you tried. Again, the service baseline for every line serving the studied stops is 2x-6x better than you would have us believe. Core Los Angeles services are already frequent; their weakness is that they run along very lengthy and sometimes traffic-choked corridors, and can therefore be prone to uneven running times and resultant bunching.

      In this context, “on-time-ness”, as reported and prioritized by survey respondents, pointedly does not translate to “follows a clock-face on a printed schedule”. Printed schedules are rarely used except at night, and most of the time one is standing miles from the nearest time-point, so even a printed schedule would offer only an estimate.

      No, in the context of the study you cite, “on-time-ness” means nothing more than “I can show up at the stop and not wait especially long due to bunching”. This response is a demand for the experience of frequency and not a request to reduce frequency in exchange for the hassle of living by a printed schedule.

      Did you even pay attention to the title or the purpose of the study, as described in the abstract? This paper showed that the ability to get around quickly, easily, and safely matters more to people than window-dressing amenities designed to distract them from service flaws. Your attempt to posit frequency and timeliness against one another exists nowhere in the study – again, unsurprisingly, because the service baseline is already reasonably frequent.

      I think the number of people making truly spontaneous “gosh I think I’ll go to Culver City RIGHT NOW” decisions is small relative to people who are either making regular, planned trips or who have a few minutes between decision and needed departure

      You think wrong. Los Angeles is nothing if not a city in which people make many, many unplanned and ever-time-sensitive chained trips, for business and for pleasure, to and fro across the urban area. As they are socially accustomed to doing so in cars, the preferred departure time is always NOW.

      I chose my example carefully. Though 7 miles apart, Hollywood and Culver City are not only both well within the central urbanized zone, but are within the same “swath” of density known as the Westside. In L.A. terms, these are practically next-door neighbors. They also share the same primary industry (entertainment), and spontaneous trips between the two are commonplace.

      Hollywood and Culver City also have no single connecting street, so any trip between the two requires forming an “L” shape on the street grid. On transit, that means taking two buses or a bus and a train, with none of the possible trip permutations allowing a transfer at the kind of major destination that would be used as a suburban-style “pulse point”. So timed transfers are a no-go. Your anti-frequency crusade would therefore guarantee a long transfer penalty, in addition to significantly delaying the start of the trip.

      There is no better example than Los Angeles of a city finally getting its transit act together after decades of wandering in the wilderness. Your dangerous and brazen misreading of a single line item in a single survey would kick that entire transit system back to the dark ages if acted upon.

      Frequency matters for urban movement. It is only over long distances that make spontaneous trips inherently difficult – long suburban commutes, intercity travel, transcontinental flights – that frequency diminishes in importance. If you don’t get that, you have no business claiming to be in command of “reason”.

      1. Thanks for the ad hominem attacks. They were helpful.

        You point out many reasonable flaws in my argument; it was hastily constructed.

        On the other hand:
        * you make a number of unsubstantiated claims, eg “the preferred departure time is always NOW.”
        * you misunderstand my argument concerning frequency — I’m only suggesting that frequency could be reduced IF one can ensure running on-time.

        The central point of my argument was that 1) there exist data which suggest on-time arrivals are of comparable value to increased frequency *in some situations* and 2) you need to measure all this stuff, not rely on intuition. The discussion above felt very soft and intuition based.

        Finally, I’m not sure from your “sim city” comment if you simply don’t believe in models, or if you have some deeper knowledge about complex systems. Perhaps you would care to share it?

        If you want to be more effective in debate, I would suggest:
        1) support your claims with data and peer reviewed research, (we can all agree the issues are subtle)
        2) read your material more closely (something I was certainly guilty of failing at), and
        3) maybe layoff the ad-hominem, you zealot. (oops.) You could have just has easily written a helpful educational reply.

      2. I’m sorry if you felt I was being a jerk. It was only after you came in with guns a’blazing, yelling “stop the presses!” and insisting that “the data do not support your assertion [that frequency matters]”.

        No one disputes that reliable, timely, evenly-spaced transit marks a tangible improvement in user experience and expectations of mobility. Where bunching is so bad that two bunched vehicles cut scheduled frequency effectively in half, then fixing that problem (i.e. restoring advertised frequency) would indeed be as good or better than double the number of vehicles but leaving them bunched.

        But in an urban context, none of this matters unless the advertised and delivered frequency is good enough to be useful spontaneously. Insufficient frequency will never cut it, no matter how clockwork-like the infrequent system runs.

        “The preferred departure time is always NOW” is not an unsubstantiated claim. The more urban the context, the shorter the distance, the less time required to complete a minimum-length round trip at that distance… the less schedule-dependent time-blocking will be needed to embark upon that trip. As a result, the tolerable wait time to leave will approach zero.

        This is especially true in a historically car-oriented city, where the wait time for shorter trips has always been precisely zero.

        There are plenty of valid models for understanding human behavior and its cumulative effect on a system as complex as a major city. Unfortunately, you argued that people were living their lives in precise half-hour blocks as they waited for the properly regimented moment to depart. That’s what I was describing as a model of Sim City simplicity.

    3. 30-minute scheduled frequency with on-time buses serves a certain subset of the population; i.e., those who are currently riding the 28 or evening 5, plus a few more who would ride it if the buses were more on-time. That’s maybe 5% of the people who travel in/out of those neighborhoods at those times. That’s a stopgap, not a long-term goal. The 28 may never be frequent but the 5 should be full-time.

      If a route is frequent and unreliable, what that does is to undermine some of the advantages of the frequency. I doubt an unreliable 15-minute route is as unpopular as a reliable 30-minute route, but it’s approaching it. The solution is not to switch to all 30-minute routes to make them reliable, but to keep trying to make the 15-minute routes more punctual and to keep making more routes 15-minutes. Because if you have to wait 20 minutes for an unreliable route (like the 44 very often!), it’s still better than a guaranteed 20 minute or 30 minute wait. Because on average you’ll only have to wait 5 or 10 minutes sometimes.

    4. I really wish people who don’t think frequency or full span of service are important would think about how they’d feel if the freeways were closed except at 15-minute intervals, or if they were closed between midnight and 6am. That’s exactly what it’s like when buses are infrequent or have a short span of service.

  19. The notion that real time information replaces frequency presupposes that one can control exactly to the minute when one is coming and going. That isn’t necessarily true for me. There will be something I need to take care of in the morning before I leave for work. Someone will want to talk to me before I head home. My daughter–imagine!–will need something from me. I like having the real time information, but having the frequency is more important.

    BART is actually not a good example of 15 minute frequency. Weekday daytimes 28 out of 43 stations have two or more lines, or at least 8 trips an hour to some destinations (not really 7.5 minutes because the headways aren’t always even). The stations with only one line are well out into the suburbs, and most of them have supplemental peak service (on the Pittsburg-Bay Point line). VTA’s light rail in San Jose has lots of stations with only 30 minute service, compounded by a slow and circuitous routing, leading to poor ridership.

Comments are closed.