13 Replies to “Podcast #76: Unplug the Database”

  1. Martin’s right about the Via math – it really is ~$60 per rider. That begs another question, doesn’t it? How is it possible they are so expensive with 1,000 riders a week? Can’t we just kick some money to Uber more cost-effectively to get people to stations?

    1. According to the Seattle Times article, “Metro’s performance standards aim for at least 1,000 Via customers a week, which computes to about $6 operating cost per user.” $6/rider is a whole lot better than $60.

      1. There’s an error in the Seattle Times math. $3.25 million per year is $62,000 per week. With 1,000 riders, that’s $62 per rider.

        Anyway, it’s weird. Seems it ought to be closer to $6 than $62 if it’s running efficiently. These are short trips and often shared.

      2. Well, you do understand that the contractor’s IPO has to be “viable”, don’t you?

        They are Amurrican Freeeee Enterprise and should be reeewarded instead of those Union slackers.

      3. There is one plausible scenario where the cost per rider starts to look more like $60 than $6. That’s if Metro misjudges the number of vans they need and ends up paying a lot of drivers to sit around all day in their cars, waiting for passengers that never show up.

        My gut instinct is that the above is not likely, though. The level of demand ought to be predictable enough for Metro to figure out quickly how many drivers they need.

  2. The missing element of Seattle’s transit system seem to be reliable last-mile shuttles. I’ve always felt that LA did it right with the LADOT DASH program: https://www.ladottransit.com/dash/

    Shorter routes have tons of advantages in an environment where higher speed transit in exclusive lanes (Link, RapidRide to some extent) provide an attractive alternative to slow “milk run” service design that extend for miles and miles. I would even observe that Routes 50, 60 and 106 are an unfortunate extension of that milk run approach.

    Rather than fund the Via program combined with long and expensive routes that go through congested bottlenecks in SODO or Downtown (and create on-time reliability issues), a few simple shuttle routes would do wonders! Each SE Seattle route needs to serve a Link station (avoiding multiple street crossings), a business district on Rainier for several blocks and areas where regular off-wire buses (because of narrow streets or steeper slopes) cannot handle well.

    It would require new neighborhood consensus to do this, but I think it’s possible. Many SE Seattle neighborhoods function as internal communities, and making local trips by transit Is as important as getting to a Downtown job.

    The final point is that short-distance and slow-moving shuttles are the ones most suitable to someday be driverless. Many technology companies are honing products to do just that. Once driverless, these routes could run often and with extended hours.

    1. Technologically, I would argue that the easiest transit vehicles to automate should actually be the Link trains. The computer wouldn’t have to control steering, just acceleration and braking, most of the route is already grade-separated, plus it helps immensely to be able to hard-code into the computer exactly what route the vehicle is going to take and how fast to go on each sections. If necessary, a driverless Link could even have humans in a control room operate the trains remotely when they pass through the Rainier Valley, switching to full autonomy only on the elevated and underground sections where the Link trains have the right-of-way completely to themselves. This approach would still require some human labor, but much less than having a driver riding each and every train all the way from one end to the other.

      Going back to the point about short-haul shuttles…it’s difficult. Basically, I see the types of shuttles you are envisioning as completing more with walking than with driving. In many cases, it is just as fast, if not faster to simply walk than to wait for the shuttle.

      I think route 50, between Othello Station and Columbia City Station is about as good as one can expect for a shuttle route, and does go to some places far enough from the station for a shuttle route to make some sense. The problem is that the 50 wastes so much time crawling through SODO and the VA hospital parking lot, which not only waters down the frequency for the part where the 50’s coverage is unique, but also ruins the reliability.

      1. Automation is also a function of speed. I could see trains having automation soon — but low-speed shuttles can more effectivrly react to avoid collisions. A shuttle no faster than 20 mph is going to be pretty safe.

    2. Yes indeedy, just like the amazingly successful Positive Train Control system that American railroads have had now for decades…….


      1. This is a little bit more complicated than positive train control, since you need to look at cameras and make sure that there’s no nut crossing the tracks against a red light. As far as autonomous driving goes, this is a relatively easy problem to figure out, especially in a semi-controlled environment where you can sensors and cameras stationary on the street, in addition to those mounted on board the train.

        Whether it’s possible to retrofit a train with such technology that was built in an era before it existed is another story, as are the implications on politics and relationships with the union (e.g. if Sound Transit tried to automate Link trains, would that cause the bus drivers and mechanics to go on strike?).

        Politically speaking, the shuttle scenario may very well prove easier to automate, simply because they’re operated by private contractors who aren’t beholden to the unions, even though, technologically, automating Link trains should theoretically be easier than vehicles doing general-purpose driving on public streets.

  3. Rainier Valley is different from the U-District in two ways. Rainie Valley is a continuous urban village that’s five miles long, and people take buses up and down it all day to shop and go tp places in other parts of the valley because it’s too far to walk. If you split the bus routes in the middle, then people have to transfer even for a short straight trip, and if the buses are only every 15 minutes then that’s a long transfer that makes transit not very effective and so people get back in their cars.

    Second, the U-District is separated from downtown by a ship canal and three miles, the first of which is low-density houses. It’s like a separate island. the mixed-use part of the valley ends less than a mile from Jackson Street and there are still businesses in between people go to. So the 7 makes more sense to serve the continuous corridor.

    1. Certainly Route 7 is a wonderful route. It should connect to Link at Rainier Beach — but overall it is good.

      Rainier Valley is however not just a few blocks either side of Rainier Ave. Also, every current Link station is at least 1/4 mile from Rainier Ave except Mt Baker. Going east- west on a bus is challenging, with areas as far as a mile east of Rainier Ave and even further from Link. Beacon Hill (1/2 to 1.5 miles to the west of Link) is also mostly residential so those residents also rely on Rainier Ave. This is why Via or other east-west shuttle strategies are needed. Route 7 and Link are just not good enough transit.

    2. Looking at the long range plan it seems like route #3996, which connects Mt Baker and Rainier Beach via Seward Park / Wilson Ave S, would be a good fixed-route version of the Via shuttle. Wouldn’t solve all use cases but I think would solve a lot of the complaints from the east side of the valley who don’t currently have a good way to get to Link.

Comments are closed.