Carter Rubin, on Los Angeles Metro's blog The Source, says exactly what needs to be said about WalkScore.com's Transit Score product. The product is in the news today because of their new ranking of US cities by Transit Score.
Everyone’s favorite mapping tool, Walk Score, has launched a new service called Transit Score and used it to determine what it believes are the best American cities for public transit. The Switchboard blog has a recap and analysis. Los Angeles clocks in at 11, just behind Portland, Ore., and ahead of Denver, Colo. — respectable enough company, I suppose. That said, I do have some constructive criticisms of Transit Score’s methodology.
To calculate a raw Transit Score, we sum the value of all of the nearby routes. The value of a route is defined as the service level (frequency per week) multiplied by the mode weight (heavy/light rail is weighted 2X, ferry/cable car/other are 1.5X, and bus is 1X) multiplied by a distance penalty. The distance penalty calculates the distance to the nearest stop on a route and then uses the same distance decay function as the Walk Score algorithm.
The first issue that jumps out at me is how Transit Score weights different modes to give a bonus to “better” modes: “heavy/light rail is weighted 2X, ferry/cable car/other are 1.5X, and bus is 1X.” There’s such an incredible diversity of service quality within each mode that this strikes me as an oversimplification. Is the Metro Los Angeles Orange Line really half as “good” as the Boston Green Line — a nice enough line, but it plods along 100-year-old tracks — just because the former runs on rubber tires?
There are arguments either way, and reasonable people can disagree. Those differences, however, are simply glazed over with a simple multiplier. And, I don’t even pretend to understand the 1.5X weighting for ferries and cable cars.
Early last year, when Transit Score first rolled out, I discussed it with WalkScore's Matt Lerner, expanded on this very critique, and suggested a better (though computationally intense) approach. It involves aggregating the content of their travel time maps – effectively "maps of your freedom" — over all likely destinations from any residence, so the two digit score is actually a percentage, a composite of answers to the question "What percentage of jobs, retail, etc can you get to in __ minutes, on transit, from here?" It needs refinement, but that's the only truly factual measure of access that could be reduced to a two digit number.
I understand why Walk Score folks settled on this methodology; it's more about the limits of compultational power than any modal bias on their part. Still, Transit Score is the only WalkScore.com product that I can't recommend. I love everything else that WalkScore does, but Transit Score assumes that a slow streetcar is better for you than an express bus, and people need to make that decision for themselves.