Detroit’s WalkScore is broken. Bikescore too?
The popular Walkscore web site gives estimates of a city’s walkability.
It’s not perfect. It’s really a measure of density rather than infrastructure. It also assumes roads have sidewalks (because Google Maps makes the same assumption.)
But there’s a bigger problem in the city of Detroit.
Detroit neighborhoods like Midtown receive high scores of 85 out of 100 — or “very walkable.”
Neighborhoods like Downtown receive an 18 or “car dependent.” For reference Auburn Hills gets a 36.
Looking at the walkability heatmap shows the Walkscore bug. Notice the red stair steps along the river? The Walkscore program is selecting restaurants, stores and other destinations in Canada. That might be okay for Swimscore.com but it doesn’t work for walking.
Unfortunately affects all the Detroit neighborhoods along the River and lowers the overall score for the City.
So breath with relief. Detroit is not tied with Garden City with a score of 50.
We’ve documented this bug as have others and shared it with WalkScore, but it hasn’t been resolved yet.
Bike Score
This week the walkscore people released a bike score. As far as we can tell, they didn’t publish Detroit’s number but our guess is it’s even more inaccurate than the city’s walkscore.
Besides this border bug, their methodology relies heavily on the American Census Bureau bike commuting numbers for Detroit, which we’ve already shown are too inaccurate to be useful.
They also assume that streets with bike infrastructure are more bikeable than streets without. That may make sense in most cities, but not in Detroit.
Which of these two streets are more bike friendly to your eyes?
Despite the motorized traffic, groups like Bikescore, the League of American Bicyclists, and the Alliance for Walking and Biking say the one of the left is because there’s a bike marking on the pavement. However, what cyclist wouldn’t choose the wide open Second Avenue instead (or the similar Third, Brush, John R, 12th, 14th, etc.)
Unfortunately the very low traffic volumes on Detroit roads aren’t measured in the scoring.
There could be a traffic factor based on vehicle ownership and road density. It would probably be a more accurate indicator than the census bikes counts, though that’s setting the bar pretty low.
Until such a factor is developed and applied in the scoring, Detroit will not be accurately recognized as the bike friendly city it truly is.
Related Posts
Tags: bike counts, bikescore, walkscore
May 15th, 2012 at 8:08 am
That bug also applies in reverse in the suburbs. It gives high scores to areas that are not very walkable simply because the concentration of business in an area. The area around Twelve Oaks Mall in Novi is a perfect example. It’s not very walkable but it has one of the highest scores in Novi. But other areas of Novi that have infrastructure like sidewalks and sidepaths score low because there’s not a lot of destinations that are within walking distance. I know that they disclose this in their scoring system but the average person isn’t going to understand that distinction.
May 15th, 2012 at 8:49 am
We have known this for years, thusly don’t use it for analysis like some people have asked us to do. great article
May 15th, 2012 at 10:24 am
@andrew: The media doesn’t understand the distinction either.
You just have to use Walkscore and understand the data and assumptions it’s making. Sometimes it’ll work for you and sometimes it won’t.
May 15th, 2012 at 12:43 pm
Since I got access to Google Mapmaker, I did add all of the sidewalks, sidepath, bike lanes, etc. within Novi in the Google street attributes. If they ever start using actual sidewalk data, it might better reflect what’s on the ground. But in reviewing that data in the area, most of that is missing save for a select few cities.
May 18th, 2012 at 1:22 am
[…] Except, as cycling blog m-bike.org observes, Walk Score is broken. […]
April 30th, 2013 at 1:12 pm
[…] algorithm is grabbing data from restaurants in Canada giving downtown a very low walkscore and lowering the overall score of Detroit. You can see the incorrect calculating along the […]
March 29th, 2015 at 6:06 pm
[…] plenty of grist for STS critique in Fitbit and Walk Score: their bugs, what they fail to capture, the privacy concerns they raise, how they have been adapted and […]
April 13th, 2015 at 11:17 pm
[…] However, Redfin and Model D are both wrong. This increase was largely due to a bug fix in the Walk Score algorithms that was well-documented in 2012 by m-bike.org. […]