Walkscore is all jacked up with bad addresses and other incorrect information. Makes it hard to get a reliable score. I'm in S. Arlington and at least one restaurant from North Arlington in incorrectly in my neighborhood. In addition, there is just massive amounts of stuff missing.

by Eric H. on Aug 16, 2010 3:49 pm • link • report

I always find it necessary to mention that for a while, Walk Score listed an IMAX theatre AT my house. They would be very disappointed with the real AV situation at this address.

by Nichole on Aug 16, 2010 4:01 pm • link • report

I live in Cleveland Park and my walk score is 77.

The house where I grew up, in a town of 3,000 in the middle of nowhere in Pennsylvania, has a 76.

Yeah ... I call shenanigans.

by Tim on Aug 16, 2010 4:49 pm • link • report

Tim, why do you suggest that small towns cant be walkable?

In fact, the traditional small town is the essence of walkability. Retail, post office, church...all in a two block area, many times with apartments on 2nd and 3rd floors. A ring of large homes surrounds the center, all within walking distance. If you live, work and play in a town of 3,000 you certainly arent getting a on a highway in your daily travels.

by JJJJJ on Aug 16, 2010 5:01 pm • link • report

@JJJJJ don't you think that Tim knows whether or not his small town is walkable or not?

The suburb where I grew up is not walkable at all (almost zero sidewalks anywhere) and it still rates a 68.

by Nichole on Aug 16, 2010 5:06 pm • link • report

@Tim

Agreed. The system is pretty good but should also rank the relative quality of nearby places (if possible). For example, having a 7-11 count as a "grocery store" is not exactly the same thing as living down the block from an actual market. In the same vein, classifying the Scottish Rite temple as a bookstore/library is also not exactly proper; I'm pretty sure I can't go buy and checkout books from their library (though I haven't tried).

by Adam L on Aug 16, 2010 5:25 pm • link • report

I just tried everywhere I've ever lived, spanning between DC, MD, PA, and NJ. I'd qualitatively say that 4 were spot-on (Logan Circle, Russett, State College, & Wyomissing), 1 was far too high (Greenbelt Park), and 3 were too low (East Falls, Lancaster County, and Long Beach Island outside of winter).

by Bossi on Aug 16, 2010 5:48 pm • link • report

Bad joke. I just checked my address in Arlington. Imaginary bookstore (the closed Olsoon's), missed the blue bus, lists my local bank as the "sumitoto bank" which probably doesn't have a ATM, my Safeway is in Annandale, etc.

It is a gimmick, and the funders are throwing good money after bad.

by charlie on Aug 16, 2010 5:53 pm • link • report

Here on the Hill, it tells me my closest Entertainment is the "Union Station Garage" (?!?), and my closest groceries are one of the Eastern Market vendors which they incorrectly place on 7th Street NE instead of SE. Great data quality, guys.

by Chris on Aug 16, 2010 8:44 pm • link • report

There's still something messed up with their algorithm.
I sampled some corners that are only a few blocks apart in Silver Spring:
52 at Sundale & East West Hwy where it's 0.83 miles from Metro and right at a bus stop.

1 block away at Colston Dr & Washington Ave is a 48 and called car-dependent

Another block away at Grubb Rd & Washington Ave is a 46 despite being right next to a nice shopping strip, school, and still close to bus.

On the other side of EW Hwy, Sundale Dr and Ross is a 54

Sundale Dr and Porter is a 69. It's equally as far from buses as Grubb and Washington. There is no shopping near it and the only thing actually closer to it's microscopically closer to an elementary school. (a day care center is called a shopping location). It also looks like it things you can walk along the future purple line to 16th st. (unless this is still as the crow flies)

I understand a bit of noise in the algorithm, but a 23 point jump over a few blocks is ridiculous.

by Dan on Aug 16, 2010 9:56 pm • link • report

Other than this software company serving as a platform for some software developers to make a name for themselves, does it serve any useful purpose?

http://www.frontseat.org/projects.html

And should I be disturbed that their funding is very opaque? Or I just being old fashioned?

by charlie on Aug 16, 2010 10:59 pm • link • report

Transit score lacks any reference to Montgomery County Ride-On buses.

by JayTee on Aug 16, 2010 11:28 pm • link • report

@ Dan & JayTee: The lack of Ride On data totally screws up the data in areas where only Ride On serves or is the closest option. In the case of Washington/Grubb/Colston in Silver Spring, the closest transit would be Ride On's 1 bus which goes around via Washington and Grubb; in TransitScore's eyes, the closest transit are the J Metrobuses on East-West Highway.

I lived in that area for 6 months without a car and I managed better than I did when I lived in Kensington (in a place with an 85 score) sans car for a month.

by Jason on Aug 17, 2010 8:52 am • link • report

Seems to work for me. Perhaps addresses involving directions (like the Norths and Souths of Arlington, and the NW, SE etc of DC) throw off the system.

I put in my address (two blocks from the corner of King and Washington in Old Town Alexandria), and the only thing that was "off" was the "Books" category. I think that there are so many non-profits that have their own small publishing entities or libraries that they show up too readily. That said, there's a public library 2 blocks from that intersection, and a bookstore 4 blocks away, so no biggie.

What I got for my address is a very accurate representation of what's available to me and how I live my life in town. I think the score (100) makes sense because there is nothing I can think of (save a video store and a GOOD Indian restaurant--don't like the one on King Street) that one would need on even a semi-regular basis that you can't find within a few blocks.

by Catherine on Aug 17, 2010 9:33 am • link • report

@Jason. Agreed. Ride-on picks up on my block. Closest WMATA bus is over a quarter-mile away.

by JayTee on Aug 17, 2010 9:38 am • link • report

@Jason

It's more than a RideOn issue for Washington Ave & Sundale. Both Washington & Grubb and Sundale & Porter have only RideOn buses going by them. In fact, Washington and Grubb is actually closer to the nearest MetroBus stop. The scores are 46 and 69. As best I can tell, the big difference is that the program thinks you can walk down the CSX tracks to Silver Spring (or is still calculating as the crow flies).

by Dan on Aug 17, 2010 9:41 am • link • report

The transit score was not very helpful in terms of providing public transit time for me (car/bike data was fine), but it explained:
"If you do not see transit data on Walk Score, then your transit agencies do not provide open data to software developers." Is there any chance of Montgomery County RideOn providing open data, I wonder.

by Tmichaels on Aug 17, 2010 10:10 am • link • report

Seems to work fine for me. It's a bit too much to be complaining that this company doesn't have omniscient awareness of every small business in the country.

The one curiosity that I noticed is that it will only display a Transit Score if the number is very, very good. For example, in Takoma, MD, or the streetcar suburbs of Pittsburgh, it just listed all of the trains rather than giving me a score. So it's not a lack of data so much as cautiousness on the part of the model.

by tom veil on Aug 17, 2010 10:26 am • link • report

I think it's a bit jacked. I live in College Park, MD two blocks from the metro and it gave me a walk score of only 57 (somewhat walkable). Mean while if I put in an address three blocks over from my house (four bloack from the metro, the score jumps to 89 (very walkable)hugh.....?? Additionally when I asked it to calculate my commute to Crystal City, it advised that I get on the Marc (which also stops at College Pwrk metro) take the Marc to Union Station, then get off and take a Loudon County Dulles to Crystal City. This would take me 1 hour and twenty minutes versus the 40 mins a direct metro ride would take. I really don't think they have the mass transit tool right (at least for the suburban rail transit.

by WDCJoe on Aug 17, 2010 12:14 pm • link • report

@Charlie, their funding isn't actually very opaque at all, though it may not list specifics on that project list. The Transit Score development was funded by the Rockefeller Foundation, as the press release and post about Transit Score states pretty clearly.

Seems like this discussion happens every time someone mentions Walk Score, but they're not generating the business listings. Those are directly from Google and their API and directory. It's hard to notice, but the listings have gotten infinitely better in the last 5 years as more people use google's service and crowdsource the incorrect data or business listings. If you want to help make those better, you can help by submitting corrections through Google for businesses that are closed or incorrectly filed.

This tool isn't perfect, far from it. But it gives a pretty good baseline number of a neighborhood's walkability. And, as Stephen mentions, they're in the process of updating the algorithm to better reflect a few other conditions that affect walkability, like frequency of intersections, block length and link/node ratio that will definitely help out the accuracy of scoring: http://blog.walkscore.com/2010/08/street-smart-walk-score/

I'm reminded of Louis CK's bit about airplanes where we're just a little meh about truly amazing things. You can type an address into the internet and get an approximation of the walkability. Cool! And yet, it seems people just want to point out the flaws. To me anyway, the far more notable thing this does is help raise the profile among the non-planning urban geeks out there of the implications that location can have on your ability to walk, to be healthier, and to save money by having multiple ways to get around. It's done TONS to help raise awareness of the demand for walkable places.

Oh, and for the availability of data, check in with http://www.citygoround.org/ for more on how to pressure your local agency to release data.

by Steve on Aug 17, 2010 1:02 pm • link • report

I suggest also checking out http://walkshed.org/, which started out with Philly but has since expanded to include NYC. I really like how you can set your own priorities for each of the factors. This tool isn't perfect, either, but I really like the direction that this one is going.

@WDCJoe-

Keep in mind that proximity to transit isn't the only factor -- it's also grocery stores and other destinations. Granted, as per my most above I still think Walkscore tends to be hit or miss... again, I think both Walkscore & Walkshed are on the right track, but they're both quite young.

by Bossi on Aug 17, 2010 1:15 pm • link • report

@WDCJoe:
Walkscore and Transitscore are built on the Google platform. Since WMATA has not joined Google Transit, Walkscore and Transitscore don't know it exists (at least for scheduling purposes).

by Matt Johnson on Aug 17, 2010 1:29 pm • link • report

@TMichaels: Ride On did give their data to Google Transit but hasn't done much of anything else in terms of releasing it to developers. Not that there would be a need to do so in theory, it'd just be nice to do given their circumstances of being a fragmented system.

Releasing that data and doing NextBus like they said they'd do a year ago are two of the few downsides about Ride On not having to do with sealed windows and not turning the AC on until it's about 80.

by Jason on Aug 17, 2010 1:35 pm • link • report

@Matt Johnson
I understand this (the fact that wmata is not part of google), however the algorythm still seems to be off. I cannot understand how an address three blocks from me, easily accessible by sidewalks can have a walk score 32 points higher, especially when one considers my proximity to metro. I think they are pretty dead on in the assessment of moost district proper locations.

@Bossi
I understand the algorhythm they utilize is based off the proximity of amenities, however a three block change on quiet streets lined by sidewalks is a bit of stretch at 32 points. Additionally, considering that the new alogorythem supposedly takes transit into account, one would think my closer proximity to metro would bring me on par with the other higher ranked location.

by WDCJoe on Aug 17, 2010 1:50 pm • link • report

@WDCJoe:
Because Walkscore can't calculate the frequency and quality of transit at College Park Metro (because WMATA is not on Google Transit), the majority of your walkscore is coming from amenities.

If you're on Calvert at Bowdoin, you basically on top of the Metro. And the post office is there too.

If you're at Rhode Island Ave and Calvert, you're 2 blocks closer to restaurants, shops, parks, a University, and so on.

That's why your walkscore is better the closer you get to Baltimore Ave (Route 1). If WMATA was on GoogleTransit, you would probably see your Transitscore increase with proximity to the Metro.

by Matt Johnson on Aug 17, 2010 1:55 pm • link • report

Understood, I just dont like that ranking. My issue really isn't a walk score one, but a wmata issue. I wish metro would get on board.

by WDCJoe on Aug 17, 2010 4:04 pm • link • report

@Matt, though WMATA's data isn't totally open yet, they do have basic information incorporated into there — that's how it's calculating the Transit Scores for addresses (in the District at least.) Search for a DC address and the Transit Score comes up at the bottom with all nearby bus lines and rail lines listed. Looks like they're pulling the information the same way that routes and stops show up on google maps even now. Routes and stops are part of Google, though timetables and detailed arrival information are not yet.

Also, to make things totally clear, these are two separate algorithms. Having transit nearby does not increase your walk score by making things a few stops away "walkable," etc. Two separate calculations.

by Steve on Aug 18, 2010 12:14 am • link • report

The idea behind this site is unique and very interesting.

However, I don't find the results to be very accurate. The algorithm gives weird results - 1.) two adjacent addressees will yield totally different results and 2.) MANY locations are just missing and don't blame that fact on Google as all the missing locations in Walk Score show up on Google Map.

For these reasons, I thing the service is useless.

by dan mckinney on Feb 1, 2011 11:32 am • link • report