Thursday, October 13, 2011

What Makes Up Walkability Anyway?

Yesterday, commenting on Strong Towns, I commented on how jogging paths and crosswalks don't make up "walkability" and the Walk Score showed that, giving the school a walk score of a cool 3. I contrasted the locally-highly-amenitized Kensington High School for the Creative and Performing Arts, on the border between Fishtown and Norris Square, with a Walk Score of 83, to demonstrate what a walkable school would actually look like.

And then I Walk Scored a commercial strip from my childhood. Here's what it looks like. It's pure sprawl, and an effort to traverse in anything that doesn't have four wheels and a motor. Yet it gets a Walk Score of 72. Why?

It occurs to me there are two aspects to "walkability" (or transit mobility in general): you have to have locational amenities--that is, somewhere to walk to, and traversal amenities--that is, what kind of effort it takes to get there. Locational amenities are places, and places such as stores, restaurants, parks, schools, churches, friends' houses/apartments/condos, while traversal amenities are the condition and quality of the trip. For pedestrians, this means wide sidewalks, good (and frequent and signalized) crosswalks, fairly narrow streets, and lots of interest and variety in one's surroundings, including variety in the built environment and of traversal paths; issues that decrease traversal amenity include poor sidewalks (if not no sidewalks), overly wide roads, poor crosswalks (if any at all), poorly set stoplights, overly deep setbacks (especially for commercial), highway boundaries (such as Bethlehem Pike in the example above or DeKalb Pike by King of Prussia; many suburban strips are, unfortunately, arrayed around these highway boundaries), lack of interest along path (few buildings, buildings too deeply set back, lack of usage distinctions, oversized parking, etc.), and so on.

Now, needless to say, locational amenities need to be weighted over traversal amenities. Locational amenity trumps traversal amenity when the area of traversal, from door to door, is relatively short (500 ft., say) and not blocked by some sort of boundary feature (fence, wall, waterway, cliff, highway, thicket, etc.) Desire lines in the suburbs routinely get etched through narrow hedgerows and woods, for example, for precisely this reason. However, there is also a horizon or sunset, beyond which traversal effort in the main trumps any kind of locational amenities whatsoever. Let's peg pedestrian sunset at 2000 ft. As the traversal effort increases, the need for locational amenities likewise increase, as a way to counteract the effort being expended and keep the pedestrian sunset from shrinking (to 1500 ft. or even 1000 ft., say). This shrink space indicates an increased importance of traversal amenity, meaning it now trumps location as the primary determinant in whether one will walk; beyond the sunset, this is in turn trumped by distance. The presence (or absence) of traversal amenity can be determined empirically, critically scaled, and then weighted against the scales of location (and location density) and distance.

The problem with Walk Score is frankly that it only measures and weights locational amenities, and then weights them by locational density as an address function. It in no way measures traversal amenity, or even traversal, for that matter; Walk Score perceives Airport Square and Montgomery Mall as being on opposite sides of the road and hence highly locationally dense; it does not register that to actually walk from the exit of Airport Square's Toys "R" Us to Montgomery Mall's entrance one has to traverse two goliath parking lots 500 feet each with a serious boundary feature (Bethlehem Pike) without any sort of safe crossing halfway between. It, in other words, fails to register that the built environment, with no pedestrian traversal amenities whatsoever perpendicular to the lay of the buildings, has effectively created a pedestrian sunset precisely as deep as the setbacks. Without pedestrian amenities, distance trumps location as soon as location ceases to be an independently viable draw.

Because of this, Walk Score is only a useful means of measuring pedestrian amenity when pedestrian amenity is there to begin with. We can see the contrast when we compare Montgomeryville's walk score with that of the closest Main Street, that in Lansdale. The implication is clear: either (a) Walk Score has to do a better job of measuring walkability viz. traversal amenities, or (b) a new walkability score is needed.


  1. Take a look at it seems to be a pilot attempt (for Philly and NYC) to do exactly what you are proposing.

  2. I WalkScored a place in Palo Alto that I stayed, and got a 68. That place was completely unwalkable, typical South Bay sprawl. I plugged in an address 100 ft away, and got a 72. WalkScore is strange.

  3. I really like the locational v. traversal idea, and the critique of Walkscore is dead on. Internationally, it seems to have even less usefulness. My study-abroad apartment in medieval Trastevere in Rome, one block from a full-service grocery store and two blocks from a light rail line, in the most pedestrian-friendly surroundings imaginable with dozens of cafes, restaurants, butcher shops, greengrocers, bookstores, etc., scored an 83, one point higher than the address for an apartment I used to live in along a 1920s-era auto arterial, where the sum total of walkable options (w/in 10 minutes) was a McDonalds and a Mexican restaurant, and buses had 30 minute headways. By Walkscore's methodology, at least within the USA, it seems anything under "90" is primarily auto-oriented.

  4. The basic critique of Walkscore as it stands today is it seems to assume (a) that all buildings reasonably front close to the street and (b) all streets are crossable. Since this is empirically false--as I have demonstrated--Walkscore must be taken with a grain of salt whenever the context shows a lack of an urban street network, and discounted totally in pedestrian-prohibitive contexts (such as the post-1980 strip). In other words, its assumptions--made most assuredly for the sake of simplifying its algorithm--render it an at best limited tool.

  5. Another problem with Walkscore is that it doesn't consider the quality of amenities. For example, under groceries it will consider anything from a bodega with limited hours to a full-service 24/7 supermarket. The nearest grocery store to my apartment I've been to maybe once, because of its limited selection and hours; for serious grocery shopping, I need to walk 15 minutes to the supermarket.