We have been working on rounding up shoe ratings for Common Running in an effort to get enough data on the shoes to try out our “intelligent” shoe recommendations and see if they actually work. However, I discovered while putting together the site data page that the site is already useful.
How? I went back to a listserv running community I belong to, where I’d asked for shoe ratings, to thank them for their help, and discovered one runner asking if there was a site which would let him check his current shoes and get a list of other shoes which were similar. Someone mentioned CR, and another list-member checked his shoes and reported back on the list of other models we reported as similar–including the other model he wears and likes.
We actually use a pretty lightweight algorithm to make those lists. It mentions the easiest possible models (the previous model in that line, and any descendants, so for the Asics GT-2120 we’d mention the GT-2110 and GT-2130), similar models by the same brand at different price points (e.g. for the GT-2120, we’d mention the GT-1130) and then models by other brands which are in that class (stability, motion control, etc.) and have a similar weight and price, like the Brooks Adrenaline GTS 7.
Each of these are easy SQL queries; the only reason the recommendations aren’t “perfect” is that our database isn’t comprehensive. And yet there’s no other site on the web that we can find which does this. There are sites where you can compare shoes by checking off a bunch and seeing them side-by-side, but this is simple and easy for the user. And it’s useful, even though we aren’t doing the kind of recommending we had planned yet.
