One of my most-used tools in life is the bicycle directions on Google Maps. They’re great. I’ve had some flaws over the years – like the time I was directed to take the “Barbary Coast Trail”, which is actually three flights of steps that were neither easy to walk up with my 40 pound mixte nor the best route – but for the most part, it’s a really great tool. I don’t use it as often as I used to; if it’s in the city, I probably know the most bicycle-friendly way to get there by now. I do use it a lot to compare routes, though, like the time I was working at an office that’s almost at the exact opposite corner of the city from my apartment. There are at least five totally viable ways to get there; the cycling directions helped me choose.
The San Francisco Bike Coalition (of which I am a paid member and could not be more supportive) has always had a bike direction tool, but the 2.0 version of it was recently released. I’m all in favor of competition, and I’m so glad the SFBC is working to make cycling easier for us in this city in all sorts of different ways, but there are some glaring flaws.
One: The SFBC tool has one big thing going for it: it gives you how many calories you’re burning. The problem? That doesn’t actually make any sense. Calorie estimates – particularly ones that take no data into account other than distance – are simply a bad idea. A 150 pound fit female will not burn the same calories as a 250 pound obese female, or a 200 pound male with a high body fat percentage, or a 200 pound male with a low body fat percentage. The differences can be completely staggering.
For example, as per WebMD, which is also a flawed tool since it’s only taking your weight into account, if you’re cycling at an average of 12-14 mph (which is standard enough in the city) for 38 minutes (which is the SFBC number I’ll be referencing in point 2), a 120 pound person would burn 276 calories, compared to the 461 someone would burn if they were 200 pounds. SFBC estimates 282.9; I’d be interested to know who their “average” person is in this instance.
Two: The times are completely, totally inaccurate. I live at 26th & Guerrero. I’m currently training for the MS150; my quick training ride is an 18 mile loop that goes from my place up Valencia, up 17th to Sanchez, through the wiggle, through the Panhandle, through GGP and down the Great Highway to the zoo, where I turn around and book it back. I figured it would be a pretty good route to use to test the tools against each other, since I’ve got a ton of data on it and ride it enough times per week that I might be cocky enough to use the word “expert”.
I ride this route regularly. I am a competent enough cyclist. I’m not a racer, but I am training to be a stronger distance cyclist for a long ride in a couple of months. I’ve been riding for years, ride a very good bicycle, and am slightly faster on average than most cyclists in the city. (This statement is based only on the number of other cyclists I pass every day when I’m riding at a speed that’s comfortable than me. It is in no way scientific and certainly still doesn’t mean that I am “fast”. At all.)
My fastest recent time on this route was 1:16:34, just over 76 minutes. My average speed was 14.1, my max speed was 25.9. These numbers come from my bicycle computer, which is only logging active time, not time from point A to point B (and in the case of my loop, back to point A again). And while this is not the place to debate such things – you’ve got thousands of better places on the internet to do that – that’s involving an awful lot of stop sign running and some highly questionable speed in certain areas. I had to ride as hard as my not-entirely-in-shape body possibly could in order to achieve that time. And yet, the SFBC tool says it takes 38 minutes to ride that route. They don’t mention if that takes into account time for stopping at lights, dealing with traffic, etc., but I’m going to go ahead and wager a guess that 38 minutes is a pipe dream for the sort of riders that would be using the SFBC tool. That’s over a 14 mph average WITHOUT stopping. That is substantial. Google Maps lists 52 minutes, as seen in the comparing screenshots below. (Click to expand.)
I have to say, things like “how long will it take me to get there” are significantly more important to me than a bad calorie estimation or how many carbon emissions I’ve offset – a point on which I don’t really have the authority to speak, but I have to imagine that’s a slightly difficult one to discuss too. SUV versus Smart Car? Is there consideration given to the amount of food I’m going to have to eat to fuel myself and where it might have come from, or the carbon used to produce my bicycle? I can speculate all day.
3. It requires a very, very, very specific format for your addresses. When I type in 3649 26th St., an address that is well within the city limits and in fact could only mean one thing, it returns “Your route could not be found. Please check your address location and try again. Address should be in the format: Street Address or Intersection, City.” And forgive me, but this is a San Francisco-specific tool created by the San Francisco Bicycle Coalition. I assumed you would know what city I was referring to. It’s probably on me for being a creature of habit, but I screw this up every single time. I’ve been using Google Maps for years, which manages to figure out what I mean, so I’m used to not having to format in a weirdly specific way. Perhaps the Google model has spoiled us with the “did you mean…” correction mentality, but it’s frustrating for a web tool in 2011 to just reject my information and not offer alternative solutions.
Cycling directions in a city like San Francisco are so, so important, particularly to the sort of new riders that the SFBC should be targeting. It’s a hilly city with some good bicycle infrastructure, but you might not know the optimal way to go if you’re new to cycling or new to the city. My route to the Great Highway takes at least twenty turns. I have them memorized now, but I can imagine it might be difficult to get out there in a non-hilly, cycling-friendly way if you don’t have those little shortcuts committed to memory.
The shortest vs. flattest route is an interesting option, but Google Maps just gives me three different options, has a terrain map, and lets me figure it out. I’m not sure which one is the better system – I’ve used Google for so long that those things aren’t taxing to me. They also aren’t disruptive to the process – if a route can be 2.6 miles, 3.4 miles or 4.0 miles, I know which one is the shortest. I didn’t need it to tell me. And by turning on the terrain map, I can determine what the least hilly way to go is, though theoretically bicycling directions aren’t going to have too much variation as far as hills are concerned.
I’m glad the SFBC is thinking about things like this, but honestly, they’re trying to reinvent the wheel with little success. Routes aren’t able to be modified (as in clicking on the route to drag it to a new area, which Google lists as traveling “via (insert thing here)”, an excellent feature), the estimations of time/calories/carbon are pretty useless and largely inaccurate, and if you don’t format both of your addresses properly, the tool simply does not work. I’ve got my criticisms of the Google cycling directions too, but I’m not convinced the SFBC spent its money correctly here. The completely infuriating thing is that it USES Google Maps, but somehow manages to accomplish everything in a significantly worse way. I’d love to see some features that justify all of this development. Hopefully 3.0 will be more successful.