The first thing I did with it was to enter my commute and see what it came up with (screenshot of the partial route below).
It doesn’t quite match with the route I prefer – for instance, it sends me along Lawrence then follows a Manhattan-like pattern parallel to Foothill Expressway rather than taking Foothill itself. I would never take Lawrence, even though it has a pretty wide shoulder: having traffic zip by you at +50mph is highly annoying, and there are too many intersections and lots of right-turning cars. Bollinger + de Anza is a highly preferable alternative, there you have less and (somewhat) slower traffic. I do like Foothill on the other hand, even though you have the same problems as Lawrence (traffic zipping by at high speed) but this is offset by a very wide shoulder giving you an increased sense of safety. And the stops and intersections are fewer and a bit safer – the Manhattan pattern alternative would slow you down quite a lot.
I guess this simple example outlines the difficulty in coming up with a good algorithm: there is a huge space of parameters which rather than digital (suitable to bike or not?) have an entire grayscale range of values, which, to make matters worse, also have a subjective quality to it, making it very hard to come up with an optimal solution. But, at least using the new bike directions will give you a good first-order stab at a decent and safe route – so far in cases where I was on the bike in a non-familiar area I’ve tried the ‘walking’ directions, which not always led to desirable results.
Interesting to note is the collaboration with Rails-to-Trails, to identify trails – it seems for instance a number of trails in Rancho San Antonio park (left in the screenshot) are included in the database.
However, there are a couple of things in the LatLong post we feel the need to strongly dissent with.
l don’t know anyone who enjoys biking up a hill, especially when you’re trying to get somewhere you need to be. Going uphill is worse than simply being much slower; it’s also exhausting and can take a toll on the rest of your ride.
Mmh, there are plenty of cyclists – yours truly for instance – who like a good climb!
But it gets worse:
Many cyclists will tell you that going downhill is annoying for a different reason: you may have to ride your brakes all the way down.
I don’t know what kind of crack those guys are on. Who doesn’t like a screaming fast downhill? Just get off the brakes.
More seriously, this bike directions algorithm is a nice feature. Since long Open Street Map (to which we contribute on occasion) was pretty much the only global online map with decent bike routes – though with strong variability in coverage depending on the region – besides of course local maps such as the VTA map for Santa Clara county; good to have another effort now (and a bit of competition?).