Please don't litter OpenStreetMap with false data to manipulate routing.
this post was submitted on 28 Sep 2024
6 points (100.0% liked)
Bloomington Indiana
3 readers
2 users here now
Bloomington Indiana Lemmy! Come stay for the fun, or snark at students, or yell at clouds!
founded 1 year ago
MODERATORS
@markstos @bloomington_in on the other side, bike routing likely gets better too
@DemonHusky @markstos @bloomington_in Bike routes will get more direct, but "better" may not be applicable, since the more direct routes also encourage drivers to speed
That said cities tend to add bike lanes only after people get annoyed with bike traffic, so maybe down the road it might add some lane miles
@pleaseclap @DemonHusky @bloomington_in I can’t see how any routes will become more direct for anyone, because no new more direct paths are being added.
Tagging a slow speed limit is a signal that cars may want to take a less direct route on streets with faster speed limits, while tagging a high speed limit is a signal for bike routing that cyclists may be less safe and may themselves prefer a less direct but safer route.
@markstos @pleaseclap @bloomington_in without knowing the local area, I was thinking not that there might be basically parallel routes with different speed limits, and car routing would choose the higher limit option, while bike routing chooses the lower limit option. But with the assumption that the limits reflect something else about the road condition and how comfortable biking is, the routing algorithm now has a better signal to choose the better route
@markstos @DemonHusky @bloomington_in
What DemonHusky said:
A map app suggesting a route for a bike will try to minimize time via the slowest streets. For a car, it will try to minimize time via the fastest streets (which are usually more direct in terms of distance)
So if the algorithm suggesting a route thinks all streets are the same speed, bikes and cars will be suggested the same low-distance routes, which will usually be routes featuring excessive speeding by drivers
@pleaseclap @DemonHusky @bloomington_in In the absence of a speed limit signal, the algorithms look for other signals, like the type of road. Cars are routed to arterials over residential streets even they are a bit longer, while bike routing would prefer residential streets.
The Pointz bike routing app is nice because it puts you in control with more options.
@markstos @DemonHusky @bloomington_in I haven't seen a routing algorithm's code, but a preference for arterial roads can be explained solely by speed limits
@pleaseclap @DemonHusky @bloomington_in I’ve looked at the code. My servers have bike routing coverage for most of the US and EU with the OpenTripPlanner and Valhalla engines. Occasionally we find edge cases that don’t route optimally and look into patches.
Lots of roads don’t have max speeds in OpenStreetMap which they both use, so other signals are essential.
@markstos @DemonHusky @bloomington_in
You've seen your specialty bike routing algorithms. I'm glad they're good algorithms, however
your first post is about misrepresenting speed limits in OpenStreetMap (which is a community resource, yes?) to effect driver behavior, and what systemic effects this would have for every user of every tool that uses OpenStreetMap
Even if you don't think this will lead to congestion on the same roads, I'm not sure it's ethical
@pleaseclap @DemonHusky @bloomington_in Misrepresentating? I found streets where the max_speed value was missing and added the correct values based on city code.
@markstos @DemonHusky @bloomington_in
Is that not your suggestion, here?
"It seems possible that adding lower-than-default speed limits to OpenStreetMap could reduce the likelihood that routing algorithms would route car traffic there, which could in turn keep the street safer for other road users."
@pleaseclap @DemonHusky @bloomington_in In context, the rest of the post says that exceptions are published in city code and provides an example link to the table in our code.
I’m suggesting to use real speed limits from city code. Sorry if that wasn’t clear.
@markstos @DemonHusky @bloomington_in So the situation is, lots of low-speed streets are currently assigned a higher speed by default in OSS, and fixing that is a helpful service?
That's my bad: sorry for being a deviant who assumes the worst
@pleaseclap @DemonHusky @bloomington_in Yes. Many routing services, including Google Maps, use OpenStreetMap data to some degree. The more accurate the data is, in general the better the algorithms can work.
If you know of times where you think you haven’t been given the best bike directions, the issue might be solved by updating something in OpenStreetMap. Which like Wikipedia, anyone can edit and it’s not to hard to get started.
@markstos @pleaseclap @bloomington_in last year a group of us we confused why Cambridge was so poorly rated on the People For Bikes city ratings and realized that the city changed most street's speed limits. After going through and fixing them in OSM, this year Cambridge jumped up significantly to be rated much closer to what I'd expect.
@DemonHusky @pleaseclap @bloomington_in Ah, so the correct speed limit needed to be lowered in OSM?
@markstos @pleaseclap @bloomington_in yeah, I think when the state database was imported many years ago it included the at the time correct speed limits for streets, though most of them were unsigned. Then a couple of years ago the city reduced most streets with signs to be safer, but no one got around to fixing OSM.
@markstos @DemonHusky @bloomington_in
Besides speed, there's probably also an "is there a bike lane" check, which would mitigate the problem a bit in places where the main roads mostly also have bike lanes
@pleaseclap @DemonHusky @bloomington_in That’s right. Many signals are involved. Some factor in cyclist aversion to hills for example.