[leeoniya]

instead, how about giving me directions that dont involve 20 turns down tiny residential streets with 10 stop signs to travel 2 miles. when it can be done with 1 turn on 2 major roads with 2 traffic lights and an extra 0.1 mile? or multi-point routes on mobile? or a maps app UI that doesnt obstruct 60% of the actual map?

so many google products seem to reach EOL at "good enough". see also: Gmail, etc.

[puzzle]

Are you sure that's not Waze? Google Maps is deliberately a lot like the latter, even if both products are from the same company. Maybe there is something in the road network (e.g. a segment with bad information) along the route you are thinking of that results in what you describe.

[leeoniya]

i have never used waze. google prefers to minimize distance rather than minimizing route complexity (for some minimal distance or time cost). my wife and i have an ongoing joke that google takes you on the most "scenic" drives through the most obscure neighborhoods.

[puzzle]

Where in the world are you? If you think Maps has complex routes obsessed with shaving seconds, you should try Waze just for a day. :-)

[cuddlybacon]

Not the person you are replying to, but I find in Vancouver it does that a lot as well.

Instead of turning left at an intersection with a light and dedicated turn lane, I find it will often suggest I turn left then right then left to skip the intersection. It always turns out worse.

[puzzle]

Very interesting, thanks. Does the map show lane information? I don't know how accurate that is outside of the US. Or maybe in this case it's overestimating the impact of live traffic data for that red light. Sometimes it feels to me as if, over time, it learns from routes that it suggests and I repeatedly avoid, but that might have been just a coincidence and it might have learned that from aggregate data, not just mine.

To reiterate, though: at least until a few years ago, according to PMs involved, Google Maps was tuned to keep directions shorter and simpler to read, describe (if you're a passenger) or even remember.

[cuddlybacon]

It's pretty accurate about which lane you need to be in. I wish it would announce with more notice, but the info is dead on.

I find Google Maps _vastly_ underestimates how hard it is to turn onto a major artery without a light. It seems to think that turning left across three lanes of traffic without a light is "free". In practice, I have been stuck making these turns for 10+ minutes before. I've noticed it underestimates bridges as well.

> Sometimes it feels to me as if, over time, it learns from routes that it suggests and I repeatedly avoid, but that might have been just a coincidence and it might have learned that from aggregate data, not just mine.

That could be the case. Given my usage of Google Maps was less than once a month, it may not have had much to learn from me. I would also rate these poor experiences as bad.

[leeoniya]

> Where in the world are you?

Chicago

[anothergoogler]

How about excluding unpaved trails from bicycle directions? I've been screwed by that too many times. A gravel trail is not a bike route, for most bikes anyway.

[magduf]

Here's my suggestion for Google Maps: how about handling toll roads better? It's all-or-nothing. Why not put a little gray box saying "20 minutes shorter, $1.50", or "1 minute shorter, $29.50"? If you turn on toll roads, it'll happily always route you on a toll road or toll lane, even if it isn't any faster. And here in DC, toll roads can easily cost you $30-50 per trip.

Maps already gives you route options to choose from, both when you first start, and during your drive; why can't they do this with toll roads? Are they secretly working with EZ-Pass?

[vmarsy]

That's actually one of the suggestions I left them on the product.

I've the choice between toll-lanes or not on my way home, I'd love to know how much faster it'd take if I get on the toll lane...

But in that particular case, I'm not sure how that information is gathered in the first place, it might be hard for them to figure out which cars are on the toll-lane, and which cars are not since both the express lane and regular lane are next to each other.

[magduf]

Surely a simple filter algorithm could figure out the toll-lane thing: if you have a bunch of cars reporting speeds on what seems to be the same road, but you know that there's toll lanes there, and the speeds reported fall into two extremely different ranges (one crawling, one highway speed), then it seems safe to assume that the slow vehicles are on the regular road and the fast ones are in the toll lane.

Plus, while they aren't perfectly accurate, toll lanes are usually far enough away from the regular lanes that modern GPS devices in those lanes should be able to generally show themselves biased in that direction (relative to the devices stuck in bumper-to-bumper traffic).

[r00fus]

I have requested and gotten added several bike paths (all paved) to Google Maps so it would route properly through a park, so it might be someone with an mountain bike who requested Google to add it?

[leeoniya]

the data quality is probably not there for this to be too reliable. a lot of bike trails/paths are just plain missing.

[anothergoogler]

It is though. Those routes are colored brown on maps, and are likely as easy to exclude as interstate highways. The route text for those segments also indicates "unpaved trail."