There is no way to mark an intersection as dangerous or to be avoided, no. By sitting there for a long time waiting to cross, you are allowing Waze to collect data which shows probably a much slower speed than other nearby recommended intersections.

Mayhap time-of-day restrictions aren't the right tool, but time-of-day de-rate might work (or at least help). Straight travel time calculations (what I think Waze does now) tend to ignore safety, and safety is at least part of what we're trying to accomplish, I think.

Taco909 wrote:The real-time speed/junction transit time information (or lack thereof, and I think the "lack" is the problem) seems to carry a much higher priority than the FC.

I guess what I was thinking of when I made the remark is a new feature or function that would mark a turn as "not a good idea" or "really bad idea" or even "don't even think about it" based on time/day. As a new feature it could be designed to have a higher weight than travel time.

The general question of whether editors should explicitly try to account for traffic conditions, rather than letting Waze do it automatically, comes up a lot.

See for example this thread regarding a left entrance immediately followed by a right exit in San Francisco. That was a bit different in that there is no way (I think?) Waze will learn how hard it is to execute at certain times. But the conclusion was the same, that it is best for editors not to force the map to discourage maneuvers that are legal but difficult or dangerous in traffic. I don't think any of us regard that as an easy choice, because, quite frankly, being given seemingly impossible or even suicidal instructions sucks for editors just as much as for any other driver.

If it's any consolation, the other nav apps have trouble with this too. In fact I would guess that Waze handles it better than most because it can learn at least in some situations.

I was dealing with a different kind of dangerous routing. In downtown Wilmington, DE, a major commuter roadway into town goes through the most dangerous neighborhood in the city (Hilltop). You're OK if you don't go off the main road -- safety in numbers and all. Problem is, the main inbound artery, Lancaster Avenue, backs up during rush hour every morning.

So Waze was routing people down a parallel side street here ---
https://www.waze.com/editor/?env=usa&lo ... ts=9250148

Now every local person knows not to dare go down there. But an out-of-towner wouldn't know and, if you look at street view, it's very easy to get yourself blocked in there with nowhere to go.

Thankfully the FC changes that happened in the state upgraded Lancaster Avenue from primary street to major highway and that has stopped the routing down that side street.

Reference: http://www.delawareonline.com/story/new ... s/8034085/
http://www.phillymag.com/news/2014/03/2 ... y-country/

I think that what constitutes a "dangerous" intersection is often very subjective. I've certainly had the experience myself of using Waze in an unfamiliar area, and coming up to some stop-sign intersection on a busy street and thinking I'm never going to be able to cross this street. But then a traffic light turns red on the next block, there is a break in the traffic, and I am able to get across, possibly with less delay than if I had waited for the traffic light on the next block with heavier traffic.

Probably there are some intersections that really are too difficult to cross during peak traffic hours, but in reality there are many fewer of these than the UR complaints would suggest. In nearly all cases, I think that letting Waze learn from actual traffic data is the best solution.

Agreed... the best we can do is attempt some kid of time of day restriction, and then we run the risk of getting it wrong.
I deal with time of day programming in setting the timing on traffic signals, and even with months of documented historical data, it is still impossible to predict when EXACTLY the time changes are needed.
Crap hits the fan on the freeway and suddenly I'm wishing I was running my 5pm plan at 11am.
Holiday Monday? No need to run the 5pm plan at all.

So rather than rely on the engineer's "guesstimate", we are moving into traffic adaptive and traffic responsive programming that will modify the synchronization plans on the fly based on actual traffic demand in real time, including reversing the directional priority if volume justifies it.

This is what Waze does with routing, and hopefully it will eventually learn to properly interpret "Suggested route frequently ignored" as user recommendations NOT to use a particular route, rather than throwing them out as Map Problems to be looked at and closed as unresolved.

bwawsc wrote:Mayhap time-of-day restrictions aren't the right tool, but time-of-day de-rate might work (or at least help). Straight travel time calculations (what I think Waze does now) tend to ignore safety, and safety is at least part of what we're trying to accomplish, I think.

Except that you can't de-rate lower than "street" and those are what I generally get routed onto when I am given really oddball detours.
So if Waze doesn't have a problem routing me off of a mH or even MH and onto a Street to reach a perpendicular MH instead of waiting at the signal to make a left, I don't think derating a Primary to Street is going to help a whole lot.

I've even elevated the FC of a number of roads to attempt to drive better routing and met with extremely limited success. The real-time speed/junction transit time information (or lack thereof, and I think the "lack" is the problem) seems to carry a much higher priority than the FC.