doctorkb wrote:The visual distinction of direction of travel isn't necessarily there. And by having both appear on the map (though not necessarily next to each other consistently), this could also add confusion.

It is for WME editors, not as much in the client where ramps are named (but TTS works better in areas where ramps are not named yet).

doctorkb wrote:Before we make ANY changes on this, I'd propose we take one of the examples from Vancouver region, and add the Alt-name to the ramps and overpass, allow for a tile update and test again. If it then chooses that as the best route, I'll concede that BDP is actively working against this scenario.

That said, I would expect that if it does NOT change its best route option, we put BDP out of the equation in determining the approach we take here.

Why would we go with the Alt-name test instead of a cardinality test? If we add cardinality and that doesn't change the results for the test cases, then we can rule out BDP.

Actually, if we add Alt names to every turn-around, won't that cause BDP to stop working at all (because everything is named Hwy 1, for example)?

My points are not ridiculous, they're logical. You're trying to run a "test" with 3 balls in the air on a skateboard. I'm trying to isolate the test case to only the bare essentials, changing only what needs to be changed so that the test case results are not impacted by outside factors. I do this for a living, but whatever.

Your test, your show, your agenda, run it however you want. I'm out.

You are correct, this is my only true argument for keeping cardinality, so let's keep the focus on that. All the other points I have made are [admittedly slight (but not ridiculous)] pros (not arguments at all) of the cardinality approach, none of which are strong enough to support the cardinality approach without the turn-around problem.

I still stand by my earlier point that a road over/under the highway having an alternate name of the highway name is illogical (hack-ish). Some additional things I'm concerned about include impact to searches, street numbering, etc. but I'm not sure they are well-founded. Since you stated they won't be applied everywhere, I'd want to hear more about the guidelines that would be used to tell editors when to apply them and when not to apply them.

However, since large sections of the country already have [currently non-standard] cardinality, why not test whether the approach has merit for the turn-around problem? If the tests indicate no merit, this argument goes away and we can begin the work of removing it.

As far as the scope of the test case itself (for cardinality test), the theory is that the wazer missed a turn, so the only segment naming that should matter are those freeway segments past the missed ramp and part of the turn-around back to the original interchange. For the test case we don't need to do large area changes.

If you think I'm correct with my scope of affected segments above, I would be happy to run the "negative" test case (currently using cardinality, test case would be written to temporarily remove it) in my area east of Toronto. Before changing anything I would write the test case and post it back here for comments and buy-in, and I would require support from a level 5 to make the changes, capture the results, and post them back here.

If the test case confirms that BDP is preventing turn-arounds, I could then also run the other test case using alt-names in the same section of highway (again, publish the test case first, then run it and post the results).

I think I see your point -- because we're not starting from a known BDP, there's no guarantee for my test case that when I remove cardinality that BDP will kick in.

Where do we go from here?

We have significant parts of the country both with and without cardinality. Do we simply recognize that in the policy and suggest to follow what has been done on other roads in the area? Possibly as an interim policy (so that no further work is done either adding or removing cardinality until we can reach a common solution)?

Or, do we want to continue brainstorming a solution for a while before updating policy?

I'm following the discussion on this topic and see that it transformed into a BDP discussion and I only wanted to bring another aspect that supports cardinal directions on Highways (where possible) and it is mainly to make the reports posted by Waze (App, Twitter, etc.) meaningful. A report on accident on Highway 401 without knowing the direction is very broad and useless.

Before we make a decision I think we should also consult with our peers in other countries, as we may want to know what is been done in other countries.
We can come up with our own rules and they could be different than other countries, but I feel that the issues that were raised by people in this forum thread are not Canada specific (BDP, map clutter, etc.).

Thanks for the initial research. It's a good start and definitely showing that there is no consistency, whether intentionally or not.

Doctorkb, manoeuvre and I will coordinate with our Global Champs peers to discuss this and see if this is a global issue or not.
While it isn't necessary that all communities around the world would be uniformed in this regards, we need to make sure that if there is an issue with BDP and this is why some communities are choosing one way or another it isn't because of something that Waze Dev Team could work on solving in other ways.

We agreed that we should have the same standard all across Canada, unless there is a good reason not to, so after we get to the bottom of this issue we should have it all the same.

Since it isn't an urgent issue, I don't suggest doing anything until a decision is taken among the Canadian Champs.

Existing policy aside, I have one specific, repeatable example: when you miss your exit.
(other example would include starting navigation for a place that is behind your current position on the freeway)

Let's say, you're doing this route:
https://www.waze.com/livemap?zoom=15&la ... t_text=Now
(The destination being in the urban fringe is irrelevant.)
But for whatever reason, you miss your exit (a wall of 18-wheelers let's say for now). Waze will attempt to re-route you.

For now, let's assume Waze will talk back to the server, and it will return what it 'deems' to be the best route. (The client re-route is a different story, but it will fall back to Waze server.)
https://www.waze.com/livemap?zoom=15&la ... t_text=Now

Waze will re-route you to your destination via local roads. Now compare this to when I put the origin pin on the next off ramp:
https://www.waze.com/livemap?zoom=15&la ... t_text=Now

Notice that Waze now re-directs you back on the highway, effectively a legal, safe turn-around. The time difference is significant--5 minutes, versus 8 minutes by local, in the middle of the night. This is likely the effect of big detour detection, where Waze will actively avoid allowing you to go back to the original F/MH/mH (original being defined as, matching name and/or alt-name), after more than one segment (i.e., it's how Waze prevents "Hwy X->ramp->traffic light -> ramp->Hwy X"). Without cardinal directions, Waze interprets the opposite direction of the highway as the same manoeuvre as "Hwy X->ramp->traffic light -> ramp->Hwy X", and penalizes that route, which is why this alternative wasn't seen in the 2nd permalink (starting point was on the highway), but was seen in the 3rd (starting point was already off the highway).

Compare that 2nd permalink to this:
https://www.waze.com/livemap?zoom=14&la ... xt=18%3A00

Cardinal directions are set up here, and voila, the desired, fastest, routing is #1. Missed exists are not uncommon--if we're going to give a route, we should give the best one, not route one twice as long. These are not dangerous manoeuvres. A Garmin can route this safe re-route.

In fact, the current policy is fairly explicit on the Wiki, although hard to find:

A road may be divided when any of the following conditions are met:
It is a [...] Limited Access Highway using the "Freeway" road type,

and

Note that the trailing N, S, E, W cardinal direction indicator is used for numbered highways and interstates which are split into two 1-way segments, per the guidelines.

So, any freeway, should be divided and have cardinal directions.
(Wild conjecture: perhaps that's why we always divide freeways too, so that this very manoeuvre can be allowed.)

Also note the "no cardinal directions" thing seems to only be in Western Canada, west of Thunder Bay, Eastern Canada seems to be doing the opposite.

Perhaps the guidelines have some missing points:

How about ring roads, or other non-linear freeways (QEW)?

As it's done in Toronto, the engineers have avoided using cardinal directions altogether, and simply signed the highway consistently with the next control city (Toronto, Hamilton, Niagara, Fort Erie). This is how it is currently mapped in Waze as well.
Anthony Henday in Alberta, is a ring road. But Albertan human factors engineers are not stupid--the highway seems to be consistently signed according to the cardinal direction that you are currently travelling in. We can name it the same.
(Heck, in Japan, ring roads and ring metro lines are signed with clockwise and counter-clockwise directions!)

What if it's Hwy 97 N, northbound?

Well this doesn't qualify to be a limited-access expressway, but if it were ever to become one, well we'll see what's signed. I'm more than happy to un-divide, and un-cardinal this at this point.

It adds clutter.

Sure it does a little, but we're getting shields soon(TM) right?
Also, this is clutter in the Android app right? Waze 4 on iOS has road names along the road itself.

It helps with shield generation.

As far as I am aware, shield matching is done with regex. I'd rather write a robust regex test suite, and test against that, rather than manually (or via a script), edit each and every segment in many parts of Canada. The US has shields--they have cardinal directions too. I don't see the problem. I've asked the US champs about it, it's apparently Waze hasn't gotten around to updating their generator. Plus, we're talking about a computer, there's no need to "massage" it to tell it everything will be alright--write proper and extensive tests, you should be good.

These last two arguments, personally, are rather superficial, and halving appropriate routing time I think would be more critical.

doctorkb wrote:#2 - The other solution to BDP issues is to add an Alt-name to ramps & overpasses with the non-cardinalized name of the freeway (e.g. "Hwy 1").

I hope you weren't serious about this. I thought we're trying to push Waze for accepting alt-names for shields too? This is a true hack, for effectively, a problem that doesn't exist if we follow the US guidance.

So, I'd say we follow (a perhaps better organized) US guidance.

Yes I did, I had it written up last night, so I didn't want to chuck it away

Well I guess it's a 'hack', but it is widely used in the US and Eastern Canada, so the style currently used in Western Canada, is a minority.

The US naming convention (according to TheLastTaterTot) was set before BDP was implemented, and it was an unintentional benefit of the convention (well I hope Waze wouldn't break routing across a country without notice either).