Page 55 of 57

Re: Road Types (USA) – comprehensive overhaul of drivable ro

Posted: Wed May 14, 2014 3:48 am
by vectorspace
I have some thoughts that may be random and naive, but wanted to share them here for comment so I can learn more about this issue. I have some observations, conclusion, and suggestion.

Nnote has taken some detailed research regarding AZ FC and tried to apply it to the state. Being his neighbor, we said NM would try to collaborate on roads crossing state lines. I am in a dialoge with Nnote, and my mentee :) irowiki who is based in the Farmington NM about how we should do this.

What conclude (for now) by looking in AZ to see what they did I see two things I will try to describe more in a bit:

(1) Phoenix, a rather large city, is comprised of a bunch of Minor Highways and Major Highways. This seems odd at first. It seems biased toward having Major Highways every half mile or every mile or so. Take a look. It seems highly differentiated.

(2) Rural areas of AZ currently seem biased to be less highways and more primary streets and streets. So the major arterial between distant cities can be primary streets. It seems under-differentiated.

So, does this make sense?

I can see some logic for (1) in that if you truly want to provide differentiation in a highly populated urban area, you need to bring in more use of MH and mH to augment all the streets and primary streets with different levels of traffic and usage. If you don't do this you under-differentiate for the amount of road use.

It seems that (2) is a problem for me. It seems that it is applying the same rules (1) in a city like Phoenix to the rural areas. In that case, fewer cars go over a major road between cities than go down a neighborhood street in Phoenix. If you use that as the benchmark, then you mark everything as a street. That leads to under-differentiation in the rural area. I think from all I have learned as an editor that you need to at least have mH connecting cities.

So that was the observation and a conclusion about the AZ work so far, which, like all of us is struggling with how to apply a new FC in different ways. I think that what we don't want to do is look at every state's FC and apply it absolutely. We are working at least a national problem here about how Waze algorithms route users. It's not about how the various state's view the world. That's why we have a Wiki FC guide. Even with that though, there could be issues.

Given the observations (1) and (2) above, it seems to me that what is needed is a sliding scale of FC that provides optimum dynamic range (sorry, the engineer in me coming out) that slides between urban areas and rural areas. Our "dynamic range" is F/Ramp-MH-mH-PS-S-Dirt. That is a dynamic range of six (6) levels. If we apply that across all roads absolutely then we don't have enough dynamic range to cover both urban and rural areas. If we interpret FC slightly differently between urban areas (like AZ is doing for Phoenix now) and rural areas (like having MH and mH betwen cities all the time), then we create a much wider dynamic range for the whole country, probably on the order of a synthetic range of nine (9) or ten (10) levels. Of course this needs some flexibility to allow the transition and special cases.

I have always argued that very rural dirt roads can be street, primary street, or minor highway. This is an example of the rural interpretation of FC that allows better dynamic range in a rural setting rather than just typing everything to be dirt or street.

Hope that makes sense.

Re: Road Types (USA) – comprehensive overhaul of drivable ro

Posted: Wed May 14, 2014 5:00 pm
by vectorspace
Great discussion. I want to make sure it is clear that I commend the AZ editors under Nnote's and Fredo-p's leadership in trying this out to see what happens and help the whole Waze community decide what to do.

I've already rambled on earlier with my thoughts, but will reply based upon Nnote's and Fredo-p's response.

I really like ply8808's comments as they have also captured many of my thoughts. His points on meeting criteria are good - but again, I would use them as a guideline that can be adjusted if required.

I worry that we should not focus and implement absolutely any state FC designation just because it exists or is easy.
  • Our primary focus should be on the optimization of Waze map utility to the routing staff and programmers at Waze.
  • Our adherence to a common but flexible Waze Wiki should come above our adherence to a state-based FCs.
That is why, given all the current constraints, I think we need a dynamic FC policy that gives maximum road differentiation in urban and rural areas, or any other area that has unique features (such as USFS, BLM, Indian Reservations, Park Service, etc. roads).

I heard that Fredo-p acknowledged the under-differentiation issue in rural areas, but the focus went back to state and city FC rules without mentioning a way to solve this (other than perhaps just accept it). I don't think we should ever consider ourselves strictly beholden to such rules as they are not always right for Waze and they vary from place to place.

Potential Process:
  • Look at local FC for guidance and an easy place to start - but don't feel beholden to these.
  • Adapt with general rules such as ply8808 suggested and as should be documented on the Waze Wiki along with what is there already.
  • Consider best ways to ensure roadway consistency across state lines. That is, some segments of road going from one state to another may have to bend to the other state's rules for some distance to help ensure good inter-state routing.
  • Consider maximum differentiation of road types and local needs add final touches.

Re: Road Types (USA) – comprehensive overhaul of drivable ro

Posted: Fri May 30, 2014 7:02 pm
by vectorspace
Hi everyone...

I'd like to address nnote's and his teams interesting experiment in FC within AZ. I just sent him a M before I discovered this thread (I am behind) so will post here. I think I posted on this a while back and will summarize here, but will also comment on FC in AZ. I think an FC experiment in AZ was interesting and may help us learn, but I think that it needs to be redressed and adjusted.

Summary from before:
It seems that the AZ FC work has increased differentiation of roads within large cities at the expense of differentiation of roads in rural areas. Phoenix has Major Highways going all over the place that are just streets, even what I would call Primary Streets. Is that right? I don't know. I get the idea, but don't know it makes sense when you look across major cities in the USA. For the rural areas some US highways are now primary streets and there is no difference between streets and highways -- they are all streets. This is a major issue with differentiation that would allow Waze to route properly. There seems to be a major disconnect between this and the FC rules for the USA and the states that surround AZ. I had also mentioned the need for a dynamic FC -- that is one that favors better differentiation within the cities like nnote was trying to do along with favoring differentiation in rural areas where long distance travelers must go.

New Thoughts

Without trying to be disrespectful, I would say that this AZ FC experiment is "naive" in that it is focused at optimizing the cities in AZ without respect to many other things: Rural areas, surrounding states, national FC policy in the WIki, etc. It does not likely account for long distance and rural routing. There can be some bizarre issues that come up for long distance routing. Sketch has already outlined some examples.

I am not for absolute rules anywhere. I don't think that we need a Waze Wiki Thumper attitude nor absolute adherence to a set of FC rules from one source in AZ that is optimized for something else or incomplete, certainly not optimized for Waze. We need guidance that we can adjust and tweak for local needs that doesn't break the national standards too much.

With regard to nnote's comment on Indian Reservation Roads, I have been doing a ton of research on this, am in contact with the primary road engineer at Bureau of Indian Affairs, Dept of the Interior, and some other places I have posted on other threads. Any road in the BIA system is open to the public. The Navajo Nation (AZ, NM UT) is as large as West Virginia. It is completely independent from state rules. We should treat Native American Indian Reservations as separate entities that are not part of states... within reason.
nnote wrote:As far as keeping things the same crossing state lines, we do need to look at this issue, but before time is spent debating it, if everyone looks they may find that their states FC does agree with the neighboring states FC where highways cross state lines. We can argue wiki etc etc, but if the FC's actually happen to agree, then what is the argument about?
The argument is really about proper routing for Waze users. It is also about working a common set of standards that Sketch and others (like AndyPoms) have put a great deal of time into in the past and recently. These are not just arbitrary rules. They are optimized for Waze, not a state road engineering office that has other purposes. We should start from the Waze FC rules in the Wiki and and then adjust for local needs. We should not start with state standards and then adjust toward a common standard. To do so would create a chaotic patchwork of the USA map. I value diversity, but not chaos.
nnote wrote:I'm all for coming to a common classification for where roads cross state lines, but lets get the official data first, without making up our own. And where there's differences, lets discuss those.
I think the point is that a state's functional classification is not official data, per se, it is a set of data optimized to something (like federal or state regulations) and focused on a particular area. It is not optimized for Waze (even though we may seek guidance from it). We do "make up our own" and that is how it should be... the result is researched by various people and published by Sketch to consider not only road types, traffic, etc., but the routing reality of Waze. This is the official data, not a regional FC document. We need to interpret regional FC documents in the context of the Wiki research that we have that optimizes for Waze.

I am very clear on what needs to be done in the rural areas. US highways need to go back to MH and there needs to be differentiation. We need to work state line issues.

I am not clear on what was done to large cities like Phoenix where all primary streets and some mH are now Major Highways. I find it extreme, but don't know enough about the effects and understand the intent was improved differentiation within the city.

If I had any brief summary suggestion, it would be this: Revert AZ back to Wiki standards except for the big cities where you're doing an experiment in increasing road differentiation. Work state boundary issues. Work on Native American Reservations as a separate entity, particularly the Navajo Nation that I have been working on and researching.

Re: Road Types (USA) – comprehensive overhaul of drivable ro

Posted: Sat May 31, 2014 10:52 pm
by vectorspace
Fredo-p wrote:
vectorspace wrote: Without trying to be disrespectful, I would say that this AZ FC experiment is "naive" in that it is focused at optimizing the cities in AZ without respect to many other things: Rural areas, surrounding states, national FC policy in the WIki, etc. It does not likely account for long distance and rural routing. There can be some bizarre issues that come up for long distance routing. Sketch has already outlined some examples.
Just to clarify, this isn't an experiment and it is, in no way, focused at optimizing the cites. All of the changes we made are in direct reflection of the Arizona DOT FC map (which is current as I have been in direct contact with ADOT). There is no way this should ever be called an experiment.

We are confident and comfortable with the changes that we have made here in Arizona. Saying it's an experiment without confirmation is a bit bold. It may seem difficult to fathom the changes we have done, but, if anyone outside of Arizona participated in our conversations, you would have understood how we came to the decision.

Before any drastic changes are done on anyone's behalf, please feel free to chat with us so we can enlighten you on how we came to this conclusion. Sometimes change is a good thing. ;)
I suppose the interpretation of "experiment" is not always clear. I completely believe this is an experiment because of a few things ... (1) You have no in-depth experience and history with the results. (2) A state FC is not Waze and the fact that you're in contact with AZDOT is quite good, but really doesn't matter as an absolute touchstone to me because it is not mixed with the Waze routing effects. (3) It does not match the Waze USA standards and was not coordinated widely with USA Champs and the opinions you're seeing here (not just from me) are a result of your experiment. This was a AZ-focused and internal effort. Some of these changes were in direct conflict with the research I was doing for Indian Reservation Roads on the Navajo Nation and changed much of the work I have already done without notice to me (but I really should have guessed it would happen).

So, I mean it was an experiment (which is good and bad -- I like bold approaches), but I don't consider it finalized. I talked to nnote and am reversing some of these things on the Navajo Nation.

I would like to point out that what was done in AZ is not what was done by Sketch in the national standards. AZ has done something much more bold, and in my opinion not completely correct or final. I really think we need to revisit what has happened in AZ and find the good in it and consider what needs to be changed. For instance, on the Navajo Nation, I have changed US-191 back to MH because it is contrary to what I have been doing for Indian Reservation Roads and the connectivity of the Navajo Nation more broadly.

Re: Road Types (USA) – comprehensive overhaul of drivable ro

Posted: Sat May 31, 2014 11:09 pm
by vectorspace
I just read the whole recent thread again, and appreciate the reconsideration of the AZ FC work. I want to mention that I am not upset about it because I appreciate the positive effort that went into this process and hope we learn from it. I actually appreciate all the research and effort. I am not advocating that everything should be changed or changed all at once.

I really like the idea of adherence to the USA standards that were published on the wiki, and then trying to change those first if it is of national value. If there is need for local variation with additional state DOT FC and other input, then make those changes in cities and parts of states, not rigidly across the whole state -- and consider blending with neighboring states.

What I think this whole approach may have helped highlight is the need for this or similar guidance that Sketch could insert into the USA FC guidelines... that is, now to adjust these guidelines for state and local FC issues.

Re: Road Types (USA) – comprehensive overhaul of drivable ro

Posted: Mon Feb 16, 2015 1:19 am
by voludu2
Then the style/best practice for this should also clearly state that any MPs which arise as a result should be closed "not identified", just to be clear that we aren't going to let an MP tell us what to do.

Re: Road Types (USA) – comprehensive overhaul of drivable ro

Posted: Mon Feb 16, 2015 7:23 pm
by voludu2
txemt wrote:
No, don't ever close an MP as "not identified." IGNs get involved then.
If you close it solved, it will return before the end of the week.

Have you found any recent IGN edits to segments?

Re: Road Types (USA) – comprehensive overhaul of drivable ro

Posted: Fri Feb 20, 2015 9:52 pm
by voludu2
I believe there are already guidelines for AGCs used for U-turns etc.

Because it is unusual to have a legal crossover/U-turn on a freeway, ordinary measures for protecting unusual edits could be used -- lock the segment, leave a UR [NOTE].

I think this is simply a rarely-seen application of a well-understood guideline. How much commentary, or commentary-upon-commentary, do we wish to see in the wiki articles?

Re: Road Types (USA) – comprehensive overhaul of drivable ro

Posted: Wed May 20, 2015 12:00 am
by voludu2
Service road. They have all disappeared from the map now,many are not available to be added.
Should we remove service road from road types?

Re: Road Types (USA) – comprehensive overhaul of drivable ro

Posted: Mon Jun 15, 2015 8:44 pm
by voludu2
I would like to clear up some conusing statements in the Parking Lot Road and Private Road sections without changing the meaning at all. ... g_Lot_Road
* The proper use of parking lot roads can also help to avoid automated traffic jam reports as well as Map Problems related to Wazers driving in unmapped parking lots. Draw in the drivable portions of the parking lot that are near streets and other roadways. This will prevent the Waze routing server from assuming you must be on the main road when in fact you are stopped in the parking lot.
* Parking Lot Roads can be used to prevent Waze from assuming drivers driving slowly or parked in the parking lot are in a traffic jam in the main road -- draw in the drivable portions fo the parking lot that are near outside roadways.
* Waze will not highlight slow speeds (automatically detected traffic jams) on parking lot roads
And then, in the private road section:
As with Parking-Lot Roads, a route over a Private Road will incur a transition penalty upon leaving it for another road type. This transition penalty keeps Waze from routing Wazers through a private area as a shortcut. Unlike Parking-Lot Roads, however, Private Roads do not suppress automated traffic-jam detection.
As of August 2014 the Waze client displays Private Roads identically to public roads. Since this may change in future releases, never use the Private Road type for unrestricted public roads. Do not use the Private Road type as a workaround to force Waze to avoid slow public roads, for example those with rough patches or construction. In special cases, however, the Private Road type may be used for a public street that is legally and enforceably signed for local traffic only.
* Never use the Private Road type for unrestricted public roads
* Do not use the Private Road type to try to force waze to route around slow (damaged or under construction) public roads. Talk to a routing expert to find out if there are any good solutions.
* Private road may be used for a public street that has a legally enforceable sign for local traffic only.
* Waze handles private roads similarly to how it handles parking lot roads, but not exactly the same.
** Similarity: Waze routing adds a penalty to a route that goes from a private road segment to a segment of a different type.
** Difference: Waze will highlight traffic slowdowns on private roads.