Talk:United States/2021 Highway Classification Guidance

From OpenStreetMap Wiki
Jump to navigation Jump to search

Just wanted to note that this proposal as is almost fully lacks locally verifiable criteria for the classification of roads. Instead it contains weasel words in abundance (importance, usefulness, best, high/medium/low population density) and refers to computable yet practically not mapper-verifiable criteria (properties of a connection in a routing graph).

It seems the main motivation of the proposal is to allow mappers to hand select cartographic importance ratings for roads to be used directly in maps. The ideas for high quality road cartography in automatically rendered maps go into the direction of computing suitable importance ratings (in dependence of map style, target audience, map projection and scale of course) from locally verifiable properties of the roads and the road geometries + road network topology and geographic context. What this page seems to show is a rough sketch on how such computation could look like and suggests mappers to essentially hand compute (or more likely: guess) such classification - in a static form independent of map style, target audience, map projection, scale etc.).

Now this is of course not unique to the US - elements of such interpretations of road classification exist elsewhere in OSM as well. But i would encourage you to - instead of taking that to the extreme - think about how to overcome this flaw and develop tagging for roads that documents meaningful locally measurable properties (physical properties as well as things like use frequency) that are of real use to allow computing meaningful road classifications for cartographic and other applications instead of tasking mappers with the impossible task of creating such a classification by hand in a consistent and universally meaningful way.

--Imagico (talk) 09:05, 18 May 2021 (UTC)

You probably know this, but one of the key challenges in sorting out highway classifications in the United States is the lack of uniformity between and within US states. I would encourage you to think of the US states as more similar to haphazardly-organized countries in the European Union rather than German federal states. The US community has tried for years to come up with a rules-based standard with little success, and almost everyone is unhappy with a physical-characteristics approach to road classification. Our signed road networks are generally based on who pays for the road's construction or maintenance and often have only a loose relationship with routing connectivity and importance. It is quite common for "state" roads to be built to motorway construction standards in places where "US Highways" are low-speed two-lane undivided roads and vice-versa. Paper maps of the US similarly are cartographically designed to show road importance in a way that corresponds to on the ground importance and is only loosely related to road network designations or physical characteristics. Traffic counts (where that data is available) are a useful way to compare two roads in a similarly built up area, but that approach breaks down when comparing urban and rural areas. There are also many cases where a highway construction project was started and then abandoned, leaving odd sections of upgraded road in places that don't make sense in the context of the surrounding road network!
Because the states are so different, it's not just hard but might even be impossible to come up with a strict rules-based approach that can work on a national or even regional basis. So far nobody has been able to crack that nut. For example, Massachusetts has a well-developed GIS culture rooted in the MassGIS system. They have an actual numbered functional classification system that maps almost perfectly to the highway classifications (and consistent with the principles laid out here) and so little thought is needed there to figure out which one to use. Meanwhile, in neighboring Vermont, there is no such classification system we've found that makes sense when mapped to OSM's highway classifications, and so local mappers there have hand-curated a list of the important cities in that state and determined the major routes between them. This works well in Vermont which has a low road density and is very mountainous so the most important routes between cities are topographically obvious. That state is also snow-covered for a significant portion of the year, and Vermont's snow removal priorities map was an important factor that those mappers considered when determining that network. Since roads cross state lines, these different systems require a bit of hammering out between states to make sure that the road network is regionally cohesive.
I realize this talk of hand-curated importance values makes you uncomfortable as a proponent of strict rules-based approaches to mapping (and indeed, rules-based approaches are obviously superior in settling disagreements and promoting community cohesion), but so far we haven't found a better approach. What you see here is the result of at least a dozen US mappers collaborating to sketch out a set of general principles that hopefully can translate into more quantifiable or rules-based approaches for each state. I assume that the US is not the only country in the world that is challenged by a lack of government-designated functional road classifications, and if you are aware of how other countries with similarly messy and complex road systems have solved this problem in the context of OSM, I would ask that you point out those approaches so that we can copy from the success of other mapping communities that have faced and solved the same problem. --ZeLonewolf (talk) 15:31, 18 May 2021 (UTC)
Not sure if you are aware of that but when i provide some reasoning as to why i think this proposal goes in the wrong direction and you react to that with characterizing this as making me uncomfortable that is kind of insulting because it reduces an argument founded in reason (which i maintain unless someone provides reasoning to the contrary) to a purely subjective expression of personal emotion (discomfort).
Anyway, I am not expressing a personal sentiment here regarding what i would like road mapping to look like, i am providing arguments and reasoning regarding in what direction road characterization and classification in OSM should develop for the long term benefit of the project (to provide meaningful data, to create the best map of the world). Whether it does will not depend on how the US mapper community decides to apply existing road classes. But it could benefit from a forward-looking approach in the US for sure.
--Imagico (talk) 18:41, 18 May 2021 (UTC)
Perhaps it would be more accurate to describe this document as a proposed set of principles rather than concrete criteria. After all, it isn't as though Key:highway is completely devoid of subjective guidance either. On the other hand, some of the state-level drafts are jumping from these principles to specific lists of affected roads. While it is important to consider the case-by-case impact of the tagging scheme on a state's road network, as a gut check, I think these drafts miss an opportunity to take the principles in the main document and recast them in state-specific terms that are more concrete and objective. Otherwise, there may still be edit wars over trunk classification, but now these disagreements will take place on the wiki (or in spite of it). By analogy, the Federal Highway Administration publishes a thorough manual on functional classification, but it leaves a significant amount of subjectivity and ambiguity for individual state departments of transportation to resolve based on their own road construction standards. Many state DOTs publish customized field manuals synopsizing the federal manual, in addition to the lists of specific classifications that result from the guidelines. – Minh Nguyễn 💬 23:20, 21 May 2021 (UTC)
Yep, the emerging consensus seemed to be that the FHWA functional classes didn't map well, and so far only Mass had a set of independent state functional classes that mapped nearly perfectly to the highway classes. @Minh Nguyen: would you be willing to spearhead an example of a state-recast of general principles, perhaps for Ohio? A concrete example like that would go a long way towards helping the various state authors! --ZeLonewolf (talk) 02:26, 22 May 2021 (UTC)

Given the mediocre British system we have inherited I think this is a good effort, & I support changing the current US wiki definition for trunk from divided highway to what you have here, in agreement with the main highway page definition. I say this despite the fact that 'trunk' is not often used in the US for a highway classification. I've always thought of primary as the top level below interstate/motorway. I can't vouch for the accuracy of this but: "The 'trunk' name is much less common in the United States with only three states, all in the upper Midwest, that call their highways Trunk Lines - Michigan, Wisconsin and Minnesota." https://k1025.com/michigan-trunk-lines/.

I keep seeing government classifications (NHS, FHWA) brought up, then rejected. But the British government system is the gold standard. Why? The NHS routes for Colorado seem like a good default for trunk except for the Denver & Co Springs metro area where that may be too many trunk roads. https://www.fhwa.dot.gov/planning/national_highway_system/nhs_maps/colorado/co_colorado.pdf . I couldn't find FHWA data or a map for CO.

I think a vote is in order since 'trunk' is sporadically used in the US, this definition is contrary to the current US highway wiki, and this will meet some resistance. I would suggest keeping the vote only on the trunk tag. --Bradrh (talk) 18:14, 31 May 2021 (UTC)

In some states there seem to be government classifications that work. Specifically, Massachusetts and New York so far have state-level highway classification that seem to map reasonably well to trunk/primary. It would be great if you could write up a bit more about what this might look like in the Colorado case and perhaps share it on Slack or on the talk-us mailing list. --ZeLonewolf (talk) 18:32, 31 May 2021 (UTC)

Clarity of level description "1-7" or "1a,1b, 2-6"

We've been talking about highway=motorway + highway=trunk forming a 'top-level' network. Would it be more clear to refer to "seven levels" of hierarchy in the way that the page currently is:

  1. highway=motorway
  2. highway=trunk
  3. highway=primary
  4. highway=secondary
  5. highway=tertiary
  6. highway=unclassified
  7. highway=residential

Or would it be more clear to talk about the hierarchy as:

  1. Top level:
  2. highway=primary
  3. highway=secondary
  4. highway=tertiary
  5. highway=unclassified
  6. highway=residential

I think this is a pretty minor distinction and is more about description and verbiage and wouldn't change any resulting tagging. I don't really have a strong opinion on this, but am trying to read through the documentation with fresh eyes and wondering if conceptually grouping highway=motorway and highway=trunk would make it easier to understand the tagging schema. --Adamfranco (talk) 15:39, 3 September 2021 (UTC)

@Adamfranco: Good idea. I think highway=unclassified and highway=residential may also need to be grouped similarly, based on the global definition of those tags, although historically some states have used highway=unclassified for a different purpose. – Minh Nguyễn 💬 18:38, 3 September 2021 (UTC)
I have remained 100% silent on this topic, preferring to watch more than participate. However, here, I'll add the following minor points. I find that grouping both a. highway=motorway and b. highway=trunk into a (sub-)group as well as a. highway=unclassified b. highway=residential into a (sub-)group is likely helpful. At least it seems so in California and possibly other states/places in the US. My reasoning for the former is that there are numerous places on California "freeways" (highway=motorway) where for a short distance to accommodate an odd, often at-grade, topographic, frontage road or all of the above roadway configuration, "regulatory" signs (black and white, which means "by law") will announce "End Freeway," then the odd zone, then "Begin Freeway" after the oddness ends. This is more common than you might think if you watch out for it. It may be that funding dried up, engineering for that segment is too difficult, expensive or will necessarily have to wait for funding a "full interchange" (with overpass/underpass and additional construction / infrastructure to be built). And it is correct to tag the freeway as motorway and the segment where it is not freeway as trunk. My reasoning for the latter is that if we are "classifying a hierarchy" (and indeed we are here with the numeric assignments), putting unclassified and residential into the same bucket is quite accurate in that both are "below" tertiary but neither one is "higher or lower" than the other. The only distinction is that residential "abuts residences" (or residential zoning / landuse). So, 1a/1b, 2, 3, 4, 5a/5b works for me, and I think California as well. I'm not so sure about the other 49 states, but we'll likely have many opinions about that. Stevea (talk) 18:36, 1 February 2022 (UTC)

Motorway and access restrictions

Hi from north of the border, but I think you might have forgot to mention regulatory signage that indicate a freeway (especially if it’s not an interstate, e.g. the Begin and End Freeway signs in California, or the typical "selective exclusion" signs placed at entrances), not to mention the varying laws on non-motorized access on freeways that has been brought up on local Slack channels. TagaSanPedroAko (talk) 21:45, 3 April 2022 (UTC)