[Tagging] Access restrictions and expressway=yes

stevea steveaOSM at softworkers.com
Mon Feb 22 10:44:43 UTC 2021


I appreciate Minh's cogent, patient, detailed, well-referenced, historically accurate, meshes-with-my-experiences-with-other-mappers in California and the USA and quite helpful-going-forward explanations here; thank you.

This feels like it might be parochial / regional to perhaps the way that the USA is structured as a "states do highways, but the federal government has two highway systems of its own" (the original US Highways and the just-about-done-after-70-years Interstates), with the federal government funding not only those (two federal highway systems) but also funding state highways to some extent, too.

Mapping in OSM benefits by worldwide structures (something like "universal tags" that work everywhere like amenity=bench or lit=yes, as well as servers that are always up and running, quick rendering, etc.), national cohesion where things can be counted on to be relatively consistent within a country-wide scope (usual or often, but not always — for example, rail can differ slightly from country to country, that's OK and OSM accommodates it well) as well as more regional ("states") and local (county, city, town) practices.  At this last level, where tagging and "standards" often come down to what one, two or a few local mappers in OSM begin to hash out as "correct and acceptable to this community," it can either go well where there is agreement and harmony or it can devolve to squabbling and edit wars.  I've seen both and untangling squabbling is difficult, often something like "authority" (a wiki, a local / state official who declares something as "just so," or the like) must be a final arbiter.  When this fails, the squabble continues, often getting "kicked upstairs" to someone like Minh to arbitrate a wiki topic or the DWG to root out and quash the bad behavior where it is found to be bad.  This is time consuming and seems virtually impossible to fully eliminate:  there will always be a need in a project like ours for "final arbitration."

The issue of HFCS is one of the most contentious, inflammatory topics I have seen give rise to disputes in OSM.  Minh's "a guideline that makes sense in California..." is spot-on.  What I'm agreeing with here is the "local sense" that mappers anywhere choose for highway designations that can and do drift from the strict definitions found in our wiki — and very certainly from those of the HFCS, to the extent those even DO map onto OSM's highway tags.  This "makes sense in the local context" can, in some cases, expand from localities to larger contexts of regions and nations, but not in all cases.  Where it makes sense for a "local sense" to impose sanity and regularity on the local road network, let's continue to allow OSM's volunteers who are local to that area come to agreement, even as this might bend strict wiki definitions.  Importantly, let us document once and for all that HFCS is a poor proxy to use for mappers to use to assign tags on highways whether those that are local to them or not.  HFCS simply doesn't work to do so, we already know this, so let's say so explicitly.

Maybe this discussion is more apt to talk-us, but it does seem that except in the UK, where OSM's highway tagging practices originated as closely as possible to be a one-to-one logical mapping, that simply isn't going to be the case in the rest of the world.  So:  let "local consensus" emerge (as it does), even as it may bend (but not break) our wiki definitions.  And let's toss into the garbage can HFCS (and similar "funding designations" around the world) and the chaos they create on mappers as "not apropos to OSM and how we map and tag."  This has been both a slow smolder and too often a raging prairie fire and we simply don't need it to burn down any more of OSM going forward.  I don't know this, but other countries may have similar problems with similar "systems," I encourage mappers there to consider taking the same, firm approach.

In a less-developed part of the world, I believe a dirt track can be "logically, locally accurately" tagged highway=secondary, especially as it connects cities or towns.  It might be wide, smooth and allow moderate-to-higher speed of a surprising volume of traffic.  But such a road would might never be tagged highway=secondary in the USA, even as that tag rather well-specifies how that road fits into the network hierarchy of travel patterns in that particular country or region of the world.  Note that as our wiki might need a bit of "stretching" to accommodate this, it actually does.  Sometimes (and in some places) OSM needs such wide latitude.

The proliferation of keys like motorroad=*, expressway=* and designation=* allow for such additional flexibility.  Our plastic tagging allows this and works well in cases like these.  "Funding," or other mis-applied "systems" of highway classification, do not.  While the USA really does have, as Minh notes, a "wild West," our map and tagging can (should, does) accommodate it.  Highways are ever-changing — they are built, modified and fit into networks in new and different ways as time progresses.  What our tagging must do is match the roads "properly in the local context," not systems from somewhere else or for another purpose such that they are contrived and therefore, wrong.  Perhaps we restore the HFCS wiki page, but simply say "don't use this, it is incorrect for OSM."

SteveA


More information about the Tagging mailing list