Since the previous poll made it to the wiki and is frequently used as an argument, I think it should be rerun with more precise questions, as I think the binary poll distorts reality. I suspect most people do not use just one single dimension when tagging tracktype (also based on the discussions). But maybe I am wrong and indeed it is one-dimensional. The other dimensional is undefinied, as it is clear it is not very clear what it is, but it is clear from other threads it exists. The question is, how strong is it?
I tag tracktype solely according to its firmness and nothing else matters.
I tag tracktype mostly according to its firmness with some regard to some general quality like usability/utility for vehicles/maintanence or construction sophistication etc.
I tag tracktype equaly according to its firmness and some general quality like usability/utility for vehicles/maintanence or construction sophistication etc.
I tag tracktype mostly according to some general quality like usability/utility for vehicles/maintanence or construction sophistication etc. with some regard to firmness.
I tag tracktype solely according to some quality like usability/utility for vehicles/maintanence or construction sophistication etc. and nothing else matters.
When considering rewriting the wiki, I think it is important to take into account not only practices but also how the resulting data will be used. Based on this summary:
I would consider, for a possible redefinition definition of new tags in the wiki, how useful and stable each criterion is and whether independent mappers can verify it reliably. Here’s what I think:
I would vote for these if it meant tracktype could be redefined to just overall quality/usability of tracks - both camps (generalist and many_taggists) would be happy :-).
Thanks for creating this poll, @supsup. The difference in this vs the previous poll is a great example of how results can be dramatically different depending how the question and answers are worded.
The linked previous poll was worded in a way that I could not even participate.
This poll is worded in a way – I pondered that for a bit – that I could select among three choices and it would not make much of a difference in my mappings, when grading tracks.
That uncertainty may stem from my regional background. Firmness comes at a cost. Bulldozing, grading, laying a base, covering the base, creating a reliably firm surface – regarding prospective users and their prospective modes of transport, maintaining that state, &c. Whether or not any or all of that and too what degree, and so on. So much to why I did vote the way I did vote.
The poll confirms it, only 1 out of 26 voters tags purely on firmness. That’s hard to ignore.
It also explains why the wiki and editor presets have diverged from actual practice. The presets push a firmness-centric interpretation, the wiki’s grade descriptions implicitly encode construction and maintenance, and mappers end up somewhere in between depending on regional context.
Rather than picking one “correct” interpretation, a more honest approach would be to update the wiki to acknowledge that grades represent a blend, provide examples from different climate zones and road types, and allow regional pages to refine the descriptions for local conditions, similar to how highway=* classifications work.
This would be a significant loss. In large parts of the world unpaved roads are the norm, and tracktype is one of the few tags capturing meaningful quality differences between them. If the concern is overuse where surface + smoothness is already sufficient, the fix is better scoping in the wiki, not removal.
I wouldn’t be happy because it would be an easier to tag but less verifiable, meaningful and useful version of smoothness
I phrased the question there under the assumption that there were two “schools” of tagging practice, and that the poll would determine how large each of them is. Unfortunately it appears that there are almost as many “schools” as there are mappers .
I also voted on behalf of mappers who tag it based only on what they read in the iD preset or the StreetComplete quest. Not a small number, I suspect
On router usage: OsmAnd and BRouter use tracktype explicitly. Valhalla and GraphHopper parse and store it. OSRM doesn’t. I believe renderers currently only act on it for highway=track.
On non-track highways: in much of the world highway=unclassified or highway=residential describes network role, not physical condition; a village road in rural Southeast Asia or Sub-Saharan Africa can be physically identical to a highway=track. tracktype is already used in combination with these classes in the database; if anyone has stats on that distribution (@ftrebien ?) it would be worth sharing. Removing tracktype would silently discard that data.
I worry more about mappers who think highway=track must be unpaved and two-tracked, and that all two-tracked and unpaved roads must be highway=track. To illustrate:
is a highway=track because it gives vehicle access to the farmland. But it was mapped as highway=service because it’s paved and doesn’t show two tie tracks.
is a connecting road (grid road), a highway=unclassified, but is mapped as highway=track because of the two tracks. The asphalt has been applied later, and no mappers have taken this as a sign to review the tagging.
Since unclassified has also been applied to 5,12 gazillion residential ways by an import in the early days, unclassified now means essentially nothing in Nederland.
Is that maybe uniquely British problem? To me as a nonnative (and anyway more influenced by the American variety, sad as it is) speaker, I know “track” as a noun almost solely from OSM. I originally thought it means “unpaved road” and never made a mental connection that you need visible tire tracks.
This poll might be getting extra votes in the most popular option because it uses an OR qualifier covering two different criteria: usability/utility for vehicles/maintanence (gets some votes) OR construction sophistication (gets some more votes). So I consider it possibly biased. Ideally this sort of poll should be multiple choice and should separate each criterion cleanly.
Quite possibly. Polls are frequently biased. The the previous poll only offered two options. One had to say that they used tracktype to indicate both “state of development” AND “maintenance”, or that they used tracktype only to indicate firmness. There were no options for only “state of development” or only “maintenance”, so voters who would have chosen either of those didn’t participate. Designing a perfectly unbiased poll is difficult and that is why we shouldn’t put too much weight behind any individual poll results. They are still somewhat useful though. In my experience it seems that a wider group of people will share their opinion in a poll (as long as it’s the first post in a topic), than will take the time to write a post sharing their opinion. So it’s a good way to sample a bit broader set of opinions as long as we don’t take the results too seriously.
As I say in the first post, it is about refuting the notion that tracktype is firmness. There is a wide agreement that apart from firmness a different aspect is used, but as the many threads atest, it is difficult to exactly say what it is.
.
