In many areas, the state of development of a track road and the firmness of its surface go hand in hand. This key was invented in one such region, as can be clearly seen in the pictures. However, this correlation is not as pronounced everywhere.
I have read differing accounts and opinions on how mappers in such other places use the key, when they even consider it applicable at all. Stuck here, not knowing how to go ahead, possible completions:
Other aspects take the lead (which ones?)
Firmness remains the decisive aspect (no tracks below grade2 in some areas?)
…
Deliberately not a poll. Please comment!
PS: German version of above:
In vielen Gebieten gehen der Ausbauzustand eines Weges und die Festigkeit seiner Oberfläche Hand in Hand. Dieser Schlüssel wurde in einer solchen Region erfunden, was man auf den Bildern deutlich sieht. Dieser Zusammenhang ist jedoch nicht überall so ausgeprägt.
For the record, I appreciated your insights on the matter and I’m still willing to update the Wiki accordingly. However, not much input about classification in “other” parts of the world was given in that thread. A notable exception was this post by @willkmis, which was so thoughtful I’m inclined to take as gospel.
In Thailand—and from what I’ve seen across Southeast Asia—tracktype is mostly used as an indicator of how well-maintained or frequently used a track is. Surface firmness is largely irrelevant, since most tracks share similar materials (ground/clay/dirt) and proper compacted/gravel pistes are rare. What really changes with low usage or poor maintenance is the encroachment of grass and vegetation. In some cases, tracks become used almost exclusively by motorcycles and end up looking like singletrack paths, even though the underlying roadbed is still wide enough for some four-wheel vehicles to pass, which adds another layer of complexity.
In Brazil, our unpaved public roads frequently deteriorate due to heavy rains and insufficient maintenance. Over 10 years ago, I was advised to use tracktype on non-highway=track ways to represent the firmness of the material and therefore its propensity for degradation by rain or of getting stuck there after rain. After a while, it became clear that tracktype closely matches surface (it is either grade1 for any paved surface type, grade2 for compacted or gravel which are quite rare in Brazil, grade3 for the common dirt, or grade5 for grass, sand or mud), so mapping them together is almost always redundant. Additionally, this practice never caught on and, obviously, tracktype was never widely supported by most tools on non-highway=track ways. So, today, I focus on mapping surface (and occasionally smoothness, flood_prone and the controversial maxspeed:practical only where absolutely needed) to guide routing systems towards the best unpaved roads (when necessary) and avoid the worst. I almost never survey highway=track directly (too busy fixing the mapping of main ways), so I almost never use tracktype on them either. If I were to do it, I think I would often wonder how two mappers working independently would reach a consensus on its intermediary values. In particular, grade4 seems to be the hardest to assess accurately.
@ftrebien Can you point at discussions leading to your change of documentation? - You did introduce “firmness” and removed “rough classification” back then.
I skimmed the whole thread. I learned: Talking heads back then and now do not bother to look at what mappers map.
At least now I understand why the gravel road that @willkmis shared a photo of in another topic is tagged surface=dirt, because in US parlance it is called a dirt road.
I do not see broad community consensus regarding your edits to the documentation.
I think the documentation should stop pretending that tracktype is anything more than a rough classification. Most topics here lately only show confusion when it comes to “firmness” as a single objective. And reading what people say here, it certainly is not what it is used for in a universally unambiguous way. In the ML thread somebody coined “constructedness”, I understand somewhat, but not sure this is a proper English term.
Also the “maintainedness” aspect came late to there. The openstreetmap editors I know, JOSM and iD, both simply call it “type”. All there is to know is in the values.
And the following is only very slightly partially true as the wiki is both descriptive and prescriptive (though there are those who think it should be either one thing or the other thing):
How it really works:
If a tag arose from a proposal, the wiki will initially endorse the terms of the proposal, not the practices (some) mappers decided to follow.
If a tag arose organically, the wiki will ideally endorse sensible uses of it, regardless of whether some mappers are confused about it.
Here is the original text of tracktype proposal, which was written 6 years before I joined OSM. You are invited to do your own research on this topic and improve the wiki based on cross-country practices and sensible, consensus-based recommendations. Of course, we can´t expect this can be done reliably just by skimming through discussions; a more thorough effort is needed to achieve this ambitious goal.
Thank you for patronizing me. But let us not get into nitpicking. The characterisation of tracktype as a “rough classification” received a lot of approval in the discourse here. I think it should be reinstated into the documentation and even promoted there to the first place. That is what it is used for, no matter whether some think that the documentation should endorse something else instead.
I’d like to see examples of that. I would guess that reliably tagging tracktype (whatever that means to the particular mapper) would require survey, and then you might as well tag surface and esp. smoothness to give map users a more definite answer to their question “Can I use this track?”
I’ve noticed that many mappers rely on coarse tags like tracktype instead of adding verifiable detail such as surface or smoothness. The same happens when access tags are misused to express whether a way is practical for a vehicle (e.g. highway=path + bicycle=yes/no). For some, this kind of shortcut may be better than nothing, but it becomes a problem once detailed information is needed and those simplified tags turn out to be ambiguous or misleading.
One can see a track from satellite images and “guess” it is unpaved, perhaps with some grass, see that it is single or double-tracked, but to verify aspects like firmness or maintenance, a direct survey is necessary, and in this case all properties (surface, smoothness, firmness, apparent maintenance) should be accessible (although perhaps not entirely clear even then).
Now if we change “when they are unable to record” to “when it is harder to record”, this opens up the possibility of mapping tracktype as way to map the state of those ways more quickly (may save a few mapping seconds per way). But is it really what the map user wants? I mean, it would be ok-ish, but usually the more details, the better, in particular details that deviate from general expectations (eg. some unexpected degree of firmness for a particular value of surface).
This is how it is done where I live. traclktype is mapped nearly on every track, but surface and smoothness are missing.
And yes, this is what I as map user want(ed). It’s what I see on the maps I use and have used. There was no indication of surface or smoothness but that changed in bikerouter.de (and Komoot) in the last years.
As mapper I’m adding missing surface and smoothness values, width and I would add firmness too.
If this was day 1 in OpenStreetMap, I would say that it would save tons of mapper time if:
The mapper was primarily concerned about mapping surface=* and adding smoothness=* and tracktype=* (which I would have named firmness=*) only where they deviate from what’s generally expected from the selected surface value
The tools supported surface for highway=track based on their default firmness, also supporting modifying the assessment of firmness when tracktype is also present
Such that:
highway=track + surface=dirt/unpaved would be rendered as tracktype=grade3 is rendered today regardless of having tracktype mapped, but could be rendered as anything else if tracktype=* was present with another value
highway=track + surface=asphalt would be rendered as tracktype=grade1 is rendered today (it would rarely be anything other than that)
How tools (particularly renderers) have been (mis)using these tags (due to tradition) may be the root cause of disagreements in the current discussions.
I am still wondering, why you want(ed) to make tracktype map firmness only. That way you will start with a dataset that nobody sanely can rely on. I guess meanwhile I got it why bradrh did applaud you: Because tracktype conveys too much more than that. May I remind you: Rewording the documentation into something that meets high standards will not heal “wrong” mappings, and there are lots of them.
That is why I propose to make the documentation match the mappings, and BTW also what is coded into the editors. And if the result is, that tracktype is a “bad” tag, the usage of which should not be “endorsed” - so be it!
PS: From what I have read here in the forum, the most promising post to help this tag to not fail as “a simple one-dimensional classification system for a complex multi-dimensional reality but are either not able or not willing to actually think this through into a consistent and practically verifiable definition.@imagico” I read here - Confused with tracktype usage - #7 by Map_HeRo - I can follow along, can you? BTW: This also does not give you a handy click bait quote that summarises the substance matter in a single readily comestible term. That still remains a job for the values.
Problem is, tracktype also applies to undeveloped/natural/spontaneous tracks, which range from very firm (rock) to very loose (sand).
Verifiability. Although the wiki does not restrict it to firmness, only places the primary focus on it.
Are there? It would be nice to see some statistics to help move the debate forward: how many (approximate %) would have been tagged differently (and how differently - a slight change or a big change?) if the definition was only about state of development (or maintenance, which are different things) vs only about firmness.
It would be nice indeed, but I don’t see how one could realistically produce it, short of inspecting thousands of track roads worldwide in a short time span.
But let’s assume that Santa Claus would hear out our appeal, engage his sleigh and reindeer for a worldwide survey, and bring us back the results. They would read, for example:
68% of tracktypes are correctly mapped according to definition X,
32% of tracktypes need adjustment, of which 27% for one grade up or down, and the remaining 5% for two or more.
Then, what could we plausibly make of that? Do those discrepancies stem from
different assessments made by different mappers at the time,
deterioration or improvement works that happened since the roads were mapped,
changed Wiki definition? [IMO least likely]
Instead, I think that the best course of action would be to expand the Wiki so as to cover all subtleties, discreetly suggest that the tag rates overall quality, not just surface firmness, and leave it at that.