More like this:
By the way according to tracktype=firmness, it might be grade3, because there is only a very thin layer of soil atop quite a firm surface. I will put grade5 instead. Of course, there should be some barrier=obstacle and impassable smoothness, but I in general do not tag those. I assume there is access=no or something, probably, have not got around to mapping this yet. On the picture, you see an overgrown road, that is very apparent when you walk over it.
It’s more like grade7… 
I think it fits the description of grade4: “An unpaved track prominently with soil/sand/grass, but with some hard or compacted materials mixed in.” It would probably also be tagged as surface=grass too.
Why not? Knowing which ways are physically impassable is very important for map users.
This tag is for a legal restriction. It’s one thing to not have the physical ability to pass, it’s quite another to try not to accidentally trespass. This is especially important for map users, many of which use maps because they are not familiar with the area.
To me, a typical grade5 track consists of mud or loose sand (or, technically, natural loose soils of various compositions). Typically, grass roots stabilize the soil beneath, making the surface firmer for walking and driving; cycling may be difficult due to the resistance of grass blades. Grass, on the other hand, can make driving risky due to reduced grip, so the risk is not of getting stuck due to sinking, but of skidding (this will depend on whether the grass is wet or not, its height, how dense or sparse it is, etc.).
I only get a " The request is blocked." message from there.
When I worded the rephrasing of tracktype I thought of that too, because, I do not drive a tractor :) Nevertheless, I wrote “way” instead of “track”, because you know, in Thailand the key gets used on unclassified as well and I did not want to rule that out. So vehicles had to be accounted for in the language.
Anyways, the change got reverted, this is not 2014 anymore, and also I am a bit burned out by how much time and effort here spent into skunking some tags beyond what they actually are.
Well that’s quite annoying. I’ve edited the post with an upload of the file as a workaround.
You literally changed nearly half the wiki page and marked the tag as “legacy”, so the revert isn’t surprising.
I’m not fond of the current content or the tag either, but it is widely used. A more productive approach is to focus on a few critical open points, get answers, and document any consensus.
I recently restructured the page and added a small Notes and regional interpretation section summarizing unresolved questions from the forums and talk page. It didn’t get reverted because it doesn’t change the tag’s meaning, it just documents open points.
The wiki already mentions usage on non-tracks. The wording could be improved, but I don’t see it as a priority and it may just trigger the “tracktype is only for tracks” crowd again. Just something to keep in mind before making wiki changes.
Trail-focused, but the maintenance and development dimensions map well enough to fill some of the conceptual gaps in tracktype 
This seems like smoothness=impassable + sac_scale=alpine_hiking or higher. Maybe only class 4 and class 5 in this scale would be mapped as highway=track. Does the document really say this, or was it an AI hallucination?
I recommend reading the document yourself, but yes it does mention walls. Based on the photos in the doc it seems it is referring to retaining walls like this:
Or short walls along the edges of a trail like this:
Neither of these block the way or make it impassible.
Regardless, we shouldn’t consider this document directly applicable to tracktype in OSM. I just shared it as example of another multi-faceted classification system that has some similarities.
I doubt you can tell this from a picture. How can you have any idea what is under? Or is that a joke on the tree haing compacted the soil on impact? Anyway, this probably used to be at least grade3 (as in “buses used to go there”, I am told, no idea about the composition of the road, how are we even supposed to know this?), it actually used to be unclassified,I think, but now after decades of course there is a soft layer on top. No idea what is under that, of course, I did not do a geological survey for the sake of OSM. You suggesting grade4 just shows how this narrowminded “tracktype is only firmness” is bonkers. Imagine a consumer that only relies on tracktype. Seeing grade4, they might think "hm, maybe this is ok, let’s drive there. I bet if we had a poll, grade4 would get less than 10 %. I think the rest would be spit between grade5 and “this is not a track at all”.
I in general only tag tags I am an user of (I would assum a lot of users are like that?) - I see them in the map and have a mental model of what they mean. smoothness is imcomprehensible to me. As for obstacle, I did not know it existed until 2 weeks ago but anyway, I think grade5 means “do not plan on driving this just from seeing it in the map”, so my needs are satisfied. Other mappers can add those tags. The same reason I do not tag surface - I do not understand the values and I am not really interested in it anyway, trails are always ground anyway and tracks are deductible from local geography and tracktype.
I know. It is a nature reserve and behind a gate (the gate I will map), so I do not think you are allowed to drive there.
As for the deduction that grass means not grade4 goes agains the literal definitions of the wiki (grass mentioned for grade5, both pictures are grassy roads!). That is totally inconsistent with this “rigid definition is all”, so I am quite confused. Also, I assume you have never been in a grassy swamp?
I didn’t say that in such absolute terms, and you can check my vote in every poll.
That’s what they have been doing for years, for cars, bikes, pedestrians and probably more modes, even for tracktype=grade5, see here:
That’s unusual. The images and descriptions on the wiki suggest grade5 should be drivable at least with the robust vehicles used where highway=track applies. From the wiki:
The tag
highway=trackis used for minor land-access roads that are mostly used for agriculture, forestry, outdoor recreation, and similar activities on open land. These roads are used by four-wheeled (two-track) vehicles and often take the form of two wheel tracks in the ground (also known as a two-track road), but they may also be graded or paved in some regions. These ways are not considered part of the general-purpose road network (which consists ofunclassified,tertiary, and higher). They are mostly not for general motor vehicle traffic, but are used by some subset. Roads used for access to permanent human settlements or facilities should generally not use this tag. … Without these tags, the only thinghighway=trackimplies about vehicle suitability is that the way is wide enough for a typical four-wheeled (two-track) vehicle. While some track roads are passable in an average passenger car, others require high clearance, or an off-road vehicle. Vehicle widths also vary. So adding these supporting tags is recommended if the information is available. … Not all track roads require a rugged vehicle, and not all roads requiring a rugged vehicle are track roads. … Some examples wherehighway=trackis generally not appropriate: a road that is minor but serves as a connection in the general-purpose road network; a minor road with permanent residences along it; a minor road providing primary access to one or several permanent residences (driveway); a minor road providing primary access to a commercial or industrial facility; a trail or path that is not wide enough for a typical four-wheeled motor vehicle
When I mentioned this picture:
I was also hinting that there’s far “softer” (looser) surfaces than what the tracktype article appears to expect. (could it be that by “soft” it means the grass? as in a comfy pillow?). surface=mud contains much more useful information for the way in that picture - and as I’ve been told afterwards, also hazard=quicksand. In this case, the quicksand guarantees this not to be a drivable track (though there are guided horseback tours there), but there are tracks with surface=mud because they stay muddy most of the time, and that’s still far less firm than what the wiki shows for grade5. And, as such (finally my point), surface=* remains capable of representing more useful information than tracktype.
“Trail” is a word with many meanings that are mapped using different highway tags in OSM, so it depends on what you mean.
Some hiking trails are covered in gravel, woodchips or others. Some agricultural tracks are made of asphalt, concrete:lanes, grass or others. Some forestry trails are covered in grass. Of course, ultimately, one could call the surface of a major highway the “ground”, but that’s almost never surface=ground in OSM.
What does “local geography” mean, exactly? Is tracktype set based on topography? Or could it be related to the proximity of bodies of water?
It also has another similarity to tracktype that isn’t visible from the documentation: it has poor user-to-user consistency in ratings. There are a number of trails that are mapped and classified by more than one national forest, usually because the trail runs along the border, and it’s unusual for both forests to agree on the classification of the trail.
I am truly intrigued by using tracktype on paths 
Some may remember, I once RfCed the community in order to create a foot_scale or foot:scale – one that anchored on something similar to tracktype=grade1, but I noticed that the community could not agree on that so I stopped my effort. The linked document was shared back then too by @erutan. Meanwhile I think the community never able to agree on anything?
I parse tracktype just the same - obviously the USDA as well Firmness is nothing but a red herring. Somebody looked at the pictures and the phrases and thought, yes, this is what the key is about and put it on the wiki. Nearly a dozen years later, mappers still map as if they never became aware of that change.
PS: Why nobody here seems to notice that tracks often have different surfaces/smoothnesses/firmnessnes in perpendicular?
Then I do not see how you can classify that picture above as grade4.
Well, that wiki clearly concerns itself with tracks that are used and tend to be more in developed world. From my observations, grade4 and defnitely grade5 is something you need a tacktor or very rugged jeep to drive on. They also often mark tracks that became unused. Still they should not be paths, they do not look like paths at all (I am firmly in ducktagging camp). In OSM terms, when I say trails, I mean path, but in everyday speech I would say about 95 or more percent of trails are ground.
I guess my own definition of tracktype comes from having started to use it in arid regions, where everything is rocky. Both picture and description do not make sense in such settings.
Well, the climate zone and geology, mostly. In temperate regions you get soil, in arid regions, you get rocks. But that sentence was written in the context of surface and assumption that most tracks are not in my experience really constructed from materials imported from elsewhere.
Honestly, it complicates things, narrow paths and 4-wheel tracks have different enough characteristics that a single grading system struggles to serve both. The USFS dimensions (tread continuity, obstacle character, vegetation encroachment, experience spectrum) map well onto paths but poorly onto the surface hardening, drainage, and vehicle-passability concerns that matter for tracks.
I’d personally rather see tracktype scoped explicitly to 4-wheel ways and a separate pathtype proposal handle the rest, that’s where the USFS material would actually fit.
Many forest ways are occasionally used by forestry vehicles or by tractors from the surrounding villages, residents driving to their fields, collecting firewood or moving beehives. Actually, many hiking paths originated that way, as can be evidenced by still visible tracks or ruts. But you’ll see a vehicle there only once in a blue moon. That’s why many tracks have switched tagging to paths and the other way round, and many still have a tracktype despite being tagged as path.
The very simple “good → bad” gradation that tracktype was designed to represent works just as well with narrower as well as wider paths and tracks.
If it doesn’t for you then I suspect your vision of what tracktype represents is far more complicated than mine.
It’s not that. It’s just that the “good → bad” grading system clearly wasn’t simple enough to work as intended, since it effectively ended up being “solid → soft” to the average user in iD. My point is that if the ultimate goal is to get back to a simple quality classification, adding another dimension to the equation will only make it harder to reach consensus.
I’ve spent most of the last 20 years in the Desert and Tropical Forest/Mountain locations running expeds, and I have definitely had to ‘tweak’ the understanding. It’s not the same and alternative photos would be good for that, but I haven’t struggled to assign a grade to a ‘way’ at any point. Also, in relation to other replies, track, path, cycleway etc doesn’t matter to me. It’s really just the surface_type. It’s called Tracktype because as well as being used to define how ‘prominent’ or ‘permanent’ something was on the landscape (or maintenance level), it was (as a secondary intention) taking the need for highway=track away, so that a right of way (path) could be on the track. There are access right tags for that but the main renderers never supported that. Likewise for any explicit tag for that access right category.
There is always seasonal variance, particularly mud with the monsoon, and seasonal ice cover, which may be a grade 2, but be a lateral ice rink at the wrong time, however the physical status/prominence of that route remain the same, and generally a lot of pretty safe assumptions can be made once the grade is known. If i set off down a grade 5 of anything my expectations are low!
It’s similar to when people muddle up difficulty, altitude, height gain and length of routes when trekking. The tracktype is just one element. The fact that it doesn’t consider all factors, doesn’t in itself mean it is incorrect, it’s just there is a need for more. And there were (are?) an array of specific tags for that, if people have the time, and renderers have the capacity to decipher the sum of those parts.
