I don't see poor definition, neither climate-specific characteristics.

Yes, it can be misused as an importance rating.

Tracktype is used in combination with key:surface and key:trail_visibility. I encounter the combination of tracktype=2 & surface=sand quite frequently and this mis-tagging is already in the JOSM validation rules set.

A proper definition of tracktype <> importance PLUS a proper definition of allowed combinations (and subsequent validation) with surface and trail_visibility should be the preferrable solution

I think it makes sense - we could also use more patterns for admin level boundaries. Paved/unpaved classification rendering would be very useful here.

I don't see it misused as importance either. I agree with removing the many dash patterns as they're confusing, and I'm the one who introduced them originally.

Regarding misuse as a generic importance rating - the only verifiable criterion for tagging tracktype that is either known or documented is 'softness' of the ground - which is a criterion that does not apply at all for much of the earth surface, in particular in dry climate of all kinds. In those areas (and partly also in regions where the softness criterion would be applicable) mappers therefore often apply the tracktype rating as a generic importance rating to indicate the cartographic importance of the track - which will often be based on a local and subjective rating of how much the track seems to be used or how the 'quality' of the track is rated (f.s.v.o. quality that is subjective to the mapper). The source of confusion might have been that i did not clarify that with 'importance' i here meant cartographic importance and not some real world physical/functional importance.

But ultimately it of course does not matter that much - there seems to be broad agreement that rendering the paved/unpaved distinction instead of tracktype would be a desirable change.

I would suggest that the current tracktype=grade1 style could be used for paved tracks, and tracktype=grade4 style be used for unpaved tracks.

@imagico In the wiki definition of tracktype the primary definition is how well maintained it is, secondary is firmness, not softness.

I am more inclined to reduce the amount of tracktypes to only three and thereby reduce the rendering style as a consequence.

I also want to mention the original tracktype proposal, which made usage of very ambiguous and subjective definitions, with characteristics "blending" over with adjacent tracktypes

Therefore (IMHO):

• Tracktype 1: Abolish. A "paved" track contradicts itself as tracks are usually unpaved. Promote this to "road" (unclassified)
• Tracktype 2: Keep. Definition: Constructed with selected materials suitable for intended use. Well-maintained, compacted layer of gravel, etc., Firm surface all year round, Suitable for general traffic. Rendered with uninterrupted line
• Tracktype 2 & 3: Combine. Lightly constructed or created by clearing the trajectory from obstacles and/ or smoothing out holes and bumps. Only locally sourced materials are used. Firmness depends on season or weather. Not always suitable for general traffic. Rendered with dashed line.
• Tracktype 5: Keep. Unimproved, badly to non-maintained. May be unusable for part of the year. Only for special of purpose built vehicles. Rendered with dot-dash line style)

I assume that a propposal like this is partly out of scope of this Github project and may (also) be discussed on the OSM wiki proposal section

@TiliaJ - off-topic here, but i suggest to look at Verifiability.

@pnorman I wonder what was your original idea behind this implementation and why you no longer support it?

That it would provide useful information and be obvious from the rendering what it means. It wasn't - we have too many slightly different dash patterns in use.

I think unifying the rendering of footway, path, cycleway, track and bridleway is a very good idea.
But what about doing it the other way round and making the rendering of footway, cycleway, bridleway more similar to current rendering of track.
Suggestion:
paved similar to type1
compacted and fine_gravel similar to type2 or current paved cycleway
ground similar to type 3 or 4 or current unpaved footway
and informal=yes similar to type5

Having read the tracktype proposal some time ago, indeed I got the impression, that this mirrors the well-known openstreetmap catagorisization of roads into primary, secondary, tertiary. Only six instead of three types. Said that, only two bins for tracks, paved/unpaved look a bit few to me. Grade 5 tracks often are tagged on stuff that is mostly overgrown but still usable for pedestrian passage e.g.

Updated numbers on this:

• tracktype on highway=track: 8.6M
• surface on highway=track: 6.8M
• both: 3.7M

I live in a region where most tracks are sand, on a few it is more compact and cycling is possible, while on most it is really soft and cycling is impossible. It is extremely annoying that I basically have to manually test all the paths in the region to find a good route, as no cartographic material is available that indicates whether a track is cyclable or not. tracktype is basically the only indicator, although in my region the tracktype is often tagged wrongly. An alternative way of determining the quality of a track would be a combination of its surface, smoothness, softness and width, but this would make it much more complex than tracktypes currently are.

In fact, I think it is particularly annoying that there is no equivalent to tracktype for paths. In our local forest there are perfectly smoothly paved cyclepaths where two bicycle trailers can easily overtake each other, and there are informal paths through the bushes where most people would never dare to go. On the map, those all look the same.

I see tracktypes as a way to classify the importance of a track, in the same way as we have trunk, primary, secondary, tertiary and unclassified for roads. The only difference is that OpenStreetMap has a primary focus on car infrastructure, hence car infrastructure is classified using the main tag (highway), while non-car infrastructure is classified using additional tags. From the perspective of a car, the classification of tracks and paths is not important, because all it cares about is that it exists and cannot be used. From the perspective of other means of transportation, the classification of tracks is really important, as in some regions it determines whether a track can be used anywhere between comfortably and not at all.

Both from my perspective as a mapper and as a user, I find the tagging and rendering of at least 5 track types (sometimes I wish there were 6 or 7) very important.

• Tracktype 1: Abolish. A "paved" track contradicts itself as tracks are usually unpaved. Promote this to "road" (unclassified)

I would like to specifically respond to this proposal that I used to live in a wealthy region where many agricultural tracks are very smoothly paved. They are only allowed to be used by agricultural vehicles and by non-motorized traffic. They are not wide enough for two cars to pass each other. They don't have any road numbers and are often privately maintained. There are no road signs indicating directions to nearby villages (except small signs for recreational hiking/cycling/riding routes) or speed limits (such as change of speed limit when entering a village). This all very clearly distinguishes them from unclassified roads. It should also be noted that obviously many of these tracks used to be unpaved at some point, and their role/significance if the road network did not change by paving them, so it would be strange to change their road type on the map when they got paved.

@cdauth - what the best way is to map certain properties of roads/paths is not a subject for this issue tracker.

Your desire to more precisely map practically important properties of roads/paths is understandable. But this is (a) a matter for a different venue and (b) something for which the existing tracktype tag is definitely not a good choice. Suggested reading how to develop new tags:

• https://wiki.openstreetmap.org/wiki/Verifiability
• https://wiki.openstreetmap.org/wiki/User:Joto/How_to_invent_tags
• https://imagico.de/blog/en/how-to-invent-new-tags-in-2021/