Tracktype as a measure of surface firmness

[rhhs (Richard)]

Now that this poll has clearly shown that most mappers (90%) consider tracktype a tag that measures surface firmness, we can discuss what to do with it so it is no longer skunked and more useful.

I propose to split the Key into one that describes surface firmness and one that describes state of development. I’m not interested in a key that describes state of development so won’t discuss it here. Please go ahead and start a thread if you are.

I think there is a demand for tagging the firmness of a way’s surface. In discussions about smoothness, bicycle users have often mentioned that there are discrepancies between smoothness as experienced by a car driver and smoothness as experienced by cyclists (and motorcyclists, probably). Most of these discrepancies are due to differences in how a broad-tyred 2-track vehicle (a car) and a narrow-tyred single-track vehicle (a bicycle) behave on soft surfaces. On a soft surface like clay, sand or loose fine gravel, a car wheel will simply flatten out any roughness (and make a rut if the surface is very soft), while a bicycle wheel will sink into it, creating increased resistance and difficulty in steering.

I’m glad to have learnt that most consider tracktype a tag that describes surface firmness, so I propose to formalise this and decide that tracktype describes only surface firmness.

I don’t think that such a decision will have much negative practical consequences. Even those that mapped it as a measure of the degree of development of a track have done useful work, because in most cases a more developed track also has a firmer surface. The biggest difference between the two tagging practices occurs for tracks that are less developed but still have a quite firm surface, so that those that were tagging degree of development would give it a grade4-5 value despite being quite firm. If we tag tracktype purely as a measure of surface firmness, those tracks will be tagged worse than they actually are, and map users will be pleasantly surprised that they are better than expected. The opposite (a track with a soft surface having been mapped as a grade2-3 tracktype) would be bad but seems very unlikely to me to occur.

I think we should also decide upon exactly how to define “firmness”. I think this should not be the actual firmness at the moment of survey, but the expected worse-case firmness in the local climate due to the composition of the surface. A clay surface can be quite hard after long periods of dry weather (and when frozen), but soft as mud after a period of rainy weather. Sand surfaces tend to be softer after dry weather and become firmer when humid. Both surfaces should be tagged with tracktype=grade5 (“soft”). A surface containing equal amounts of soft (clay/sand) and hard (stones of various sizes) materials should be tagged tracktype=grade3, and a surface that is almost exclusively hard material (bedrock, mostly large rocks, or paved) should be tracktype=grade1. A surface of freshly laid loose gravel or fine gravel should be tagged tracktype=grade4-5 because of the effect these surfaces have on bicycle wheels.

We should update the wiki:

• We should spend a few words on the history of the tag, and that it is now a tag that only describes surface firmness. In fact we could consider using the opportunity to rename the tag to firmness=*. This might be a lot of work now, but will avoid future confusion.
• We should make it clear that it can be used on any unpaved surface (paved surfaces are grade1 by default)
• We should replace the photos in the table with better ones that show close-up images of surfaces (not necessarily from tracks).
• We could add a column in the table, or add a separate table, that describes the effects of the surface firmness on various vehicles (heavy trucks, cars, bikes, correlation with tyre width and weight) and pedestrians.

In fact I’ve been considering whether we should move the values of tracktype into the surface space (i.e. tag a surface firmness as surface=grade3). The values describe surfaces after all, so why do we need descriptive surface= values for unpaved surfaces when all that matters is their composition? For paved surfaces, the surface= value has some value for orientation (an asphalt road looks clearly different from a paving stones road, so it helps to recognise roads). But for unpaved surfaces, we have long discussions about what to call a certain surface, so apparently it is not very valuable to recognise a road surface when we can’t even agree on what to call a surface.

For a map user to decide whether a way is usable for him, all he needs to know are its smoothness and sometimes its firmness, and once he knows that, he won’t care if we call an unpaved surface compacted, gravel, dirt or ground… for practical purposes, it’s all the same for him.

[Vinzenz_Mai]

@rhhs

I’m glad to have learnt that most consider tracktype a tag that describes surface firmness, so I propose to formalise this and decide that tracktype describes only surface firmness.

That is a very bad idea. It would be much simpler and better to describe the firmness of a track with the new tag firmness as @ezekielf said:

@ezekielf

But please stop attempting to redefine it. If you want a tag to specifically record only how firm the surface is, firmness=* would seem logical, though many surface values already give a pretty good idea of the firmness (sand vs compacted for example).

[bradrh]

This is a good idea. The wiki already says this, but then goes further and says too much only confusing the issue. The wiki could be cleaned up. As the poll shows, tracktype is about firmness. In hindsight that’s what the tag should have been

[Hungerburg (Asa Hundert)]

@rhhs

we can discuss what to do with it so it is no longer skunked and more useful.

This tag has 10 million uses - Key:tracktype - OpenStreetMap Wiki. The picture is the same for 20 years now - Proposal:Grade1-5 - OpenStreetMap Wiki - It did influence an awful lot of mappings. Obviously people could relate to what it shows? A handful of people discussing its merits will not change its usefulness or even de-skunk it.

@rhhs

I think there is a demand for tagging the firmness of a way’s surface.

The perfect way ahead firmness=* – Looking forward to RfC of that key!

[ezekielf (Zeke Farwell)]

Attempting to re-define a 20 year old key with 10 million uses is futile. Please create a new key instead. The wiki can certainly be improved, though. It should probably say something like like this:

tracktype is a rough classification of track quality. On the high end it represents a high-quality reliable road and on the low end it represents a low-quality, potentially unreliable track that is only faintly recognizable as a road. As a rough classification, the choice of grade is affected by several different factors including how well built the track is, the surface material, and how rough the terrain is. To add more detail, these factors can be tagged explicitly with surface and smoothness.^[1]

---

1. and potentially firmness if that catches on ↩︎

[ezekielf (Zeke Farwell)]

@rhhs

Now that this poll has clearly shown that most mappers (90%) consider tracktype a tag that measures surface firmness

It shows that 35 mappers think this. Meanwhile Taginfo reports that objects with the key tracktype were last edited by over 120,000 different users.

[rhhs (Richard)]

@ezekielf

It shows that 35 mappers think this. Meanwhile Taginfo reports that objects with the key tracktype were last edited by over 120,000 different users.

There’s always the question if poll results on this forum are representative for all users, but if you know a better way to find out, go ahead!
I suspect that the results of the poll here are skewed towards the practice of more active and experienced mappers, and many users of StreetComplete (that’s aimed at inexperienced mappers without OSM knowledge) are not represented here. It would be interesting to know the share of StreetComplete users in the edits on tracktype (they can be recognised by the changeset comment “Specify tracktypes”.

[Matija_Nalis]

TL;DR: You never, ever change a meaning of highly popular OSM tag after prolonged use of it. Ever. Just trust me on this.

---

@rhhs

Now that this poll has clearly shown that most mappers (90%) consider tracktype a tag that measures surface firmness

@rhhs

I’m glad to have learnt that most consider tracktype a tag that describes surface firmness, so I propose to formalise this and decide that tracktype describes only surface firmness.

Note that (given a key usage) it is quite bad percentage, and not a good one as you seem to imply. If that holds with wider mapper community, it means about a million of tags would become incorrect if you were to change definition to only mean firmness.

It is a horrible idea, unless you personally plan to identify, visit and fix those million ways^[1].

@rhhs

we can discuss what to do with it so it is no longer skunked and more useful.

As noted in that other thread, de-skunking is impossible by definition. Adding another redefiniton only makes a tag even more skunked. Please do not do that.

@rhhs

We should update the wiki:

The only change that might make the wiki more useful in that regard, is adding Questioned template, and linking to newly invented tags which better replace it (until such time as tracktype can be deprecated), as noted here:

@Matija_Nalis

Only solution to “skunked tag” is inventing a new tag(s) much better defined from the very start (hopefully through long and insightful proposal process), with plenty of examples both in text and pictures what the certain tag covers (and why) and what it isn’t (and why).

@rhhs

There’s always the question if poll results on this forum are representative for all users, but if you know a better way to find out, go ahead!

You’re missing the point, which is: You never, ever change a meaning of highly popular tag after prolonged use of it. Ever. Not even if 100% of the mappers in a forum poll said it was a best idea since the sliced bread.

You can use poll for finding out what would be a best way to invent a new tag. I’d suggest going with proposal process for that, lest you repeat the problems of existing tracktype.

See e.g. wiki article about problems with deprecating tags and Everything is more complicated then expected – and note that changing a meaning of a tag includes all of those problems and then a million extra problems.

---

1. and if you do plan to do that, it is likely even worse idea, for reasons that should hopefully be obvious for most given a length of human lifespan ↩︎

[rhhs (Richard)]

Maybe a way to proceed with the diverging tagging practices of tracktype that cause it to be skunked, while keeping as much as possible of the value of past work could be:

1. Automatically add tags firmness=gradex and development=gradex (x=1-5) to all ways with tracktype=gradex (except maybe only add firmness=gradex to those that were set with changeset comment “Specify tracktypes”; these were set using StreetComplete, answering the question “What’s the surface firmness of this track?”)
2. All mappers who ever tagged tracktype to review their edits, and delete either firmness=gradex if they were tagging development or development=gradex if they were tagging firmness on the ways they tagged. It would be nice to have an easy to use tool for this that lets even inexperienced mappers find and review their edits one by one. Not all mappers will do this (many will no longer be active), but I’m wondering if that is a problem (let’s think about this and discuss).
3. Notify data consumers of these changes, so they can decide on how to use the new tags
4. After one year (or more?), automatically delete all tracktype tags, and officially deprecate the Key.

[rhhs (Richard)]

@Matija_Nalis

You’re missing the point, which is: You never, ever change a meaning of highly popular tag after prolonged use of it. Ever.

I think the point here is that what appears to be a large majority of mappers thinks it’s not a change of the meaning.

[Matija_Nalis]

@rhhs

I think the point here is that what appears to be a large majority of mappers thinks it’s not a change of the meaning

No. As others pointed out several times too, that poll just shows that few dozen mappers thinks it align with their mapping practices of primarily firmness (myself included). But mappers that actually use that tag – there is more than hundred thousand of them.

So all we know is that less then 0.03% of mappers using that tag think it means “firmness” – which if several orders of magnitude less than your claim of 90% majority. You simply cannot use mathematical induction^* here to make a leap of faith that the percentage of forum readers is the same as the percentage of actual mappers.

That would defy the very bases of scientific method. There are zillion differences between those two groups of people, so you cannot draw a conclusion that they are behaving the same any more than you could assume that whatever works for cats also works for dogs, since they’re both mammals.

---

^* That reminds me of a joke:

Mathematician, physicist, engineer, and programmer are asked to prove that all odd numbers are prime numbers:

• mathematician - 1 is prime, 3 is prime, 5 is prime, 7 is prime, and therefore, using mathematical induction, we can see that all odd numbers are prime
• physicist - 1 is prime, 3 is prime, 5 is prime, 7 is prime, 9 is experimental error, 11 is prime, 13 is prime, etc. Evidently, all odd numbers are prime
• engineer - 1 is prime, 3 is prime, 5 is prime, 7 is prime, 11 is prime, 13 is prime, 17 is a prime etc. So, all odd numbers are prime
• programmer - 1 is prime, 3 is prime, 5 is prime, 7 is prime, 7 is prime, 7 is prime, 7 is prime…

[Shaun_das_Schaf]

As already discussed in Confused with tracktype usage, surface firmness is certainly part of the decision of which tracktype to use, but it is definitely not the only factor. It may not even be the most important one.

[Matija_Nalis]

@rhhs

1. Automatically add tags firmness=gradex and development=gradex (x=1-5) to all ways with tracktype=gradex

Bad idea. It would automatically skunk the firmness and development tags at the very start of their life too, as on most (if not all) ways at least one of them would be tagged fully incorrectly.

@rhhs

2. All mappers who ever tagged tracktype to review their edits

Frankly, this is very naive at best. Who is going to make them? Even if you offered $100 to each of them (from which fund?) to review their changes, it’s not going to happen to all of them (or likely not even vast majority of them).
Who knows how many of them do not even map OSM anymore or read the messages (if users who fail to respond to followup Note questions or changeset discussion are any indication, there would be waaay too much of them).

You cannot simply shovel those problems under the rug with “I’m wondering if that is a problem”. Might as well wonder if Sun is mostly made of hydrogen. Maybe not, but the current body of evidence seems to suggest otherwise.

And then, additionally, unless their current mapping happened to perfectly align with one of the new tags, and did not align with other at all (which is questionable to which percentage of them it would happen), they wouldn’t need just to review, but actually to resurvey all of them, which is even less probable to happen.

But anyway, since point 1. is no-go, this is not even worth discussion as this point could never be reached. You simply may NOT add knowingly incorrect data to the map via automated edit, and then design a MapRoulette-alike challenge so people “can go and fix your mess”. Please read Automated Edits CoC.

---

What one can do is:

@Matija_Nalis

Only solution to “skunked tag” is inventing a new tag(s) much better defined from the very start (hopefully through long and insightful proposal process), with plenty of examples both in text and pictures what the certain tag covers (and why) and what it isn’t (and why).

… and then you use those new tags, and promote it to others that they should use it, propose it to editing software (and/or write pull requests) and other data consumers so they should support them in addition to tracktype, and then work on promoting new tags some more.

And only then, when new tags are so popular^[1] that basically all of tracktype is either fully deleted or dual-tagged with new tag, have you completed your task. This is the way.

---

1. because the proposal was so good that it fixed all the problems and ambiguities and have not introduced any new ones so people went wild to use them ↩︎

[Hungerburg (Asa Hundert)]

@Matija_Nalis

Bad idea. It would automatically skunk the firmness and development tags at the very start

At the time the notion of it mapping, among other things, the firmness of the surface, got introduced to the documentation, the tag already had 3 million uses.

A sister tag to smoothness=* like firmness=* should not start with a base supposedly grown on bad advise but instead from blank slate and go full proposal process.

BTW: Did I say, I do not find the tag skunked? The old photos still stand out as picturing something that makes me enjoy openstreetmap - being out and wandering e.g. the countryside. The offer granted, here some bins to roughly represent what is different, still seems worth the little effort.

[Duja]

@Hungerburg

BTW: Did I say, I do not find the tag skunked? The old photos still stand out as picturing something that makes me enjoy openstreetmap - being out and wandering e.g. the countryside. The offer granted, here some bins to roughly represent what is different, still seems worth the little effort.

The thing is, development of a track road highly corresponds with its firmness. For most use cases, track roads exist in agricultural and forested areas, which tend to have rich, thick, rather soft top-soil layer. To improve firmness and accessibility, such roads need maintenance, and occassional improvements in form of a layer of aggregate artificially added. The original 2007 discussion obviously apparently had (only) those use cases in mind. From today’s viewpoint, we may regard it “parochial”, but still, it produced a widely used – and quite useful – classification.

As with any key, there are edge cases where “development” does not correlate with firmness…

File:Lava fields, Krafla - Iceland - panoramio.jpg

..the “road” to the right of the photo is apparently “solid”, i.e. very firm, but I doubt you would like to drive your car over that. On the other hand of the spectrum, in polar areas there are roads that are quite useful and maintained in winter, but on rainy or summer seasons they just turn into mudbaths. I would argue that those cases simply fall out of scope of tracktype and should be better described by other tags.

[Hungerburg (Asa Hundert)]

@Duja

I would argue that those cases simply fall out of scope of tracktype and should be better described by other tags.

Personally, I’d leave the choice to the local community. After all, this tag renders in OSM Carto and probably lots of other consumers use it (which may explain a bit why all the talk?) Still, people nevertheless encouraged to create new globally applicable and universally unequivocal terms to spell out the minutes!

[rhhs (Richard)]

@Matija_Nalis

No. As others pointed out several times too, that poll just shows that few dozen mappers thinks it align with their mapping practices of primarily firmness (myself included). But mappers that actually use that tag – there is more than hundred thousand of them.

So all we know is that less then 0.03% of mappers using that tag think it means “firmness” – which if several orders of magnitude less than your claim of 90% majority. You simply cannot use mathematical induction* here to make a leap of faith that the percentage of forum readers is the same as the percentage of actual mappers.

Are you serious? With that reasoning, we would never be able to take any decisions, because the number of mappers that vote in a poll here or in a formal proposal procedure is always a tiny fraction of all mappers. Unfortunately there is no way afaik to reach the majority of all mappers and ask them for their opinion so we can take decisions by majority of all mappers. The best we can do is to assume that the mappers that vote are a representative selection of all mappers, and discuss and correct for any bias in the sample if we suspect there is one. Otherwise anyone can say that “I don’t like your proposal, and the 100.000 mappers that didn’t vote are on my side, so the proposal is rejected”.

[Mammi71 (One feature, Six mappers and still More ways to map it)]

@Duja

On the other hand of the spectrum, in polar areas there are roads that are quite useful and maintained in winter, but on rainy or summer seasons they just turn into mudbaths. I would argue that those cases simply fall out of scope of tracktype and should be better described by other tags.

Key:winter_road - OpenStreetMap Wiki

[Matija_Nalis]

@rhhs

Are you serious? With that reasoning, we would never be able to take any decisions,

There is ENORMOUS difference between attempting changing the meaning of long establish tag with many millions of uses, and “being able to take any decisions”.

For the former, yes, I’m quite serious. A majority vote of few dozen mappers on the community forum cannot decide by themselves that, say, female=yes tag suddenly applies to all non-male people. (or to change tracktype meaning).

Once the popular tag has been used so much with certain meaning for so long, its meaning might as well be written in stone. It’s amend-averse. That tag core meaning isn’t adjustable anymore. It’s beyond radical revision. It’s chiseled in granite by a stern medieval mason with no sense of whimsy. It has ceased to evolve. It has been ratified, notarized, and stamped “FINAL.”

In short, it’s stiffer than a Norwegian Blue that’s been resting since the Late Jurassic.^[1]

@rhhs

Unfortunately there is no way afaik to reach the majority of all mappers and ask them for their opinion so we can take decisions by majority of all mappers

Which is exactly why I would (again) re-emphasize the enormous problems with trying to redefine POPULAR long lived tags, because it is enormously harder then it looks.

Did you even look at those links that I posted earlier that explain why? If you did, where is your response for each of the problem points and how you propose to approach them? I didn’t see it, at without at least suggestions on how to fix each of those^[2], it is hopeless to even try to considering modifying the meaning of already existing popular tags.

@rhhs

The best we can do is to assume

No, effectively destroying^[3] the work of hundreds of thousands of mappers to make a few dozen mapper feel like they’ve made the place tidier is not a good tradeoff. Please resist the temptation to skunk existing popular tags, and instead concentrate on finding other ways to do better tagging.

IMHO, the best we can do is put up a lot of effort to make a proposal for a new tag(s), which would be much better defined and non-ambiguous, and start using that. And when they become popular enough alternative to be a viable replacement, we can talk about deprecating existing popular tag tracktype in favor of them. Not before that.

@Hungerburg

A sister tag to smoothness=* like firmness=* should not start with a base supposedly grown on bad advise but instead from blank slate and go full proposal process.

Agreed.

@Shaun_das_Schaf

As already discussed in Confused with tracktype usage, surface firmness is certainly part of the decision of which tracktype to use, but it is definitely not the only factor

Agreed. I voted firmness too, because for my use cases it plays the most part (as it does on current wiki). But it by no means even suggests that this is the only meaning and that my vote somehow means I would support the idea that tracktype should be redefined to mean to have the same meaning as hypothetical firmness=* tag (I certainly wouldn’t!)

It is just that poll didn’t offer any better options, and this answer was somewhat closer to my use cases then the other one. And I certainly didn’t expect that poll results would be interpreted against my intentions, and then such gross misinterpretation of results used as an excuse for weaponizing the poll to do something that was not even mentioned in the poll as its purpose (i.e. redefine meaning of tracktype).

Now, if the poll had options:

• tracktype means exclusively surface firmness and absolutely nothing else
• tracktype means mostly firmness, but other things too
• tracktype means exclusively state of development^[4], and nothing else
• tracktype means mostly state of development, but other things too
• Something else (leave comment)
• It’s complicated

I would certainly NOT have voted exclusively firmness in that case ^[5], and I would bet neither would 90% (actually 83% at this moment) of the others. I’ve now reverted my vote there, as a means of vote of no confidence.

@rhhs

large majority of mappers thinks it’s not a change of the meaning.

That’s great to hear! If there is indeed no change in meaning intended, then there there is no need to go wiki-fiddling at all, and we can leave tracktype as it is, right?

---

1. with apologies to Monty Python ↩︎

2. ideally ones that might work, but really any suggestions would be a prerequisite ↩︎

3. i.e. making it more unclear what they tagged ↩︎

4. add definition of that term “state of development” firstly, of course ↩︎

5. but probably mostly firmness or it's complicated ↩︎

[Matija_Nalis]

In fact, let’s try it:

• tracktype means exclusively surface firmness and absolutely nothing else
• tracktype means mostly firmness, but other things too
• tracktype means exclusively state of development^[1], and nothing else
• tracktype means mostly state of development, but other things too
• Something else (leave comment)
• It’s complicated

The poll is not intended as a means of changing the existing definition of the tag on the wiki, but just to get an estimate of misleadingness of previous poll.

---

1. add definition of that term “state of development” firstly, of course ↩︎