From OpenStreetMap Wiki
Jump to navigation Jump to search

Ben's thought

Hi, I’ve just read this entire page and it’s taken some time. I scribbled down notes as I went and I’ve compacted it down, but it’s still quite long. I proposed tracktype a couple of years ago, and did so after mapping tracks for some period before hand. It does the job, I can defiantly still say that, but it doesn’t do the same job as has been proposed here. Here’s my opinion on some points made above.

  • ‘Tracktype never getting anywhere’ is incorrect. This irritated me a bit just because of who said it, and the certainty in opinion not resulting from experience in its use.
  • If something is impassable how on earth does it get surveyed?
  • Being pedantically specific when making tags is just unpractical in reality. I try to map with as much detail as I can, but I really wouldn’t be able to map with some of the proposals here, and I believe I could count the people who would on one hand. When people suggest wheel size for example...the amount of work required there...uurrr
  • Voting – why is it that everything seems to be about votes now? This is becoming more and more so with OSM, and it really shouldn’t. Discussion should get us places.
  • Tracktype being replaced by smoothness – Can’t agree to this. There not the same.

Tracktype – Average surface type and prominence on the landscape Smoothness – What means of transport it’s suitable for.

Tracktype also is advantages for aesthetic reason, smoothness for route planning.

  • Tracktype being unclear. Grade 1 – solid, Grade 2 – Broken hardcore, Grade 3 – Primarily broken hardcore but sand/soil/grass also being significant, Grade 4, Mostly sand/soil/grass, with small amounts of hardcore. Grade 5 – No hardcore.

The complication is that highly compacted materials tend to become grade3. Compacted broken hardcore breaks down, and compacted sand and soil becomes harder. This is the only exception I have had to think about.

  • Grades changing with snow: - Incorrect. Grade 4 in winter would be mud/stones, and grass/stones in summer..still grade 4. If snow falls on surface= tracktype= or smoothness= it has the same effect..the tracks the same, but it has snow on!
  • Changes over time and with weather - The lower the grade in tracktype the greater it would be effected by weather conditions.
  • "tracktype=* gradeX are not much intuitive," I agree. In the original proposal I asked people to come up with intuitive wording. I merely proposed the need for that tag. No ideas were put forward, and it remains as that to date. Please improve it…I couldn’t care less if it’s accepted/rejected or on/off the suggested tagging scheme…lets improve what needs improving.
  • Hawke: "I think its values were extremely poorly thought out." (in reference to tracktype). This just isn't so. I was using it for about 6 months before proposing it..It had around 400 hours or riding/walking and note taking before proposing in which it was tested. Therefore it wasn't slapped together. I don't think it covers everything. In fact I see need for various things stated on this page, but it says a lot about a track in a single tag.
  • Tracktype5 being impassable is incorrect. It’s not talking about how passable it is, or it’s smoothness (used in this topics sense).
  • "you'll find grade1 is duplicate of highway=unclassified " incorrect. Highway=unclassified is a road..part of the highway system, and graded as a minor/unclassified/lesser road. A grade1 track is a strip of cement/tamack…it may have a footpath going over may be a private driveway. Both the assumptions that can be made with highway=unclassified such as it being public, and maintained by the country/council etc. can not be made with tracktype.
  • tracktype=* is bad just because it only covers highway=track and not other highway=* tags. " Trueish, but also a point missed. The tracktype is specific to tracks. It was done so because I saw a need to split political and physical. A footway (the right to walk there) doesn’t mean it's a path. Therefore highway=footway, tracktype=x...if tracktype didn't state it was a track then it would have to be highway=footway, highway=track, tracktype=x. Impossible.

I can see the objections to this, and I shall just say…highway:byway is physical? Highway=bridge used to exist as well…and that was separated off for this reason, it’s stating the physical. Highway isn’t just physical, it’s the right to go there.

I prefer this tag to bike=yes car=yes foot=yes also. This tagging method is really long winded, and it’s never really been cleared up if it means ‘can go’ or ‘may go’. For ‘can go’ this would combine potentially infinite yes tags into 1 tag, so I defiantly see an advantage there. Highway=covers ‘may go’.

Personally I think it’s very important that common tags aren’t either too long, or all the values aren’t very similar if they are long, and if neither of them, then the difference is minor and auto-complete plus quick tweak sorts it. This means there quick to type and quick for the auto-complete in JOSM to sort.

In conclusion to the page: I think that there are 4 different points that are being spoken about here. There is the:

Smoothness…which actually covers 2 things. Hazards on the route, and the suitability of the route in general which would effect transport choice. There is the surface material and/or surface description and prominence on the landscape the track has. Then there is the actual right to rome, or suggested ‘route’ that goes that way also, which may be a footway.

I therefore think hazards need separating, and smoothness should be a more specific tag to add with tracktype. I think the potential of smoothness isn’t for rendering but route planning, where as tracktype is primarily rendering and using tracks as a landmark.

Smoothness would then state the ‘weakest’ means of travel that could use the route, and all the above in the list could be assumed also. Width or track/path would obviously filter off some as well.

For the amount of values under smoothness I would suggest 8. (as thats how many I can think up) I don’t think the amount of variables is the complication here.

Also, the greatest factors that I think would effect skill levels would be how steep a route is, how straight/bendy and hazards such as boulders. These could be extracted from hazards, and the other map data, so I don’t think need to be considered here.

Finally the issue of the conditions changing. I think a lot of averaging has to be done, and the worst state of the track stated as a hazard (e.g. flooded) but I have found for tracks, and also in the case of rivers (widths) that if I ignore data that is gathered when it’s a drought, or after large amounts of rain, then the variation left over is minimal and is going to be correct a high percentage of the time.

In general I’m in agreemenet with where this has lead, but that’s pretty much everything I’d like to be considered here. As long as it's kept as simple as possible then it may be plausable, but I have a bad fealing that this tag may involve a large increase in surveying time. I’m done! Ben 21:42, 7 December 2008 (UTC)

Doh ! This is a long thought !

I've been waiting long for your answer but thought you were not anymore participating to the wiki. I'll try to make comments not inside yours, but step by step.

human concerns

Maybe we haven't been "keen" on tracktype, this wasn't so clever, and this was not against anyone. Maybe tracktype came at a time where it was needed, and we have just considered some of it's defects (not completly as gentlemen maybe).

your bold words

  • never got anywhere The sentence you spotted is obviously false, since tracktype is used a lot.
  • impassable : this is a rarely used case, it's only use is for variation see : Talk:Proposed_features/Smoothness#early approval and last unilateral modification
  • Being pedantically specific when making tags is just unpractical in reality : there I can only agree, but since at least a hand of people can, I don't see the need in forbiding them to.
  • Voting : because ONE user went on an edit war while the feature was approved, and all this have turn and merge with the approval process being bad.

Sletuffe 11:04, 8 December 2008 (UTC)

now back to tracktype

Much concern to me is : what we want to know about a track

I'm still not changing my mind, I don't like tracktype (not because of the name, or gradeX or else) because it doesn't answer the questions I'm concerned with :

  • what is the track made of
  • what is the usability of the track

I have tried using it, but back later on the track I had mapped in the past, I wasn't able to use the grade4 I have added to answer the two above questions. I then found it rather useless and stopped using it. (I just said "Me", if someone is able to get an information from tracktype, I have no problem to keep it and let people use it)

Back to the inside of tracktype

  • grade1

I have a problem with it, in my mind, a track is defined by the fact it is unsurfaced, how could garde1 exists then ? You suggest grade1 is defined by the fact it is not part of the main road network, then this was no mentionned in the proposal ! the description is rather poor then, it says "paved track or heavily compacted hardcore." Why not tag an "off network" but paved a highway=service ? So I feel people will just look at the photo and think "ok it's a paved road but badly maintained"

  • grade2

the grade1 says "compacted hardcore" while grade2 says "densely packed dirt" what is the difference ? maybe I don't understand the "compacted hardcore" what could that be ?

  • grade4

it says : "with plant growth in between." is that needed ? is the same thing without grass a grade3 ?

I could continue, but for me words like "a mixture of" VS "mostly, but some" and "Quite compact" VS "densely packed" are not clear enough to make most distinction.

So I have to assume this is a subjective scale of which I will get the "general look of the track" but I won't be able to assume any information from it, beside : "what does it looks like" (I cannot for example define if grade1 will be paved or unpaved, so I need to add a surface=unpaved/paved tag )

Deprecation or not

Since surface/tracktype/smoothness have a different goal, maybe there is no need in the end to deprecate anything. Users having the choice of tags will tag what they find usefull, and time will tell if some appears to be useless.


You propose that, like it is today, tracktype should be rendered while smoothness is not. This entirely depends of the purpose of the map we are displaying. I am creating an hiking map, and for me, it's clear that I want the information of usability, general "look" of the track is secondary. Even surface is secondary, in the end I don't really care if there is gravel, dirt or sand, what I want is the usage I can have with.

steepness issue

I don't agree with you there. For sure a incline=* could improve thing and give even more information, but, like you mentionned earlier, using more and more tags will defeat the purpose of simplicity. When I am mapping smoothness, I make resonnable guess has of the steepneess, hazards, and surface to determine passability. Tagging an incline is very hard, you need measure of you GPS tracks and good supposition that not every one will do the effort for.

example : Suppose I went on an hiking week end, I went to a place by car to start with, and back to home I probably won't have had the time or courage to measure the incline. If I did, then perfect, I will add this information, but if I didn't should I tag just :

  • highway=track ?

Too bad, I have a crucial information I wanted to tag : I went there with my passenger car This is usefull in it's own. But that's a implication of a tracktype, a surface, a max incline, etc...

Probably what I want is smoothness being a duplicate of the sum of other tags, but that respect the KIFS Sletuffe 13:58, 8 December 2008 (UTC)


Hi Sletufee. I do seldom participate in these wiki conversations anymore becuase for a time there seemed to be no progress and pointless things were discussed, but there seems to be comments on the smoothness page that contribute to progressing forward with this, which is defiantly what I want to see happen for OSM.

  • 1. being pedantic
I agree – If they wish to map that way they go nuts, I really dought it would suffice though regardless of it's complexities because the average editor just wouldn't be willing to work in this way.
  • 2:a. what is the track made of
This one is covered roughly. It’s not specific. Surface= is for that. There are 2 issues with surface=. 1, a surface is a often a mixture of things and not really taggable if it’s not something like cobbles. 2, it’s very time consuming to map this way, and I haven't yet seen this tag been used to cover areas as extensively as tracktype is used for.
  • 2:b.what is the usability of the track
I agree here also, this is not the purpose of tracktype, and this is why I do see a need for smoothness (or whatever the ‘key’ is agreed as). Therefore this isn’t really a criticism to tracktype, as it’s not it’s purpose, just as tracktype doesn’t tell us the maxspeed.
  • 3. how could garde1 exists then?
This is basically the same as the question I answered above ‘"you'll find grade1 is duplicate of highway=unclassified "’ I proposed tracktype as a means of tagging a linear way by it’s physically characteristics. This is probably less clear to those not tagging in places such as the Uk, where there is the right to rome, and the land is open for all, but in the UK a footway/bridleway/byway is the ‘right’ to go somewhere. A track is a physical thing on the landscape. Where as in most countries a track on the land is just a track in terms of use, in the UK (excluding Scotland for the most part) a track on the land may be a footpath in it’s legal use. Therefore tracktype is a way of mapping all the tracks that are off the highway, which are private by default (UK). So highway=byway, tracktype=grade1 is different to highway=unclassified. Physically the same, but still different.
Grade 2: The descriptions on the tagging page aren’t that simple really. A grade 2 is gravel or very compacted material such as crushed gravel soil/sand etc.…it’s the next best thing to a tarmac/cement track.
Grade 4, a grass/soil track with a low percentage of hardcore (gravel). ‘plant growth in between doesn’t mean too much, and I don’t consider that in mapping.
I may have some photos of examples, where a road appears to continue, but a sign sits half way along saying 'bridleway' for example. The tarmac continues, but where as before it was a road when a car could drive, now only horses/bikes/walkers could go unless they have specific permission to go down there. It's also the point at which data wouldn't be required for people just seeking road maps. At this point the highway=unclassified is demoted to a footway, but the trackype=grade1 which is assumed default with an unclassified road now requires stating as it is by no means default for a footway. I'm never completely clear on it, but I assume this does exist in other countries (Again such as Scotland) when walking across someone's garden, rather than just general 'land'. (curtilige/curtilidge)
  • 4. Deprecation:
I agree. I think they all have there usage. Tracktype does not cover what you are proposing that smoothness would, and/or what surface does and vice versa.
  • 5. Rendering:
I’m referring to the generic maps that currently exist. Obviously anyone can render what they want.
  • 6. Steepness:
Hmm, I see your point, and I agree as the topographical data we have probably won’t be good enough to really warn you of everything.
I’m not really clear on what your asking for the last part. Highway=track really isn’t necessary ever. I have never needed to use it. If tracktype= exists then it must be a track.
  • 7. Also...
To add to this, I think one issue that exist here is that tracktype says a couple of things, as well as addressing an issue that exists within the UK. That doesn’t mean it’s a UKcentric tag, or in anyway impractical in other places such as Scotland, but without this method it is impractical when mapping in England. Track type covers 2 things anyway, it says
1) It is a track…physically. (track: a route laid for or made by a 4 wheeled vehicle)
2) A rough description of the surface
There are also paths that are not highways in England, so really there needs to be a tag for paths. It may be that to simplify things and make it clear to people that instead of adding 1 tag we add 2. More work; but maybe neater in tags.
So state the 'physical' : track/path/(other? (maybe wider paths suitable for horses/bikes more)
Then state surface or tracktype, which would in effect make tracktype purely a simplified less specific surface=, which I would still use for it’s simplicity in mapping and rendering.
In a country such as Scotland, which I assume to be similar to many other places in this respect, you can rome anywhere. Therefore a track could just be 'physical' : track, and access=public could be assumed. (This is assuming that access rights on ways can be varied by country in the route planners).