I’m personally not a big fan of coded hierarchies. They tend to fail miserably.

Even hierarchies coded in a human language, and used by everyone (like the primary-secondary-tertiary classification) fail quite often. People need to invest a big amount of time in them to understand them. So newbies get them wrong a lot.

However, I agree that highway=path is missing information, it would be better to give names to paths instead of codes.

I’m thinking about stuff like having a path=* key like * path=forest_trail -> A natural trail through a forest, expect it to be muddy, with tree roots sticking out, and likely steep or slippery, unless the details tell otherwise * path=forest_path -> A man-made path through a forest, thus with a good surface, however, it’s still likely to be steep, and leaves can make it slippery. Sometimes roots also grow through the pavement. * path=grass_trail -> A path through or along meadows. Generally flat, but quite often wet * path=park_path -> Flat path, with a good survace (compacted gravel or paved), and a nice scenery * path=city_shortcut -> flat path with a good surface, but in-between buildings or hedges. It’s quiet, but doesn’t have a nice scenery.

Note that this is not a final proposal, it needs someone to actually check which paths exist (both rural and urban), and to see how the classification can be worded easily. Since I live in a flat country, I have only experience with the flat type of paths. Thus for me, the surface is the most important (you can’t go on a surface=dirt path without boots in the winter, it’s just too muddy).

Ugh, and of course, those examples went missing completely. Here is that paragraph again:

I’m thinking about stuff like having a path=* key like

• path=forest_trail -> A natural trail through a forest, expect it to be muddy, with tree roots sticking out, and likely steep or slippery, unless the details tell otherwise

• path=forest_path -> A man-made path through a forest, thus with a good surface, however, it’s still likely to be steep, and leaves can make it slippery. Sometimes roots also grow through the pavement.

• path=grass_trail -> A path through or along meadows. Generally flat, but quite often wet

• path=park_path -> Flat path, with a good survace (compacted gravel or paved), and a nice scenery * path=city_shortcut -> flat path with a good surface, but in-between buildings or hedges. It’s quiet, but doesn’t have a nice scenery.

Really interesting thoughts… I’d like to see that idea developed to see if it would work better than my suggestion. My worry would be that there would have to be too many categories… as a surveyor I find that this kind of tag only works if the categories are really easy to remember. Maybe take a look at the photos I provided in the proposal… can these be categorised sensibly in the way you discuss?

Most people encounter new tags by finding them in use. Someone who finds a way tagged as highway=secondary might be able to guess that there’s a hierarchy of roads and that this is the seconds best. Someone who finds pathtype=grade3 or pathtype=b2 would need to look it up to have any hope of understanding it.

A value on it’s own of grade3 would ask more questions that it answers. Is it from 1 to 10, 1 to 3, 0 to 3, etc.? Is 1 the best or 5 the best?

And I think the alternative proposal is even worse because you have two scales in one. If a1 is the best is the second best a2 or b1?

Peter - very valid thoughts. I learned myself often from looking at what others had done… but my argument is going to be that if there isn’t a simple way to tag (primary, secondary, etc) which corresponds to well recognised ideas (in the UK, A road, B road, etc) then we have to either do nothing (rely on tagging objective facts) or to do our best with some simple categorisation… even if it’s not ideal. It might be valid to argue that adding the word ‘grade’ (‘category’?) to each category makes things clearer, but I couldn’t make this into something sensible for my proposal (‘grade_a2’ isn’t much clearer than ‘‘a2’ and raises lots of other issues.). After all it doesn’t matter if paths are added without a grade… (beginnner mapping).

I sometimes think that we overdo our contributions so allow to say why:

I am 75 years old and am still rambling around collecting data for OSM: It gives me something sensible to do as my wife does the housework and gardening.

Can you remember paper maps (such as Ordnance Survey)?

How were path attributes in these paper maps describe?

Answer: Mostly, not at all!

I assume that I am writing this for ramblers that regularly walk through fields, forests, etc. I do so 2-3 times a week and all I do in addition to picking a route and a restaurant (OSM in its present form fulfills these specifications admirably) is to consider the present weather conditions and the rainfall of the past few days and equip myself accordingly.

So I ask you:

Do you really think that hikers will look for the path attributes that you have suggested when planing their tour?

IMHO the only attribute that is required and then only in geographical areas where it is required is the SAC SCALE.

The wheelchair=no should be a default for paths and therefore not necessary.

NOTE:

I know that we have the combination highway=path / foot=designated / bicycle=designated

which I don’t like.

Why not highway=footway / bicycle=designated / segregated=yes/no?

I quite understand that others contributors will disagree with me, so I ask the following question:

What do the users of OSM data think?

I sometimes think that we overdo our contributions so allow me to say why:

I am 75 years old and am still rambling around collecting data for OSM: It gives me something sensible to do as my wife does the housework and gardening.

Can you remember paper maps (such as Ordnance Survey)?

How were path attributes in these paper maps describe?

Answer: Mostly, not at all!

I assume that I am writing this for ramblers that regularly walk through fields, forests, etc. I do so 2-3 times a week and all I do in addition to picking a route and a restaurant (OSM in its present form fulfills these specifications admirably) is to consider the present weather conditions and the rainfall of the past few days and equip myself accordingly.

So I ask you:

Do you really think that hikers will look for the path attributes that you have suggested when planing their tour?

IMHO the only attribute that is required and then only in geographical areas where it is required is the SAC SCALE.

The wheelchair=no should be a default for paths and therefore not necessary.

NOTE:

I know that we have the combination highway=path / foot=designated / bicycle=designated

which I don’t like.

Why not highway=footway / bicycle=designated / segregated=yes/no?

I quite understand that others contributors will disagree with me, so I ask the following question:

What do the users of OSM data think?

Hi Little Brother

Your comparison to paper maps is really interesting… because actually I think there actually is a hierarchy used on such maps. The choice of what paths to show and what paths not to show is/was made for each particular scale of map, each map according to its purpose. What’s different here is that we’re trying to record the data which others then use to create those maps. So once upon a time a person (or at least a member of a team or organisation) would have surveyed with the final product in mind (a direct connection from surveyor to map maker)… but now we’re doing something very different… there’s very little connection between the surveyor and the map maker. Therefore the surveyor has to record enough information to allow the map maker to be able to choose what to include and what to exclude.

Typical hikers, I agree, won’t look for the data I’m suggesting should be recorded… but those creating the maps (or equivalent services) for the ‘hikers’ potentially will do so… allowing them to produce a map at a high scale showing a city park and all the paths within it suitably rendered according to significance (for example), or a map at a low scale showing a whole city, and only the main paths in that same park.

Hello Rostranimin. Thanks for the reply. I can understand now that such information may be of importance to the mapmaker providing a specific map for a specific purpose, e.g. wheelchair users.

So I am not contra your proposal but I myself would not contribute any of your suggested attributes to the database. That is purely the case because I cannot see any usage for them in my areas of influence.

I wish you luck with your proposed feature.

Thank you :-)

Don’t forget that “well recognised ideas … in the UK” are usually classified as “wat?” by worldwide mappers. I grew up in Scotland so most OSM tagging seems vaguely sensible, but explaining it to my fellow Canadians …? Nah.

That’s a really interesting line of thought Scruss. I tried to write the proposal so that it would work internationally, but I’m painfully aware of having little or no proper awareness of worldwide differences in OSM tagging. I’m not thinking of English speaking nations really when I say that - I’m thinking that very different things could be going on in non-English speaking countries without me being aware. I did ponder looking through other language versions of the wiki documentation armed with Google Translate… but haven’t tried yet. Could it be that elsewhere people have been developing good ways to tackle the issues I’ve raised?

Interesting. One way that the distinction between official paths and shortcuts can be made that I haven’t seen here is by using relations. This is often used outside urban areas to define long distance paths. See www.waymarkedtrails.org for a rendered map, the wiki for tagging. You could easily define the main paths in a park as part of a very local network.

That said, having walked and mapped paths in different parts of the world, I do agree that a simple classification would be useful. The approach to cut properties up in very many different tags appeals to my inner nerd, but it increases the learning curve and makes for a complicated decision process for data users. Your proposal on the other hand looks nice, but it does try to measure many dimensions at once and fit them in a single scale. That means you assume e.g. that an unobvious path will always be hard to walk (or easy?).

That’s why I like Sander’s proposal. It allows for a few very common, intuitive categories that actually have some features that do go together (forests tend to be muddy). It looks easy for data contributors and for data users. It might even cannibalize the specialty sac scale.

Hi Joost Schouppe. (Please forgive the length of this reply and poor grammar- it’s been a long day but I’m interested in your thoughts and want to comment this evening)

Route relations are good… I’ve used them a lot… but they serve a particular purpose I think. I can imagine that for a defined park it might be possible to use this system to say something useful - but I suspect that this starts to do something which is discouraged for relations (using them to categorise things rather than to map something definite which just happens to involve more than one feature). I also struggle to think how things would work once outside a defined area. Nice idea though - feel free to try to convince me/others further here if you like.

The proposal to name a set of path types as Sanderd17 suggests also appeals as I said earlier. But after thinking more I can’t think of a way to make this work in a sensible way. What we think would be a small set of path types I think would quickly mushroom into a very wide range of path types.

Unless someone can come up with around 6-8 categories (no more!!) which capture the kind of hierarchy that I’m aiming at. I don’t think that ideas like ‘forest path’ and ‘urban path’ will work - they raise far too many questions I think. Scanning in my head through the areas I’ve mapped I don’t see that I’d be much wiser… for instance I mapped a large park area on the West of Scotland which is on the edge of a small town but under woodland. You could call all of this urban path or forest path or formal path - some of it informal path or desire line or whatever. Fundamentally none of the sets of words i can think of would allow me to show you what the main paths are and what the faint desire lines are. Give me a bit of paper and I’ll draw you what is the main path, what’s secondary path, and what’s fairly faint desire line… but I can’t give you any sensible set of words to say the same thing (beyond what I just did).

Something like this would appear to work at first… primary path, secondary path, desire line, but it gives absolutely no information about path quality (so a primary path in one place would be a tertiary or secondary path elsewhere). And what happens when what is a primary path in one place becomes a tertiary path as it enters an area where it’s not changed but it’s no longer the primary path. I would like to think that something like ‘formal_prepared_wide_flat_path’ and ‘informal_narrow_steep_path’ would work (or simpler words to mean this), but I really can’t find a way to do what I think is necessary.

Can someone come up with the necessary categories? If the categories are to be simple they ought to be simple to conceive of and fairly easy to agree on.

You’re right that my system is intentionally hierarchical… so as you identify it asks us to accept that the ‘unobvious’ nature of a path takes precedence over everything else (whether it’s walkable or usable by wheelchair becomes irrelevant in the system). I concluded that this is an inevitable feature of creating a hierarchy. I do wonder if the hierarchy in my head is just the one in my head… would a competing system put walkability above obviousness for instance?

I was surveying some paths today and wondering about this proposal while I did so. One thought was whether some kind of numbered score would allow for the creation of a hierarchy. So 2 points for an obvious path, 1 for indistinct, 0 for an invisible one. 2 for a wide path, 1 for a narrow path. 2 for a flat path, 1 for a gentle angle, 0 for a steep angle…. but then I started thinking that my original proposal brings in these things, but also surface quality and smoothness. Now it starts to seem far to complex: 2 points for good tarmac, 1 for a bit rough, and 0 for very rough (BUT we already know from the smoothness tag proposal that this is difficult to define). 2 for a bound surface, 1 for something loose or slippery, 0 for ??

Far too complex. Not memorable.

And one other thought - related to yesterdays comments. I was following woodland paths for a while. They were all woodland paths, but all very different paths. What was the main path and what was the informal hidden path was really obvious - but I can’t currently tag enough data for anyone to easily see this on OSM. What strikes me is that in any one place a change in only one feature makes the difference - in this case the real issue was just something about how well packed the surface material was. Width stayed the same. The basic surface material stayed the same. As things stand if I map these paths nobody will be able to tell that some are key good paths and some little used.

I keep thinking about this as well. Relations are probably my the way to go in most cases. However, within a national park it is ok in my opinion to add all official trails.to relations, and other de facto trails not. In the settings where I map, that should be enough.

I believe the analogy with primary, secondary etc roads is very good indeed. Within most developed countries, this gives a very intuitive result, as road quality (number of lanes, surface) and road function (important, less important) we very clear. However, in countries with a less developed road system, this is not the case. For example, in Peru, all the most important roads are defined as trunk (to indicate they connect major places, following government classification). As there are hardly any real freeways, that means anything from a dirt track to a four lane paved road with separate lanes us tagged as trunk. Cross the border to Ecuador or Chile, and here they have decided to give priority to form. So even if the road is the only way to get to a town, it will still be tertiary if it is unp?aved. Or a long primary road might have a small section tagged as tertiary if that part isn’t paved. I believe the best way would be somewhere in the middle, though that would mean you need a visualization that shows both highway and surface tag.

In the case of paths, in most cases form and function will also tend to go together. A similar scheme with primary, secondary etc. paths might thus also work. I kind of do this already, mapping as footway “official” trails and “path” as trails that just exist. Unfortunately, the definition of these two depends very much on where you are. So Maybe a combination of highway=path & path=secondary might be better. The lack of a clear definition is a strength: within a certain context it will work (a primary path will look very different crossing the Alps from crossing Central Park). And when this info is not enough, extra tags (analogous to surface for roads) are in order. That way, you can have much simpler ways of tagging quality, as they don’t have to measure importance too.

And now without the typos (working on smartphone)

I keep thinking about this as well. Relations are probably not the way to go in most cases. However, within a national park it is ok in my opinion to add all official trails to relations, and other de facto trails not. In the settings where I map, that should be enough in most cases.

I believe the analogy with primary, secondary etc roads is very good indeed. Within most developed countries, this gives a very intuitive result, as road quality (number of lanes, surface) and road function (important, less important) tend to go together. However, in countries with a less developed road system, this is not the case. For example, in Peru, all the most important roads are defined as trunk (to indicate they connect major places, following government classification). As there are hardly any real freeways, that means anything from a 4wd dirt track to a four lane paved road with separate lanes is tagged as trunk. Cross the border to Ecuador or Chile, and here they have decided to give priority to form. So even if the road is the only way to get to a town, it will still be tertiary if it is unpaved. Or a long primary road might have a small section tagged as tertiary if that part isn’t paved. I believe the best way would be somewhere in the middle, though that would mean you need a visualization that shows both highway and surface tag.

In the case of paths, in most cases form and function will also tend to go together. A similar scheme with primary, secondary etc. paths might thus also work. I kind of do this already, mapping as footway “official” trails and “path” as trails that just exist. Unfortunately, the definition of these two depends very much on where you are. Maybe a combination of highway=path & path=secondary might be better. The lack of a clear definition is a strength: within a certain context it will work (a primary path will look very different crossing the Alps from crossing Central Park). And when this info is not enough, extra tags (analogous to surface for roads) are in order. That way, you can have much simpler ways of tagging quality, as they don’t have to measure importance too.

Joost Schouppe - thanks - I’m glad to know that other people are continuing to think about this. While you’ll be able to tell that I think I may have come up with a workable system (as proposed) I’m actually less worried about the detail and more about making sure that this gets sorted in a workable way one way or another. So long as people start worrying about this and testing ideas we ought to be able to sort it.

There’s been one key thought knocking about in the front of my mind for the last week; we get away without having to worry about a categorisation of roads for one reason and one reason only. Someone else has done the categorisation for us.

There are all sorts of real issues with categorisation of roads, but the whole reason that any(many) of our maps make sense is because they are categorised. The categorisation always has flaws, and is always based to some degree on relatively arbitrary decisions… but we get away without having to worry about these arbitrary decisions because some official body has done this.

So we get to map lovely comfortable ‘facts’ and to pretend that opinions don’t belong on OSM.

(I should say that I’ve come across a limited amount of discussion about the issues with road classification - but 99% of people seem quite happy that road classifications are used as a way to understand/draw maps and to influence routing algorithms.)

I was out again mapping recently - and once again found myself being absolutely forced to map my opinions rather than facts. Some paths had to be missed off the data because adding them would lead to confusion. In an effort to support clarity in mapping I had to decide what deserved to be mapped and what not… an arbitrary decision at the highest level - all because there is no way to categorise.

I now have a favourite path in mind as proof that we can’t just map facts and leave categorisation to those rendering maps. On my local hills there’s a path… it’s about 8 metres wide at one point, is usually about 4 metres, and occasionally narrows to 1-2 metre. This happens over about 800m of path length. It’s surface is primarily grass in terms of proportion of surface material. For much of its length what actually defines its use and appearance is that short but significant sections are very muddy. The muddy sections dry up for some short but relatively significant (but variable) periods in summer. The path is clearly not used ever for vehicles, but in other respects could be called a track (it’s certainly wide enough). Early on the path is consistently grass, wide, and because the grass is firm it’s not particularly obvious (being defined really by a line of new trees). Later the worn parts become gradually separate narrower paths which are too close together and too complex and too numerous to map individually, but the grass is no longer used for walking on.

The key question is: how do we tell potential path users something useful about this path?

We cannot map the facts (using a human mapper and working at the scale that OSM recognises at least). This is entirely an impractical idea. There are not even any average facts which help.

I can put things into simple words which define it well though…. this is a wide key path, created by wear by walkers, cyclists and horses, initially on smooth firm grass, later with significant usually muddy sections, and later deeply rutted with surrounding rough grass.

Surely we can’t continue on OSM to simply call this ‘a path’ and be happy with this?