Feature Proposal - Voting - highway=scramble

The SAC scale is a hiking scale and any scrambling/climbing sections that may occur on T5 or T6 paths will only be of UIAA grades 1-2, which are relatively easy. The difficulty on such paths usually comes from other factors like exposure, glaciers, loose terrain etc.

Bigger rock climbs indeed shouldn’t be tagged highway=path.

1 Like

Slightly OT, the SAC itself considers such comparisons dicey and points out that the higher hiking grades may in real life terms be substantially more difficult than low difficulty climbing grades and there is no continuum between hiking and climbing grades (besides the reasons you mention, absence of protection in any form may be an issue).

See https://www.sac-cas.ch/fileadmin/Ausbildung_und_Wissen/Sicher_unterwegs/Sicher_unterwegs_Wandern/2020_Berg_Alpinwanderskala_EN.pdf

Yes, I didn’t want to go too far into that, that’s what I was alluding to with the difficulty coming from other factors. I just wanted to clarify that this is not about classical rock climbing pitches belonging in the highway=path space.

Conversely some out of shape, morbidly obese ex-Covid patient in the United States won’t be able to walk up slight incline without crawling because their knees went out (something that I just saw happen a few weeks ago BTW) and decide that it’s a “scramble.” So, I don’t really how highway=scramble is any better then highway=demanding_path. Not that I think that tag is great either BTW, but they both clearly have the same problem of vagueness and the potential for random “grandmas” to miss-use them.

Although I’d argue highway=scramble has the problem more then highway=demanding_path would. Since there’s legitimately more chance of someone having to use their hands on a normally otherwise benign path then some random grandmother being like “hey, there’s a cow. Wow this perfectly flat cemented walking trail is rather demanding!” or whatever.

1 Like

I’d argue that average person today is “couch potato” or “grandma”.

Expect the minor detail that in all that time almost nobody succeed in making it work in practice.
Maybe it isn’t “in scope” of what proponent has initially envisioned, but it is hugely overlapping issue, that could (and indeed, IMHO should) be solved at the same time. Because if we approve highway=scramble today, more inclusive highway=demanding_path that gets suggested would be rejected as overlapping (or would need to deprecate highway=scramble, leading to even more friction).

Agreed. They are still demanding even if not lethal, hence fit highway=demanding_path + demanding_path=scramble just fine. Their technical difficulty / danger could be further refined with sac_scale=*.

Agreed. Which is exactly why I didn’t propose it that way (e.g. separate highway=scramble, highway=jungle_vegetation, highway=almost_unpassable_mud etc.) but instead as one highway type (highway=demanding_path or something similarly generic) enhanced with extra tag (like demanding_path=scramble or demanding_path=jungle etc.)


That’s actually a pretty legit idea. One of the problems I have with highway=scramble is that you can’t really enhance it with extra tags. Like ‘Scramble=jungle’ just sounds weird. I think extra tags could work with highway=demanding_path though.

1 Like

For me it is much simpler than this and not really about how hazardous or difficult a way is. Its about how one functionally uses the way. A highway=scramble is a way where an able bodied person can scramble with hands and feet, but walking, riding a bicycle, or riding a horse is not possible (except perhaps for highly skilled stunt riders). Mappers sometimes add bicycle=no and horse=no to trails where it is not possible to ride a bicycle or horse, but this is technically an incorrect use of access tags since bicycles and horses may not necessarily be legally prohibited. For example riding a horse or bicycle is allowed on most, if not all, trails in US National Forests. However, this includes plenty of scrambles where it is essentially impossible to do so.

Separate tagging for hazardous paths is also worth discussing, but seems more difficult to explicitly define.


If that is only purpose, doesn’t sac_scale suffice for that? e.g. sac_scale=alpine_hiking says “Use of hands needed in order to advance in certain places.” which seem to do exactly that.

The main advantage of different value of highway=* instead of extra tag (like sac_scale=* or scramble=yes or foot_and_hands=yes) is AFAICT only in that it does not show / route on regular maps, but only on very specialized maps which need to invest extra effort to add that extra support for it. Or do you disagree?

No. Just as I don’t need to look for the presence of bicycle_scale tag to know that a highway=cycleway is a way I can ride a bike on, I shouldn’t need to look for the presence of a sac_scale tag to know that a scramble is something I can’t ride a bike on. If it’s tagged as highway=scramble, the primary tag makes this is clear.

The advantage is that highway=scramble alone clearly and simply communicates to data consumers:

This is a way you’ll need to scramble on

On the other hand highway=path alone communicates to data consumers:

This is a way that you might be able to ride a bike or a horse on, might be able to walk on, might only be able to scramble on, probably can’t drive a motor vehicle on, might be quite wide, or might be very narrow. However I can’t say anything for sure about any of these things, so go look at a variety of other tags to find out how you can use this way. Oh none of those other tags are present? Sorry I’m not a very useful tag.


Ohsome dashboard gives 2014 for the combination of path and difficult_alpine_hiking of those 2240 that taginfo reports. So ~90%; of the rest 96 are via ferratas, especially the A/B ones often can be hiked without the gear. What you say about local consens certainly does not apply to the German community. Many T6 paths in my area were mapped by Germans.

The SAC has no climbing scale, it relies on the UIAA in this. It has a Hochtourenskala, and this is what the comparison alludes to.

The English document is relatively new, but compared with the 2002/12 one, that was the base of the OSM tag sac_scale, mountain hiking seems to have increased in difficulty since then.

Back on topic: In this new English version, it is not called climbing, but scrambling instead :slight_smile:

I fully don’t understand the beef about attributes. I am a mapper, and I want to map what is there, in the most uncontrived way. A recent vote complained, that this proposal repeats the mistake of highway=steps, which correctly should be highway=path+steps=yes. Who else is sure about that?

If there were steps for primary, secondary, tertiary, unclassified, residential, service, track, …, I’d certainly vote for the latter.

1 Like

Actually, you do have to look for that extra tags for bicycles, if you are any kind of bicycle-related data consumer.
True, it might be mapped as highway=cycleway (due to obsolete historical reasons), but it also might be mapped with much more popular newer tagging scheme:

  • highway=path + bicycle=yes|designated, or
  • highway=residential|secondary|... + cycleway:right|left|both=lane|track|... (or cycleway=*)

And bicycling paths are (arguably) more common and more wildly used than scramble paths!

If it were suggested to be created today (and this comes from the avid cyclist!), I’d argue against creating highway=cycleway (as it brings nothing new to the table, and all data consumers still have to parse other bicycle-related tags, so it is just more work and TIMTOWTDI).
Note that even with highway=cycleway it may also be that that pedestrians or horses or motor vehicles are also allowed on that “cycleway” (as it simply means highway=path+bicycle=designated; due to the fact that it comes from time before *=designated were a thing, but that is historical artifact, and not an invitation to create a separate new highway value for each and every possible combination, e.g. highway=cycleway / footway / footway_and_cycleway / footway_and_bridleway / footway_and_bridleway_and_cycleway / footway_and_cycleway_and_bridleway_and_scooterway etc.).

Hey, there are even some thousand highway=cycleway on which you can’t actually ride bicycle (i.e. bicycle=dismount); how about that?!

Exactly as does highway=demanding_path + demanding_path=scramble, does it not?

The point is that highway=cycleway makes it clear the way is for bikes. If I need to know more I’ll look at other tags, but maybe I don’t. There are many different use cases.

1.2 million uses and climbing is neither obsolete nor a historical artifact, but lets keep this topic focused on scrambles.

This combination may communicate the same thing (I don’t know because the proposal hasn’t been written), but the key word in my statement was “alone”, as in only one tag needed. So no, not exactly.

But it does not. If you look only at highway=cycleway, you’ll miss most of the cycleways (i.e. get only 1.5 million cycleways out of more than 8 millions of them). So you absolutely need to look at other tags used to mark cycleways, or you’ll end up with completely useless map for bicycles.

more detailed calculations
highway=cycleway 1528k
bicycle 6735k-1449k no
cycleway 944k-267k no
cycleway:left 203k-78k no
cycleway:right 584k-223k no
cycleway:both 795k-732k no

1.2 million uses and climbing is neither obsolete nor a historical artifact

But it is. It only seems its usage is growing because of the way iD editor choose to handle the situation (i.e. mark cycleways both with highway=cycleway and bicycle=designated). If you replace those redundant highway=cycleway with cycleway=path, you’d get a much lower number, which continues to fall.

but lets keep this topic focused on scrambles.

Well sure; you’re the one who brought highway=cycleway into discussion! (It’s not my fault that cycleway numbers happen to speak against highway=scramble - /I’m guessing/ somewhat contrary to your expectations :smile:)

So, the gist of your support (as I understand it) of highway=scramble specifically (compared to highway=demanding_path + demanding_path=scramble) is that you would prefer to type-in manually one tag instead of two?

Should I assume then that you don’t use iD, JOSM, or Vespucci? As those editors (amongst others) support presets which automatically set any number of tags when you select them (so exact number of tags some preset have is irrelevant to the user doing the editing)?

That is quite a reductive mis-charactarisation of my position. I’ve done my best to explain my thinking on the matter, but you don’t seem interested in actually understanding. I’ve already stated in the other thread on this topic:


Please, do not imply malice on my part. Because I do try to understand your position, but as I didn’t (in this thread which I follow) found other arguments you might have against highway=demanding_path + demanding_path=scramble in this thread, and you didn’t mention it when asked about, I assumed there were none. Please do let me know if you have some arguments against such solution (other then already mentioned “two tags are more to type than one”)?

I’ve already stated in the other thread on this topic:

Perhaps that might be the reason for me not finding your other statements on the matter: I’m not omnipresent :smile:

I see I have another 77-long-messages thread to contemplate, hopefully some gold nuggets might be found there which are missing here. It will take some time for me to process it, though, so please do let me know if I reply in a way that seems like I’m intentionally ignoring your arguments (I assure you, any such failure of mine is due to omission, not malice).

Oh, have I known it sooner, it would’ve saved some replying here! So, I gather from this short quote that you are not in principle against such idea? (I’ll read up details from other thread when I get some time).

To quickly comment on it as to not leave you hanging: Yes, highway=demanding_path + demanding_path=scramble is wider proposal than the original one by @Hungerburg, but that is very good IMO – main purpose of Proposal process is indeed to bring together potential tag users from different backgrounds so the more generic tag solution can be found (as it is much better if 1 proposals of 2 tags can solve 5 different needs, than if there need to be 5 different proposals, each aiming for 1-tag solution just for their narrow field of interest).

When voting turns into “a comments section” many times that is more indicative of the responses to the RFC input not being actually addressed:
“The Rationale is not negotiable.” x2 - the submitter of this proposal straight up said that they were not flexible to user feedback during the RFC process.
1- Suggesting that highway=path + an add on tag was sufficient
2- A RFC specifying that the tagging is non-intuitive where the commenter lists three different definitions for what constitutes a ‘scramble’ that don’t seem to align

“The proposal cannot and shall not deprecate that.” - thus admitting that highway=path is prefer but insists a replacement is still necessary

“only map scrambles where on location signs of actual use by humans” - I’ve done a lot of mountain hiking, even blowing my knee after a summit, and while mountain hiking/climbing books refer to scrambles I’ve never seen one with a sign as it’s simply one of those “I know a ___ when I see a ___” things

“I myself cannot imagine scrambles in other domains as hiking, but others may.” - A single query for “scramble” in TagInfo will pull back 100+ tags that already exist for crossing:scramble (i.e. the “pedestrian scramble” - where all cars get red lights and pedestrians can walk all directions including diagonally. Far more applicable to OSM and only one ranking behind hike scrambling in Google search rankings making them on par with one another and the call for distinguishment between them to be reasonable.

Under “Subjective” the proposer responded, “Additional pictures have been added to exemplify, what use of hands is about.” The issue is that the pictures he pulled happen to be the same from sac_scale=demanding_mountain_hiking AND sac_scale=alpine_hiking BOTH meaning the way they addressed the subjectivity issue of ‘scramble’ is by pulling pictures that span what is already less subjectively defined by the existing sac_scale, which I might point out already has 697,571 tags in OSM use already and is far better suited to the task at hand.

The proposer stated as a resolution, “I hope, that resorting to values in the UIAA scale, as has been in the text from the beginning, will do.” - Then UIAA exists as a tag and UIAA=1 would be the equivalent to a “scramble” although I still think sac_scale does a much better job at deliniation but I see no reason both existing tags with thousands or tens of thousands of uses each can’t be used and don’t already fulfill the need.

“The proposal text puts now even more emphasis on the similarity with highway=steps” that a response to " Existing tags like sac_scale=demanding_mountain_hiking can be used" because highway=steps are clearly definable and have clearly defined starts and stop and clearer still impact on their use by bicycles and the disabled. While ‘scramble’ has none of these distinguishing factors simply pointing to something else doesn’t address the original comment which is that sac_scale not only can be used but already is being used to describe this exact “thing” and in more clearly delineated detail and with a scale and not a binary yes/no flag

“When this goes to vote, I would not be surprised, if the ‘too late’ argument or the ‘we can do that with attributes on paths’ argument may prevent approval, so maybe voters should be asked to write that in the vote, to ease counting of reasons.” - I realize the argument being made prior is that OSM-Carto makes some poor decisions based on highway=path but that isn’t a problem with highway=path that is a problem with OSM-Carto. This goes to “don’t map to the render” as OSM best/good-practice. This doesn’t mean that it’s too late but it also doesn’t disqualify “we can do that with attributes on paths” as an argument as a scramble is more or less a path especially ones that are well trailblazed. In fact, the insistence on sticking with highway=path is largely in line with the one feature, one tag ideology, as well as these, are paths first, hiking/climbing paths second, and at best a scramble third if the first two don’t do the job, which I feel they do.

Bottomline, is if a proposer listens to RFC feedback and actually changes their proposal to address the concerns of those taking the time to comment then those same comments should not need to be repeated in the voting process. I should never hope to see, “The Rationale is not negotiable.” spoken ever as an RFC response and highlight the problem, not as one with the voting or the RFC process but lays those issues at the feet of the proposer who by their own admission in the RFC process expected and discounted the responses they know would likely come in advance.


The value of a simple tag mainkey=X for an observable feature X which is generally called X by most people is, that it makes life easy for mappers and data users. One attribute tag X=yes comes in as second on the easyness scale, and mainkey=garbagecanvalue garbagecanvalue=X is third.
Combining attributes which may or may not imply X is at the low end of the easyness scale.

I would prefer to tag a scramble section of a trail, path or route as such as a feature exactly where it actually is, as directly and as simple as possible, and in a way that data consumers can use as directly and as simple as possible.

If a trail contains scramble section, it is probably tagged with a sac_scale value that says there probably wil be scrambling sections. As a hiker, I would like to know for sure where exactly these sections are on my planned route, in order to plan a workaround. I know other hikers who would seek those sections out for their trips. As it stands, they can’t, not with OSM-data.


The problem is that scrambles aren’t observable features. Otherwise, how would someone go about observing one? By sitting at random places along a trail that they think looks “scrambly” or whatever until they see X amount of hikers falling and using their hands to crawl up the rest of it?

Like where’s the threshold? Say 10 people used their hands and 7 didn’t is that a scramble? what about if 7 people used their hands, 4 didn’t, but 4 of the people who used their hands are obviously out of shape. Is that a scramble?

What about if a mapper just sees a hill on satellite photos or in person, thinks it looks like something they can’t hike up without using their hands, and tags it as highway=scramble. Would that be OK? Or is that not a highway=scramble because they didn’t actually observe an “able bodied person” hiking it?

What if someone wants to verify a highway=scramble after it’s been mapped. Do they go through the whole observation process again or what?

1 Like

If “use of hands” cannot be observed, all references to “use of hands” have to be removed from the sac_scale key documentation. This will leave some glaring holes.

BTW: In Australia scrambles, in the sense of having to use hands, are mapped as “demanding_mountain_hiking” (T3). So much on the praised precision of sac_scale.

1 Like