RfC: Highway=Scramble

Some reading later, I conclude, that the Welsh use the term in just the same way as Jake from the US: From the point on, where hands are needed, walking ends and scrambling starts. Use of hands therefore is the single most defining feature, be it for keeping balance, be it for pulling up. Trail visibility is a nice add-on attribute, in the same class as trail blazing or assisting measures.

On a side-note: This reminds me of the history of alpine hiking, a 150 years ago: The Britons had quite some say in this; For them, use of hands was not something, that a lordly person would do, in order to advance. So locals performed long searches, to find ways for their customers, on which to reach the summit in a walk-up. Quite often here, these sections are mid-route, one has to get by one or more “steps” in the terrain, where continuation is not immediately obvious and the itinerary seems puzzling at first.

Here it says https://www.walkupsnowdon.co.uk/snowdonia-walks/crib-goch/

3 The path from this point on becomes a full on scramble…

One may read this: path=walk turns into path=scramble. I’d say, the difference warrants the scramble a top level entry in OSM tagging. Data consumers should be able to rely on path=walk being the default and not having to look at extra attributes that override the assumption.

Thank you all for the valuable input! Maybe meet on the Wiki!


I think this idea has merit, but it’s important to note that an unintendended consequence of introducing this new tag highway=scramble would be the fragmentation of existing hiking trails. Many mountain trails include sections that require scrambling, and they aren’t just at the top. As these sections get converted from their current highway=path|footway to the new tag, they will appear as gaps on many existing hiking maps until they start rendering the new tag. It is stated in this thread that some renderers would likely never display highway=scramble. That being the case, the gaps would be permanent. Perhaps the idea is that this new tag would only be for full scrambling routes that really aren’t trails in any sense, and short scrambling sections of trails would be excluded. That seems like a tricky distinction to document, though.


I’d say, the renderers most likely to never display this new tag, are also the ones, that do not display hiking route relations now. Gaps indeed will show up badly. A welcome incentive to update data import strategies, for those who care :wink:

For trails, if they are mapped as route relations that needn’t happen. Here is an example which is shown like this in a pedestrian map style and like this on “Waymarked Trails” (which is pretty much the gold standard for hiking routes).

1 Like

A similar example from Ireland: “To reach the summit, you’ll need to spend a short time balancing on hands and feet in the final scramble to the top”. I found lots of pages about hiking this mountain (a popular hike for beginners/families in Ireland) that use very similar words. The proposed highway=scramble seems suitable to me to map the paths at the top of that mountain that are currently tagged as sac_scale=demanding_mountain_hiking. I don’t think it would be problem if those sections disappeared from “generic” maps - if it discourages people from using unsuitable map renderings for outdoor recreation, all the better.


I agree with those proposing a new highway=* value. I’m fine with highway=scramble or some other value.

One of the challenges we face in OSM is that highway=path is so overloaded and expansive that it can mean anything from a paved concrete sidewalk through a city park to mountaineering routes requiring ropes and anchors to protect from falls. The ends of this spectrum are fundamentally different even if there is a continuum between them.

A corollary is the road network: There might be a lot of gray-area between highway=primary and highway=secondary , but in most cases no one is going to confuse a road tagged with either of those with highway=residential or highway=motorway. Breaking the spectrum of roads into broad chunks isn’t always straight forward (and can require retagging due to misconceptions), but doing so makes the road network more intelligible to data consumers.

The proposal being discussed to segment off one end of this spectrum based on some relatively clear distinctions seems like a useful way to begin to reduce the scope of highway=path to something slightly more constrained.

Additionally, as was mentioned in another thread, there are real-life consequences to having naive renderers and routers treating all highway=path in the same fashion whether they are paved sidewalks, maintained walking trails, or mountaineering routes. By moving the most dangerous types of foot-traveled routes out of highway=path and off into highway=scramble data consumers will default to a “more-safe” gap if they don’t understand how to handle such potentially dangerous ways.


We have contributors that have mapped highway=path, trail_visibility =no, sac_scale=demanding_mountain_hiking for years (if not more than a decade).
This has been crafted by hand, and also by feet and sweat in this particular case.
Proposing highway=scramble or else as a synonym will end up as it is: an alias, and you will find either way of tagging in the database for the next decades.

If some data consumers fail to recognise long established keys such as trail_visibility and sac_scale, what do we hope to achieve with a new tag? More case to dealt with in a style sheet? More omissions?

I always found that the best way to deal with helplessness is education.


An OverPass query for
highway=path and (sac_scale=demanding_mountain_hiking or sac_scale=alpine_hiking or sac_scale=demanding_alpine_hiking or sac_scale=difficult_alpine_hiking)
does a good job of finding candidates for retagging. According to taginfo there are 59,939 ways with the four highest sac_scale=* values and with the top three values only having ~20,000 uses. Retagging these from highway=path to highway=scramble would not be an insurmountable job if a proposal like this goes through. With a bit of local knowledge tagging could likely be updated without requiring a re-survey.

While education is ideal, many map renderers simply don’t care about mountaineering details – many build maps targeting an urban user base and show paths to get their park renderings looking good. Their users then try to use those same map renderings in the wild and get themselves into trouble because wild areas are crisscrossed by undifferentiated paths.

1 Like

So, you say that 59’000+ scrambles on the three highest sac_scale path is an easy remap? Be my guest :wink:!

Even where it is at its best, OSM is far from excellent in the wilderness, as opposed to the next door park.
I would say we must be caution to change mapping habits of the few of us that deals with remote areas. Maybe they are dedicated enough to both their hobbies to remap 59’000 ways, but maybe not, and we certainly don’t want to disgust them in any way.

They would probably be pleased to see their work on every map around, but just with a proper rendering, don’t you think?

Actually, that is not my intention. In my opinion, a scramble is pure fun. Yet, I know, that for others, these are insurmountable obstacles. The sword is a double-edged though, so your comment welcome still. Maybe I should mention an upper limit, eg. climbing:grade:uiaa=III is only in, when not exposed? I rather keep the proposal tightly worded and not a weekend lecture. Maybe stop at UIAA II, like sac_scale would command, but mappers seem to just ignore.

Update: That is an easy one, grade II will be the max, whatever some mappers argue, to have mapped as path, that goes above. Scramble will be a subset of what can be mapped path, at least as far as I could honestly do.

Yes on hiking maps that render route relations this is not a problem. Sadly there are many hiking maps that do not. So for this new tag to be successful, outreach to these map makers would be needed. Otherwise users of those maps will see the invisible sections as errors and change them back to highway=path.

1 Like

As you say, it depends a lot on how popular the tag gets, to bring downstream vendors on board. I’d say, there should be a sufficient number of mappers, that value concise tagging above OSM-Carto rendering. Chances are not so bad, IMO.

Surely, any map that does not render hiking routes is not in any sense a hiking map?


In North America at least All Trails, Gaia GPS, and Cal Topo are three of the most popular apps for hiking. The default basemaps for all three are OSM based and do not display route relations*. I don’t know how popular these apps are in other parts of the world, but their maps cover the globe so I imagine they are used outside of North America as well.

*edit: actually I think GaiaGPS does use route relations to display certain well known long distance routes at low zooms. They aren’t differentiated at high zooms though.

1 Like

All Trails and Gaia have representatives working with OSM and some US Federal agencies to work on improving trail tagging especially with respect to informal or social trails. I strongly suspect that they are looking at more than just highway=path when rendering. It would not surprise me if they are using the SAC scale and trail visibility tagging when deciding how a trail should be shown or even shown at all.

1 Like

If a scramble is defined as trail or route where hands are needed, then would a ladder be highway=scramble? If not, how would you define highway=scramble to exclude ladders. Many trails have ladders on sections that would otherwise require scrambling or full on rock climbing if the ladders weren’t there. I wouldn’t call ladder climbing “scrambling” in normal speech, but climbing a ladder surely does require the use of hands. Would ladders need to be a yet another primary tag (highway=ladder)? In some ways a ladder is more similar to highway=steps or highway=via_ferrata, but those tags also don’t fit.


In a highway=scramble, a ladder will be adequately mapped as a node, after all, ladders are mostly vertical, so node is a good fit, and the tag already exists, it even is approved. But if you want, of course you can map ladders as a miniscule way, if you are into that kind of detail.

Re-reading my post I see that I could have been clearer. I wasn’t thinking about mapping a ladder within a highway=scramble, I was thinking about mapping a ladder as a highway=scramble within a highway=path or a highway=footway. Although steep, ladders are often not vertical and I could easily see micro-mappers using a short way instead of a node. With the definition being “hands required”, I could see mappers tagging ladder sections of a trail as highway=scramble (because hands are required) much like they would tag sections of steps as highway=steps. Hand rails and cables are also often found on steep or overly exposed sections of trail and I could see some mappers interpreting these sections as “hands required, therefore highway=scramble”. Would you consider usage like this as appropriate? If not then I’d suggest stating in the proposal that assisitive devices like ladders, handrails, or cables do not indicate a highway=scramble.

as the reason for this idea seems to be safety considerations, what about this: https://luftschubser.de/wp-content/uploads/2018/02/Elbsandsteingebirge_Minimulis_501.jpg

it’s not difficult at all, nor is it particularly narrow, clearly it would not qualify for scramble, still many people would want to avoid it, especially with kids or those with acrophobia

Not a native English speaker here, so have to listen on what those who are conceive of a term. If you say, you never scrambled a ladder, I believe. We here, we do “climb” ladders. Which is even more obnoxious. UIAA is silent on ladders, but in my opinion, a 10 or more meters high ladder, as seen on the picture you posted, has great potential to make the trail a scramble, from the point on, where you cannot escape that ladder.