Would it make sense to treat glacier:type=rock as preferable to geological=rock_glacier? These two values appear to express the same thing and recommending more established one as preferable makes more sense to me.
See also Rock glaciers retagged as moraines
I created Tag:glacier:type=rock - OpenStreetMap Wiki and Tag:geological=rock_glacier - OpenStreetMap Wiki - second page currently hints that alternative tagging may be preferable but is not marked as deprecated
In general, I am in favor of keeping different tags in OSM where there are distinctions in meaning, however subtle, to preserve the richness of information in the map.
In this case, though, the two tags appear to be synonyms with no significant differences. The glacier:type=rock tag is well documented and much more widely used (408 elements) versus the geological=rock_glacier tag (15 elements).
As far as I know, neither tag has been used for rendering. I would be interested to know if there are any data consumers using either of these tags.
When I included “Rock glacier” in the Glossary of landforms, I documented the glacier:type=rock tag for the reasons above.
I did not include geological=rock_glacier as a possible synonym or less preferred tag, but that could be added if there’s a consensus to do that.
@Hungerburg raised a good point in the Rock glaciers tagged as moraines thread that rock glaciers may be tagged as “scree” rather than “glaciers” because the surface layer is rock rather than ice.
One of the main characteristics of a rock glacier is a surface layer of rock that insulates a layer of permafrost at the base of the feature. In many cases, this surface layer is accumulated from scree, talus, or glacial deposits such as moraines. We would mostly tag this as natural=scree in OSM, but some rock glaciers consist of rocks too large to be called “scree.” In this case, the currently undocumented and rarely used natural=boulders tag might be better.
I figured it would be good to survey current tagging for rock glaciers, and these are my results from Overpass today:
glacier:type=rock: 439 elements
+ natural=glacier: 436 elements
+ surface=scree: 103 elements
+ surface=snow: 7 elements
+ natural=scree: 3 elements
+ surface=ice: 2 elements
natural=glacier + surface=scree: 135 elements
glacier:type=* + surface=scree: 108 elements
geological=rock_glacier: 15 elements
+ natural=scree: 4 elements
That is not to suggest that any of the current tagging is correct or preferred, but just to look at what mappers have chosen to use previously when there has been little documentation.
Are there other tag combinations we should consider?
Another thing to consider is that a “true” ice glacier may carry a medial or supraglacial moraine, which we might tag as natural=scree, but this does not make it a “rock glacier” because the body of the glacier is still ice.
I feel uncomfortable with glacier:type=rock as subtag of natural=glacier, as most people understand a glacier as mainly composed of ice.
Most of the examples of rock glaciers presented here are more of the kind “rocks moving similar to glaciers”. They may contain a considerable percentage of ice from the underlying and melting permafrost, but ice is not visible at the surface and perhaps no longer present in the lower parts. Without digging in they are not distinguishable from unstable still moving landslides or scree fields.
A “true” glacier even when covered with scree, rocks or middle moraines shows its ice core at crevasses or at the glacier’s snout
This is a reasonable argument, and possibly a reason to prefer geological=rock_glacier even though other tags are more widely used.
The main natural=glacier tag could mislead data consumers into interpreting a rock glacier as an ice flow. We’re not tagging for the renderer, but as an example of misleading a data consumer, OSM Carto does render natural=glacier + glacier:type=rock + surface=scree the same as natural=glacier without the additional tags.
(I just wish we could apply the same logic to dry lakes tagged with natural=water.)
are these intermittent/seasonal? Or always dry?
It depends on the case. Some “dry lakes” have never (within recent geological history) had standing surface water at any time. Some may have standing surface water in small areas during periods of significant precipitation. But then, we don’t tag the puddles in parking lots as natural=water + intermittent=yes.
then natural=water is simply wrong for them
Those three are all in Tyrol - where I live. The reason for the natural base tag being scree is of course, that glacier:type=rock got only added later to the element. Conversely, all the rock glaciers mapped are at the same time tagged scree. That is what non-geologists see there. A few might be mapped fell.
There are around 3000 rock glaciers in Tyrol, I bet quite a lot of them are mapped scree as of today, only missing the extra rock glacier. If I’d do as @Mateusz_Konieczny proposed in the fine wiki article I could turn the mountains quite blue in the Standard View. I rather not do that.
I am fine with glacier:type=rock for the extra bit though - after all, here they are called Blockgletscher. A culvert isn’t a tunnel neither, (nearly) nobody has a problem.
I know the feeling. We have some very blue areas in the local Colorado, Mojave, and Great Basin deserts even though there is really no water there.
I think it makes sense to tag rock glaciers as natural=scree since that is the natural surface of the landform.
And I might propose that we prefer geological=rock_glacier over glacier:type=rock for two reasons.
glacier:type key is less appropriate.glacier:type key implies that it should be combined with natural=glacier where instead we prefer natural=scree.From reading the definition, a rock glacier mingles gravel, water and ice, while it flows like a glacier of pure ice, further considering, that scree/talus is gravel that accumulates below cliffs, it seems fine to me to make natural=scree the so-called base tag for them - which everybody can apply from looking at the matter - This does not misuse the natural=scree tag, does it?.
I find geological=* a bit unwieldy, scree:type=rock_glacier perhaps 
Sorry for hiding behind the smiley - I am a bit overtasked with this.
UPDATE: Maybe the unwieldy geological= is fine after all, and reasonably preferred over creating an OSMism. Subject matter will forever be something of interest for a very small part of the community, no need to curry favour with the non-geologists, e,g. myself.
UPDATE2: Browsing the academic literature - there is quite some amount - 8 to 9 of block glaciers in Austria are of the scree type. (Yes, there are several block_glacier:type=*.) - That leads me to pick up this statement:
Now, who is going to take care and remove that misleading association from the documentation?
I don’t mind updating the Wiki pages. I think we have consensus here and in established usage that the correct primary tag for a rock glacier is natural=scree, so that can be documented in the Wiki.
I’m not sure that we have consensus on a secondary tag for rock glaciers, though. Using scree:type=rock_glacier seems a little unwieldy to me, but we only have opinions from four people in this thread and the opinions that matter are those of people actually tagging features in the map.
I may summarize the opinions from this discussion in the Wiki documentation and leave it to mappers to decide which tags are most appropriate.
In the interest of full disclosure, there is one Wiki user who seems to be unhappy with the changes I made today. If you like you can review the changes at:
This Wiki user is of the opinion that, “There is so far no substantial use of natural=scree for explicit mapping of rock glaciers to be found in the database. Use of natural=glacier for this purpose, however, both with (in particular South America) and without glacier:type=rock (like in Afghanistan), goes into the hundreds.” (See Talk:Tag:glacier:type=rock - OpenStreetMap Wiki)
@Mateusz_Konieczny and I tried to encourage this user to join the discussion here, but they don’t seem interested.
I wrote Hungerburg's Diary | Rock glaciers in the Eastern Alps | OpenStreetMap with references to dependable data - the result enforces the first impression of mine when joining this topic.
So “use of natural=scree for rock glaciers in the Eastern Alps easily goes into the hundreds”
Five mappings worldwide makes this rare indeed. I like the idea. I imagine split opinions - some map scree, some map bare_rock. I enjoy “Blockkletterei” (this is not bouldering), as well as “scree rides” 
Definitely not rock climbing, but no riding neither.
I guess this is a different topic, but I’m in favor of the natural=boulders tag for other cases where the formation of boulders is due to different processes than the formation of scree or talus.
See, e.g., Blockfield - Wikipedia
Also, just because a tag is rarely used does not mean that it is not a valuable distinction for mappers to make.
Edit: natural=bare_rock is exposed bedrock that is still intact. This is distinct from natural=boulders which refers to an area of large, loose, stones that are not directly attached to the bedrock (i.e., a large area covered with individual natural=stone features).
In contrast to the other alternatives this is definitly wrong for rock glacier formations.
bare_rock is the fixed surface of underlying rock, rock glacier is moving.
@Kai_Johnson @seichter No worries, I won’t tag something bare rock where there are loose boulders or gravel - Yet, it should not only apply to Yosemite Granite - that would make it hard in other places, lots of summits here rather look like a pile of stones, eg File:Lüsener Fernerkogel (20200115 143514).jpg - Wikimedia Commons
I did a vote with my feet today and mapped some rock glaciers. During that, I learned of Tag:geological=moraine - OpenStreetMap Wiki - makes me like the key some more. I did double tag, glacier:type and geological - the latter matches better, in my opinion.
Here to the oldest glacier:type=rock mapping
Here to the oldest geological=rock_glacier mapping
Nice work!
I notice that the Key:glacier:type - OpenStreetMap Wiki page didn’t even exist until June of 2020. The key was mentioned on the natural=glacier page earlier than that, but the page didn’t say anything specific about rock glaciers at that time.
If you’d like to see some cool history of moraines, check out Devil's Lake (Wisconsin) - Wikipedia, or a more formal geological history of the area.
Devil’s Lake sits above the surrounding terrain, with no apparent inlets or outlets. It’s flanked on two sides by the quartzite bluffs of the former Wisconsin River gorge, and capped at the other ends by terminal moraines left by ice age glaciers.
It’s also a great place to go climbing and bouldering.
.jpg){kind=link}