So, unhewn beats natural and cobblestone beats cobble/cobbles.
So, how about unhewn_cobblestone then?
Glad to see some good suggestions, and at least to my non-native ears, unhewn_cobblestone sounds like a workable option that gets the distinction across. 
Then should we also define hewn_cobblestone in order to slowly phase out cobblestone?
Absolutely! I think since an agreement has been reached here, what’s left to do is simply to document it in the wiki.
I will then also add it into the next StreetComplete version.
I haven’t seen “hewn_cobblestone” in this thread yet, so can you elaborate a bit what you would use this value for? Perhaps using the images used before (e.g. here) as a reference point?
Oh, I assumed that he mistyped it and meant unhewn_cobblestone. Basically, hewn cobblestone would be sett.
Nevermind, hewn_cobblestone would really be the same as sett.
Ah, good to have that cleared up. Glad that we’ve reached a consensus then. 
What remains is updating the wiki and getting mappers and devs to adopt this tagging. I’m happy to do my part for the latter by adding support to OSM2World.
Mh, perhaps ftbrien or Tordanik could add this to the wiki, since both of you extensively analyzed the wording over the time in the wiki, so you have the best idea how to exactly phrase it.
I am ready to do my part by letting StreetComplete use the new tag, but I would want to have it documented first.
Done. New definition, let me know if you agree/disagree:
The definitions sound good to me, thanks for adding them to the wiki.
What about the cobblestone:raised value you added a while ago, though? Wouldn’t that also fall under unhewn_cobblestone using our new definitions?
I’ll let you remove that if you think it should go. I added it because (1) nobody complained about it on the tagging list when I suggested it, and (2) many have pointed out, when referring to that image, that such pavement significantly affects traffic, particularly pedestrians and cyclists, and that such pavement seems to be intended to deter traffic.
I think I’d remove that then. The raised cobblestones would then really just be a subcategory of unhewn cobblestone and there is no clear line to distinguish these two (less raised vs more raised). Distinguishing unhewn from sett is already a good step forward.
Done: https://github.com/westnordost/StreetComplete/commit/bf387f21e75ef7b4c5e1738f25bdc16dfd59c48b
And removed cobblestone:raised from the wiki.
I have seen that in the wiki are strikethrough out and do not advise using the type of surface cobblestone:flattened to avoid confusion with sett and unhewn_cobblestone.
I think that in the discussion did not take into account a way of making pavements that is used in the south of Brazil and northeast of Argentina (see in overpass: http://overpass-turbo.eu/s/x19 ), which is called “brazilian cobbled style”, which is a variant of the known as portuguese pavement (see: https://en.m.wikipedia.org/wiki/Portuguese_pavement ).
The brazilian cobbled style is made with basalt stone and fits what can be defined as “cobblestone:flattened”: natural stone, crushed, smoothed surface, irregular, binder witk earth.
As you can see in the images:
https://wiki.openstreetmap.org/wiki/File:Empedrado_brasilero_1.jpg
https://wiki.openstreetmap.org/wiki/File:Empedrado_brasilero_2.jpg
For this surface, I think that the type sett (stones with regular format) or unhewn_cobblestone (uncut, with rigid binder) is not suitable.
For this, I request that consider modifying the wiki and not removing the “cobblestone:flattened” option to be able to use it on brazilian cobbled style pavement.
This is polyhedral pavement, not Portuguese pavement.
The query returns 829 (0.08%) ways out of 995314 ways in the same area tagged with surface=*. I took a random sample of 19 ways to check whether usage was consistent and found that most were incorrect assignments:
- 6 should have been mapped as sett
- 5 were polyhedral pavement
- 4 should have been mapped as dirt or earth
- 2 were actually Portuguese pavement
- 1 should have been mapped as paving_stones
- 1 should have been mapped as concrete:plates
So only ~11% of those ways represent actual Portuguese pavement. To me that means that the dataset should be evaluated in greater depth before we can jump to the conclusion that the suggested usage is widely adopted locally.
The value cobblestone:flattened has never been defined explicitly in the English wiki with those characteristics, as you can see in this edit history summary. There you can also see that the images used to represent this surface changed significantly over time, and the images present during the period of greatest adoption resembled sett much more closely. The Spanish wiki did not have an entry for cobblestone:flattened until 10 October 2015, when it was first described simply as “empedrado liso” (flat stones pavement), with no picture. The same description remains to this day, having received the last English wiki image (#7) via transclusion since 23 May 2016. Most of what is mapped as cobblestone:flattened in the world today was already mapped by then. The Portuguese wiki article on surface was translated for the first time in 4 January 2018, so for most of its history it simply copied its definition from the English wiki and was inaccessible to non-bilingual local mappers.
Also, Google Images is not returning any samples of either Portuguese pavement or polyhedral pavement when searching for “empedrado liso”. In fact, it does not look like a commonly used expression, which was the same issue with “flattened cobblestone” (and explicitly written as description for cobblestone:flattened in the English wiki since mid 2015). Searching for “empedrado liso” with quotes on Google returns only 234 results.
Portuguese pavement is hewn and flat on top like sett, shaped to fit like paving_stones, placed with tight gaps like paving_stones, and smooth like paving_stones.
Polyhedral pavement is hewn and flat on top like sett, shaped sharply like sett but irregularly like unhewn_cobblestone. Crazy paving shares the same pattern, but is very smooth and usually mapped as paving_stones.
For actual Portuguese pavement, I think it would be better represented using additional tags such as paving_pattern or paving_stones:pattern, both still in early stages of discussion (please leave your opinion there!).
As for polyhedral pavement, I’m using sett since it’s similar in smoothness and made of the same material, the only real difference is the pattern.
Well, I said that is what we call “brazilian cobblestone”, not Portuguese pavement.
Some differences that I wiev: brazilian cobblestone (what you call polyedral cobblestone) is less smoothnes that sett. Sett is binder with rigid material, brazilian cobblestone is binder with earth. Sett is made with grannite, brazilian cobblestone is made with basaltic stone. Then, the pattern is not the only real difference between they.
Brazilians technically call it polyhedral pavement, or informally by terms such as “rough stone” (pedra tosca), “hand stone” (pedra de mão) or even “brat’s foot” (pé de moleque), the last two alluding to the manual work involved in its construction and the precarious conditions of the workers (formerly slave children, nowadays low wage workers).
The definition of sett in OSM encompasses any kind of natural stone, so granite and basalt are both included.
Also, in Brazil, both setts and polyhedral/Brazilian stone pavement can have the gaps filled or unfilled according to this Brazilian civil engineering reference. In the South of Brazil, the most common variety is the one with masonry (rigid binder) filling.
The smoothness of both regular setts and polyhedral/Brazilian stone pavement, with gaps filled or unfilled, is highly variable and it is best to represent it using the smoothness tag, which is further surface detailing recommended on the surface tag article.
Hello, please ask me to reconsider this type of surfaces, as explained by Carlos Brys.
Users wishing to indicate the type of surface in the Province of Misiones Republic of Argentina, in a large part in the Republic of Paraguay and in part of the Republic with Brazil in specific border with Argentina and surrounding areas, does not have the means to indicate correctly the type of surface. If you wish, I can indicate images with Mapillary where you can see this type of surface, but photos have been provided that are sufficiently clear.
Thank you
I just noticed that unhewn_cobblestone has “Joined by a rigid binder, unlike pebblestone.” in a definition.
I propose to amend it to “firmly attached to a ground” - I think that it does not matter exactly how stones are attached. Sometimes stones are bound just by compacted soil/clay but stay put at their location, I think that it is not necessary to invent a new surface value for that.