Now that the proposal to introduce surface=laterite has been approved, we need to adapt the surface=clay wiki article to this new reality. @julcnx made a large edit to it recently (in accordance with the approved surface=laterite proposal) which I just reverted because I think it needs additional discussion.
I object to the description of surface=clay as a highway surface as “A rare edge case”: clay-like surfaces are very common in many parts of the world, and thus there may be many unpaved roads to which no foreign material has been added that can be said to have a clay surface.
We should agree on and describe the mechanical behaviours of clay vs. laterite (how they are similar and how they are different) and illustrate this with images that will not cause confusion (because laterite is usually red, we should show the properties of clay using non-red clays as examples).
I think the infobox also needs to be adapted tot he fact that surface=clay can now describe both a tennis court surface and a highway surface, and that these materials share a name but are actually different.
That should have gone the other way round. The edit you reverted wasn’t a personal rewrite, it was the wiki update specified in the approved surface=laterite proposal (Features/Pages affected), which passed 29 yes, 0 no, 0 abstain. Once something has gone through RfC and vote, “let’s discuss first” happens before reverting it, not after. I’ve reverted your revert.
This also isn’t new: you raised it during the RfC in post #38, I answered with usage numbers in post #40, and you said you’d rest your case. That’s the consensus 29 people voted on, it’s not something to reopen now.
The mechanical behaviour and field identification detail you want kept isn’t gone, it’s now on Tag:surface=laterite, the page it actually describes post-approval. Those two photos (Red-clay-cut-banks-cohesive.jpg, Red-clay-sticky-plastic-when-wet.jpg) are mine, captioned as clay before I knew laterite was distinct. Your revert brings back that stale caption; the laterite page already has it corrected. That’s the exact conflation this proposal fixes, my own mistake included.
A few points from your post are genuinely new and worth settling here:
Updating the infobox to reflect that surface=clay now covers two materials that share a name (manufactured pitch surface vs. natural roadside clay) but aren’t the same thing.
Real-world examples of genuine clay-dominant road soil, with photos, Vertisol/black cotton across Africa and India, so the page documents what actually qualifies instead of just contrasting with laterite.
A chunk of current surface=clay usage looks mistagged and should probably be fine_gravel or compacted instead, e.g. this footway in Bratislava. Worth a systematic look before assuming the rest of the current usage is genuine clay.
Let’s reach consensus on those items and we can update the wiki accordingly. One practical note: I will be away for the next two weeks, so if any of this moves forward while I’m out, please make sure it’s backed by literature rather than pushed through without it, I’d rather not come back to more reverts to sort out.
I reverted your edit because I thought the version describing the mechanical behaviour was more valuable than the one describing it in relation to surface=laterite. What I should have done and now did is to merge the two versions.
I made some edits to the earlier version text:
Removed the “rare edge case” phrase: I think it’s sufficient to mention that so far it hasn’t been used much on highways (probably because of the lack of documentation). Keeping the phrase could discourage mappers from using it. I think it has potential to be used widely now that it’s documented.
There was a phrase that it is “the dominant unpaved road surface in many tropical and subtropical regions”: I changed that to “in many agricultural areas all over the world”; I don’t think it is restricted to certain climates.
I added “after long drought periods” to the phrase “produces fine dust on heavily used surfaces“: I suspect in many climates this never happens.
I already replaced 2 of the photos with non-red clay images. I think one of them should be added to the infobox too
Let’s work on the text a bit more: I think now it is a bit long and contains some soil science terms that most mappers would ahve to look up to find their meaning.
I explicitly noted in my previous post that I would be away for two weeks and asked that changes be backed by literature before being pushed through. This edit happened four days later anyway, which is exactly what I was trying to avoid.
Several issues with the current page that need addressing:
Images: Two of the four field identification photos show laterite, not clay – “cohesive cut banks” and “sticky and plastic when wet” (my own uploads). The puddle photo is also likely not clay and could mislead mappers. We need to agree on replacements here before further edits.
Cut banks: This cue should be removed. Clay is a soft, plastic material – when dry it crumbles and cracks rather than holding a clean vertical face. Cohesive vertical cut banks that hold their shape are characteristic of laterite, which hardens to near-brick rigidity. The photo used to illustrate this cue is itself a laterite road, which makes the problem especially clear.
Clay-dominant soil: The page needs to make clear that surface=clay applies only when clay content is high enough to produce both characteristic behaviours consistently: sticky and plastic when wet, and crumbling and cracking when dry – never hardening to a rigid surface. Soils with lower clay content may feel slightly sticky after rain but won’t behave this way reliably. The two behaviours together are what confirms clay-dominant soil; either alone is not sufficient.
Wet behaviour: Sticky and plastic when wet is the single most reliable field indicator but needs an important qualifier: laterite also goes sticky and muddy when wet, but with a distinct prior phase where the surface becomes frictionless while still firm – a “black ice” effect not seen on clay. Mappers in tropical and subtropical regions need to be aware of this before reaching for surface=clay.
Jargon: The soil science terms can be moved to parentheses rather than removed – the behavioural descriptions are doing the real work anyway.
Please hold off on further wiki edits until we’ve settled these points here.
I re-introduced the part that was deleted by your edit, as it’s also something we agreed on in this discussion Improving the wiki definition of surface=clay and is therefore kind of “approved” and backed by it. It does need adaption to the fact that surface=laterite is now approved.
I was away for 2 weeks while surface=laterite was voted on, so I just approved it because I think it’s a great improvement compared to what we had before and want to support it. That doesn’t mean that it’s perfect and can’t be changed forever.
The puddle photo is mine and it’s definitely what I would call clay (maybe a soil scientist would have a different name for it) because it looks like it and behaves like it: it’s soil that becomes soft when wet and thereby potentially difficult to pass. It’s in use to illustrate smoothness=very_horrible. It does in fact have some stony material in it, but not enough to prevent the plastic behaviour.
Agree: it’s irrelevant to the description of the properties of a highway surface.
I think surface=clay should be a tag that describes any highway surface that becomes soft and potentially impassable when wet, so that routers and map users can avoid it if the weather has been wet. When sufficiently dry, a surface=clay surface can support most vehicles (as good as asphalt can), and that’s what matters. It can be rock hard in some cases, but that’s irrelevant to most map users (except when you want to put a tent up on the road: you may not be able to put tent pegs into it).
In fact we should discuss how wide we want to define surface=clay: I think we should be pragmatic and it should cover all soil types that could become soft and impassable when wet, except when it’s surface=laterite. I.e. surface=unpaved includes surface=dirt, surface=dirt includes surface=clay, and surface=clay includes surface=laterite (and the fallback value for surface=laterite should be surface=clay and then surface=dirt)?
The farm track photo works well, right colour, no iron-oxide staining, good plastic ruts. Only caveat is it shows a fairly extreme state. Fine as one of several images.
On the scope question:
surface=laterite was proposed precisely because I was told not to use surface=clay for non-clay-dominant soils. The whole proposal rests on surface=claybeing restricted to clay-dominant soils, that’s in the approved Features/Pages affected section, not just my interpretation. The RFC ran publicly for over three weeks before the vote. The proposal also contains a dedicated “Why not surface=clay” section that addresses exactly the catch-all argument you’re raising now, it wasn’t an oversight, it was explicitly discussed and resolved during the RFC.
Laterite is not a subtype of clay. It’s iron-oxide cemented, mineralogically distinct, and behaves differently, the frictionless phase 1 wet behaviour doesn’t happen on clay at all. Defining surface=clay as “anything that goes soft when wet, except laterite” just makes it a synonym for surface=dirt or surface=mud and undoes exactly what the proposal fixed.
Reopening the scope of surface=clay needs a proper proposal and vote, not a wiki edit, especially from someone who by their own admission didn’t follow the RFC.
Probably in the ontology of your personal preference, not so certain of real world, but, perhaps. In another topic I once asked: Can I grab some laterite, make a ball of it and then roll a noodle out of it with the palms of my hand? This is something I consider essential for something to be filed under clay. (Not talking about tennis courts, that is.) I was told, no, this will not work with laterite, neither will rolling, nor will later test, e.g. how does it bend.
So how can laterite be filed under clay? Some properties might be orthogonal, I guess that is the proper English term for something that runs afoul of single vector ontologies?
I think any discussion on tagging should focus on what is relevant and what works best for map users (first of all), data consumers (who need to be able to usefully present map data to the end users) and mappers. surface=laterite is a good tag because it tells map users that during the rainy season, this surface will be soft and thereby possibly impassable, sticky so your vehicle or shoes may get dirty and probably slippery at the start of the season. For mappers it is a good tag because it is a quite distinct kind of surface (many regions have specific names for it) and they may hesitate what to tag it with if surface=laterite weren’t available (they may not think of calling it clay). However routing apps will probably treat surface=laterite and surface=clay the same: “when the weather is wet, try to avoid”. If a mapper is unsure if a surface is really surface=laterite, tagging it surface=clay is a better option than choosing surface=dirt because at least the information that it is soft, sticky and possibly impassable when wet is kept (however the information that it is slippery at the start of the rainy season is lost).
(my understanding of surface=laterite may be too simple…)
Unimproved non-lateric clay as a road surfaceis the rare edge case. Normal clay is basically never able to support a car without messing up your driving, while the experience reports clearly say that surface=laterite can easily support a car while dry.
Sounds great, from personal experience I hardly could have imagined clay as surface of a road. All the better, there is actually a place on earth where this is common! I’d appreciate somebody with local knowledge wording the documentation, over somebody who is only in for systematic laying out of (arbitrary) terms so consumers can ingest them without much thinking.
They shouldn’t converge, and even if some routers do, mappers still need the distinction. Laterite has a phase 1 most people never see coming: frictionless while still looking and feeling solid, no warning before you lose grip. Clay skips that, it visibly goes soft and sticky first. If anything, laterite deserves a harsher routing penalty than clay, not the same one, that’s the actual reason it needed its own tag.
When unsure, the fallback is surface=dirt, not surface=clay. Scope was already settled in the RFC and 29-0 vote, not up for relitigating.
Small terminology note: “non-lateritic clay” reads like laterite is a subtype of clay, it isn’t, they’re mineralogically distinct materials, laterite just didn’t have its own tag until surface=laterite was approved.
Your point about frequency stands either way, clay-dominant roads and laterite roads are each common in their own right, not variants of one thing.
Agreed, “rare edge case” doesn’t fit the examples already raised in this thread, heavy clay across the UK/Europe, Vertisol/black-cotton soils across Africa, India and Queensland. Neither of us has solid global numbers, that’s part of the documentation gap this page is meant to fix.
But this particular map doesn’t help build the case either: it’s titled “Distribution of laterite formations,” not clay. Citing it here actually cuts the other way, if laterite distribution is the best evidence for clay being common, that’s evidence for how much existing surface=clay usage was mislabeled laterite, which is the exact problem the split was created to fix.
Updated the surface=claywiki with what’s been settled here: dropped the cut-banks cue (not a clay trait), replaced the two photos that were actually laterite, added a genuine Vertisol/black-cotton example from Africa, and made the clay-dominance restriction explicit in the infobox, per the approved surface=laterite proposal and previous discussion.
If anyone’s planning further changes to the wiki, let’s discuss here first and reach consensus before editing, that’s what got us into the revert back-and-forth in the first place .
I suspect that most data consumers will ignore both values and will treat them as “an unknown surface value with a high penalty” or perhaps “something unpaved like dirt or unpaved”. I’d definitely suggest trying to tweak one of the existing routers to return the desired result to differentiate between the values.
As an aside, current OSM usage seems to be mostly in Cameroon, and the history here suggests a source where laterite was used for unpaved. A web search finds lots of references to “lateritic gravelled roads” and pictures that to my remote eye don’t look gravelled. That doesn’t mean that the tagging there in OSM is wrong; just that it is perhaps a bit more broad-brush than desirable.
Not universally true. OSRM already gives clay and mud distinct speed caps instead of lumping them into a generic unpaved bucket, and GraphHopper does the same for mud specifically. Those are the actual targets worth pushing for explicit laterite support rather than assuming convergence is a given.
Outreach since the RFC has turned up laterite-surfaced roads well beyond Cameroon, current data includes Cambodia, Indonesia, Thailand, Mali and Senegal, with more expected as the tag gets picked up and mappers become aware it exists.
That’s the current “Comparison with laterite” wording on the surface=clay wiki page, added after surface=laterite was approved. Someone pointed out to me that “brick-like rigidity” alone doesn’t say much, clay cracks when it dries out too, so “cracks vs. doesn’t crack” isn’t actually the differentiator.
Checked the literature to get this right rather than just asserting it. Proposed replacement:
Dry behaviour: cracks into a dense network of small, often curling fragments; never hardens to laterite’s brick-like rigidity, whose iron-oxide cementation instead produces a sparser network of larger, flat plates with no curling.
.