[Tagging] Feature Proposal - RFC - Discouraging the use of deprecated schemes

stevea steveaOSM at softworkers.com
Wed Mar 24 07:51:10 UTC 2021

On Mar 23, 2021, at 6:06 PM, mail at marcos-martinez.net wrote:
> Fully agree. As I said: Complex rules for a complex world. If there is consistency based on community consensus and documented on the Wiki - perfect. This is all what I understand Sören wants to reinforce: to strongly discourage tagging that is not voted or agreed and documented. Of course world wide valid tagging schemes are desirable but when not possible local schemes need to be implemented - and followed.

Thank you for your thoughtful reply, Marcos.  We discuss a spectrum here.  One end is “perfect” community consensus, documentation and actual usage in OSM 100% reflective of that, with zero divergence to other tagging practices.  We can all certainly agree on that, as we do!  The other end of the spectrum is the cacophony of say, landuse=forest / natural=wood, where there are (at least) 6 different semantics for these tagging pairs, one of which is “as originally designed” (Approach 1), some of which are more prevalent than others, and one or two which are in a distinct minority.  In the middle of the spectrum is where things are interesting and we “fully agree,” such as regional differences among highway=* tagging to harmonize to a particular set of local preferences and “standards” for highways in that region.  Those can be messy, yes, but an “allowable messy” because they are a “necessary messy."

> There's nothing wrong with inventing new tags as new tagging needs will always arise. The emphasis lies on NEW. You can't find a suitable tag for what you need? Invent one. Ideally you should inform the community via the mailing list and create a proposal that hopefully will later be approved. From that moment on it should become "official" and of course anybody can challenge the vote and start a new proposal to modify the previous one. It seems obvious to me that any editing of the same reality on the ground but with different tags should be STRONGLY discouraged. I still can't think of any possible advantage of tagging against the wiki.

Yes, for NEW, we (generally, ideally) follow the process you outline.  It is somewhat messy:  “rule by committee” (the membership of which is always different and constantly fluid) and “consensus must emerge” (yet it doesn’t always).  But this process creaks and grinds ahead and (largely) serves us well.  It is a process which has emerged “by consensus to have consensus” and thus, will be difficult to change — though it could be changed, it would take time.

Let’s be careful here, especially as we describe such “thou shalts” and “thou shalt nots" at a rather high level.  We DO want to “tag against the wiki” in the limited scope of “well, this is a region where the tagging expressed by the wiki needs some modification to fit reality here, so we do that” (e.g. local highways have their own logical mapping onto highway=*).  In that limited scope, we (carefully) say that’s OK, if we develop a scheme that is sane and we document that (on a country-level wiki, for example).  As for tagging against the wiki in other circumstances, I have an inclination to agree with you that we shouldn’t do that, so I say “yes, let’s discourage that.”  I do not say “let’s STRONGLY discourage that” because I have the experience (tagging in OSM, Proposal development in our community and on our wiki, multilingualism in several natural languages, academic training in linguistics, half a century of life showing me that new utterances are constantly happening, knowledge of cartography that all sorts of cartographic concepts exist are not now well captured in our map…) that to be that restrictive (here at a high level) too easily breaks the sort of flexibility that is sometimes needed to develop new ideas and tagging schemes.  I would much rather have a bit of ambiguity and cacophony (a cost, but it happens) than be so restrictive by saying “let’s STRONGLY discourage that” that we stifle the ability to make new utterances.  "Better to sometimes smear a little than not be allowed to paint at all.”

Feel free to disagree with me, I certainly don’t want to come across as a know-it-all.  I do think that a bit of flexibility is better than being too restrictive.

> In case you don't want to start the whole process (which is perfectly fine as well) there will eventually be competing undocumented schemes around and these kind of inconsistencies are acceptable as many contributors don't want to be involved in OSM bureaucracy. Now, AS SOON AS somebody else makes a proposal and this gets approved, we should make sure only the "offical" one is used.

I’m not so sure we should.  There IS the admonishment that we not engage in (semi-)automated edits:  why is that?  It is because of the knowledge that there really is such a thing as the natural attrition I spoke of earlier, and the tendency (whether strong or not is debatable) for the older, or less-robust, or undocumented in wiki, or were-never-proposed flavors to get used less and less.  (And conversely, those which are NOT these things, which follow our processes, WILL get used, more and more).  So I’m not sure I agree with your “we should make sure” part of the “official” tag gets used.

So, while what you call ‘bureaucracy’ might feel like that, it’s really an “officialization of process” that has naturally evolved. It, like writing into the wiki, is as much a process in OSM as is reading FROM the wiki.  Yes, some Contributors shy away from this (as “too much ‘bureaucracy’”), that’s simply how things are.  But what isn’t required is to further insist that they or others “should make sure” the official tag gets used.  Note I said “isn’t required” rather than “won’t work.”  There can be some “policing” activity, say, informing an errant user that a better tagging scheme not only exists, but is documented, so please use it (instead).  But this isn’t required.  Eventually, this will “sink in” (to the consciousness) of errant Contributors, sometimes over a longer-term, yes, to where (eventually) the wrongly-tagged will become the no-longer-tagged.  This really happens.  (A personal example I gave is farm -> farmland).

> 	• Such "alignment" doesn't happen by fiat.  Well, not very often, anyway.  It happens as usage gets noticed, picked up by others, used more widely and emerges as the accepted standard.  This is always happening in all living languages, tagging in OSM is no exception.
> OSM in many ways is/should be different from languages.

What I mean (and believe I said?) is that TAGGING in OSM is like a language:  syntax (tagging) representing semantics (what we mean to map, what we mean is in the real world), expressed as semiotics (rendering).

> As languages, OSM in some way tries to "communicate" the reality of the world.

Yes, here we agree.  A rendering is the representation of the real world through the language of tagging on geometric primitives (data structures).

> Nevertheless, language inherently has the purpose of allowing individuals to express themselves as such in a creative way. But we don't need book writing or intimate poetry skills, nor songs or speeches.

Again, we agree:  natural languages need these, an artificial language (like OSM’s syntax, tagging) does not need to write poetry or sing.  (Unless that is its express purpose:  there are AIs that write music, for example).

> In language terms, synonyms and rhetoric figures are desirable, in OSM we should aim for ALWAYS repeating the same tag for the same element.

We SHOULD aim for this in the crafted, organic language of OSM, but it seems we agree that sometimes, along that rather wide spectrum, sometimes we do not.  Let's be grateful for those “100% agreed perfect” cases where we are on that first part of the spectrum.  Let's ALLOW for those smeary middle parts (because the world is messy and we really must).  Let's avoid, but acknowledge that circumstances will happen where we get cacophony and ambiguity.  Let’s allow the latter, as a “safety valve” of creativity.  When it happens, and it does and will, let’s gently clean it up with our polite policing (informing the errant author:  here is a documented, community-agreed-upon method:  please use it instead) and let’s avoid heavy-handed tactics like STRONGLY discouraging these.  People are human, we make mistakes (including “oops, coined something new without looking at wiki for existing tags — now that you’ve called that to my attention, I’ll redact my errant ways and 'do it by the book,’ thanks”).  And, OSM must allow this flexibility to exist, BEFORE wikis are written, consensus has emerged and Proposals are hashed out (not in that order).  STRONGLY discouraging pours a lot of cold water on that flexibility.  “Simply” discouraging allows the human component to remain human, civil (“please go read the wiki…”) and avoid being heavy-handed.  This is what Frederick means by “weakly” discouraging.  That’s about all any of us as individuals can do in a project this large and diverse.

> What we need is the creativity and combined effort to best reflect the world in a database in the most universally understandable way. OSM should be breathing and alive by quickly including new concepts and needs but not in duplicating and triplicating tagging schemes.

What we’re discussing is the proper balance to keep a “weighted spectrum” PROPERLY weighted:  not too heavy on sloppy and with a cacophony of triplicate tagging schemes when one is proper and documented, but not too discouraging of the flexibility to coin new tags.  That isn’t an easy balance to achieve for everybody, but once again, discussing it like this (maybe bringing it out into the open like this) is often the first step to achieving it.

I don’t think I’m saying anything fantastically new, but I may be saying some things in explicit ways that either aren’t ever discussed, or are only seldom discussed.

Thank you for your patience to all who continue to extend it to read these lengthy discussions.

More information about the Tagging mailing list