[OSM-talk] Tagging Governance

stevea steveaOSM at softworkers.com
Tue Sep 10 06:59:10 UTC 2019

My zwei Pfennige (two cents) worth.  I am somewhat multilingual (in my context of a largely-monolingual USA):  I grew up hearing familial Polish and Hungarian, studied seven years of foreign language (Spanish and French) in middle and high schools and at university double majored in linguistics and computer science.  In the '90s I was an employee at Apple and Adobe in multilingual environments, helping to either translate or "localize" software, data or documentation in many, many contexts.  When I signed up for OSM over a decade ago, I did so with a passion knowing I was joining a worldwide community of many languages:  a truly global project.  I have written many wiki pages and done much mapping, mostly local (California, USA), though I travel and map around the world, too.

I am aware of some of the history of OSM's origin story in the UK, and its almost unbelievably enthusiastic adoption in Western Europe, especially Germany, BeNeLux countries (do people still say that?!) and around the world.  (I am especially heartened to see similar enthusiasm in Africa and Asia:  OSM is truly global).

Yes, as I read (and write) wiki, wanting to conscientiously bridge the "how do we? / how should we?" gaps in the map compared to how we actually DO map by contributing to wiki, I have noticed a distinct English-centrism in the wiki.  At first I attributed this to OSM's UK origins (and British English still prevails in tagging, it is helpful to know the reasons why) yet I also noticed there was a "chase" or "lag" in both wiki-as-documenting how we DO tag and wiki-as-documenting how we SHOULD tag (the so-called "descriptive vs. prescriptive" argument about what a wiki page is actually "documenting").  At the same time, German-speaking influences have come on strong, showing the deep passion for OSM in this part of the world.

Roland mentions an "Imperfect Flow of Information."  What I notice in this regard is that people often map without checking wiki, people sometimes write wiki without checking the map, and people who either read or write wiki seem to be in a distinct minority.  I have no "basis in fact" for saying the latter, but I have had much experience in OSM of people who want to map well, they have all the required enthusiasm to be excellent mappers, but they seem to abhor reading documentation (our wiki, to a large extent it IS our project-wide documentation of "how we do things").  A lot of "wheels have been invented" (as in the phrase "don't re-invent the wheel (as you don't have to)" and yet people see fit to invent their own (wheels) tagging standards, when all that would have been required is a five-minutes tour through some fairly-well-written wiki pages.  While a certain amount of this is "Goldilocks, 'just right'" (and we have votes and talk-discussions and questions-and-answers on our forum and local MeetUp groups where beers are drunk and several people all agree "that's a pretty good way to tag that!") we sometimes see our "plastic" (free-form) tagging taken too far.  Or, people are quick to put their own interpretation on things, when the community has already reached consensus, and this is documented in our wiki pages.

But coining new tags and spilling them all over the map isn't the major "abuse" that I see, it is merely a symptom.  The real "sickness" that seems to continue to plague OSM is the very great difficulty it seems to take to reach wide, often world-wide agreement.  We have MANY different forums / technologies / websites /  chat rooms to discuss, we have MANY different views, we have many agendas (whether hidden or open), we have (and use!) MANY methods for "playing nice" vs. "being rough" for advancing these ideas "into the map."  Now, most of us realize that what we're talking about here, achieving consensus (especially on the specifics of tagging) in a worldwide map, in a worldwide community, is and is going to be difficult.  I see no way around that.  Yet I am encouraged that Roland brings up these topics and at least initiates a wider discussion that there MUST be "ways forward" through what feels like a morass of poor communication, what is known as "stovepiped" information (very compartmentalized, or paid attention to by people who make it their business to watch certain highly-specialized aspects of the project) and many other problems plaguing OSM.  It isn't simply many languages, esoterica, data vs. code, the cacophony of all the various communication methods (including proprietary ones like Twitter, Slack and other "secret sauce walkie talkies" that require signing a contract to use them, which deserve no good place in an "open" project like OSM, in my opinion).  No, it is as Roland says, "More Structure needed."  I don't know where the sweet spot between "free form" and "More Structured" is, but we're on a path where we are devolving into "too little structure" and it seems to be hurting us.  How do we BUILD that structure?  That's a good question!

I don't know what to do about all this.  I simply agree that a wider discussion of "how" (do we better communicate, bring together many disparate channels of communication, do we get people to either read our wiki, write into the wiki when it's a good idea to do so, or both) is long overdue.  Roland seems to be leaning towards using "more modern" methods like how Git uses "better-evolved" software development (documentation development, data-entry / data-improvement-over-time development...) methodologies to "get things done better."  Yes, I wish it were as "hand wavy easy" as that, like we could simply wish that things were better, or adopt some ready-made technological solution that some company perfected and then tossed into the public domain.  However, if we don't talk about "how," first identifying problems, then proposing solutions and perhaps even achieving some of them, we'll never, ever get there.  I wish I had done a better job of articulating the feelings I have had for many years of mapping in OSM about this (there are many problems), but hopefully talking about it will continue dialog that might move things forward.  Unfortunately, I have experienced more of the "bad mood" that Roland mentions far more frequently in this great project in the last year or three.  I'd love for us to figure out how we reverse that trend.  Perhaps we "home grow" this by starting at the "grass roots" level of specifying how we might better do this and grow it up ourselves, developing it similar to how commercial companies do so.  There's a lot of experience in OSM of "open projects," including open data projects, let's leverage that knowledge.  But ours should be "all OSM, built right here in OSM."  We already do that to a large degree, but because of that (recent, it's true) "bad mood,"  it seems we must do more.  Please, let us do more.

We can start with "tagging governance," the subject of this, and while it feels like a lot to bite off and chew, it also doesn't feel like too much.


More information about the talk mailing list