In my ~15 years of this project, I’ve seen many “version X has simply become version Y” instances, they are nearly too large to enumerate. Many (like wood / forest, a flavor of landcover…) still “rage on.” My reaction has been to be an “old-timer” and “get used to changes.” I do not complain, and up until now, have said very little to nothing about this — I simply absorb a fair amount of re-adjustment to new schemas. In short, this is what is expected of OSM, not only as a contributor to its data (literally changing how we tag compared to past taggings) but to “suck it up” when such changes occur. I attribute this to the plastic nature of our map, though at some point, things can become so “goopy and liquid” that the map might become unusable with many ambiguities of its haphazard development.
Without getting too lost in the weeds, many times I’d say these changes result in “the new is better than the old.” There truly is a reality that we sometimes don’t exactly get things right the first time, and a 1.0 (schema, way of doing things) becoming a 1.1 (or 2.0) with significant improvements is welcome and refreshing. Yet, other times, there are seemingly intractable problems, often rooted in deeply semantic linguistic and cultural differences in both the concepts we attempt to map and the tags we choose to do so. (Wood / forest, landcover / landuse, the underlying meaning of what is and what could be, are perhaps foremost example of these). A few years ago, boundary=forest_compartment and related tags (which took three ballots to succeed, IIRC) was a small victory. Yet, we still have more work to do.
There has been friction. I suspect there will continue to be friction, though I hope we can keep that minimized to the extent it induces rancor instead of progress. We might develop some sort of more-formal versioning approach, I think this shows a lot of promise, as it can be both a technical and a social solution to what often freezes OSM into brittle, sometimes outright broken data models and process which seems difficult or intractable to improve.
“One-off” proposals (like this one) are simply bandages on a longer-term wound that doesn’t seem to be healing very well in OSM — the chaotic nature of our anarchic process. We would hugely benefit from a worldwide, project-wide methodology to make fundamental changes to rather ambiguous tagging that has evolved, and skidded into fairly ugly ambiguity in some cases. A “1A” first step would be to identify (major) problems (like landcover), a “1B” would be suggested methodologies on how we go about fixing these. It’s possible we leverage existing methodologies (Proposals / voting, wiki tuning, mechanical edits…), it’s possible we wholesale re-invent, it’s possible we develop some new versioning scheme, it’s possible we find a blend that works. But this “technical debt” continues to pile up, and longer-term, it hurts OSM both presently and into our future.
I don’t have ready-made solutions to offer, but I do roll up my sleeves in earnest as I attempt to identify that this is truly difficult (social, technical) work, and it needs doing. I don’t want to rush this, I would much rather we “get it right.” Landcover is only the tip of this iceberg. Thank you for reading.
