[stblack]

I have a hard time imagining a more appropriate tool.

The best tool would offer

  * Discoverability
  * Updatability
  * Transparancy (who, specifically, is behind it)
  * Tracability
  * Time stamping 
  * Linkability, with über-stable URIs
  * Public issue tracking.
  * Documentation, including the possible crowd-sourcing thereof.

How is this not GitHub?

Edit: In this case we have identifiable and passionate individuals behind the initiative. This is far from faceless and cursory, as most data-dumps are. What's not to love here?

[yellowbkpk]

The only issue is that with huge data dumps (the buildings dataset here as GeoJSON is ~2GB uncompressed and ~1GB as shapefile) it becomes difficult to make direct pull requests against the data. Indeed they zipped the JSON file up before uploading it so it's impossible to make pull requests (I originally suggested GeoJSON because a pull request could be read by a human as opposed to a shapefile diff which could not be read).