File talk:London Mappers so far in 2017.png
voiding the collective contributions of the many individual users
Unfortunately this has the bad effect of voiding the collective contributions of the many individual users. I would have prefered a diagram that shows not individual users but users by amounts of changes (number of changesets, or total number of creation of OSM objects (nodes+ways+relations)+GPS tracks+notes? I think that even notes are valuable input to the project, even if the reported modifications to do are performed by other people
As well I would have separated contributions made by known bots (most imports of external data sources, not fully checked and that frequently need many individual adjustments such as missing tags, disambiguation when the sources are imprecise, deduplication/merges, normalisation of tags, translations, updated data from more recent/more precise sources or from local surveys)...
- Well you'd have to ask Andy Allan about that. I think he said he did eliminate bot activity. -- Harry Wood (talk) 11:10, 16 May 2017 (UTC)
- What would be interesting is to track how many contributors are needed for filling half of the data currently available in OSM.
- One problem of the measures is that every edit in a single object tracks only the last contributor, even if there was a change of a single property (adding a node to a way, moving a node for higher precision, adding/removing/changing a single tag, fixing a typo in a tag value. To avoid this bias, it would be necessary to count all contributors found in each object's history (but discard those edits that were reverted later).
- This would require a complex analysis of histories and a lot of data (most often not available on all data mirors but only in the main OSM database), and this can be computed on relatively small areas (like here only in London, but which London was used is unspecified: City of London, Inner London, Great London, or some bounding box?).
- These measurements are not comparable to very different areas (notably not in those where open data is not so developed and where import bots cannot do a lot). And it would be useful to compare to other areas (including in UK: the London area is a bit exceptional because mot useful datas were already part of open data that were easily and massively imported and are still updated by bots). The situation would be very different in poor areas (notably those involded in HOT projects, even if in these areas there are also important contributors that have done a majority of the work).
- Those massive edits are the easiest ones, this is not something for which OSM will be most useful, where we'll need many more small contributors that will fix few things on few points spread over a much larger area where their contributions will not be valued with this type of diagram. I wonder if it is possible to group users by total number of edited objects. It is desirable to go to a situation where this diagram would turn almost completely grey (due to delimiting borders), except a thin yellow border on the top and left sides. But QA tools will keep this border thick because there will still be lot of automated edits to fix the many indivisual contributions, even if there were an insignificant amount of unwanted edits (illegal imports or spammed edits that were redacted are already excluded, they are normally not visible in the history): there will still be various disagreements between contributors, and frequent errors (e.g. typos in tag names or values), or deprecated but still proposed in some editors. — Verdy_p (talk) 14:04, 16 May 2017 (UTC)