All edits are not equal - so counting them as such can be misleading

A graph from 2007 shows an ever-increasing percentage of edits that are either reverts (reversing, say, vandalism) or the edits that were reverted. The implication is, hopefully, obvious - what we should be interested in, in terms of trends, is constructive edits - that is, edits that improve Wikipedia. If an increasing percentage of edits are vandalism, and/or fixes (reverts) for vandalism, then trend analysis that ignores this (as in, "total edits counts are roughly flat from 2007 to the present") is misleading, because it misstates whether Wikipedia is attracting and keeping editors who improve Wikipedia.

I'm fully aware that some edits are reverted as part of edit wars, or by over-zealous editors who want to protect articles, but I'd guess that this percentage hasn't increased over time. In any case, the matter is empirically testable: take samples of 100 reverted edits from (say) January 2007, January 2008, January 2009, and January 2010, and categorize them as vandalism, constructive, etc. That should only take a couple of hours.

But the larger point remains: reverts and reverted edits aren't simply minor background noise - they are a very significant (and, as of late 2007, increasing) percentage of Wikipedia editing activity, and treating them like other edits is misleading. John Broughton 23:04, 2 November 2010 (UTC)

Dear John,

Thank you for your constructive thoughts. I agree with you that not all edits are equal and that ideally we should discard such edits. I have been reading a number of papers [1][2] about how to detect reverts and I will try to add such functionality to the Editor Trends Study/Software.