Compare https://beta.musicbrainz.org/statistics/timeline/main#+-c-core-entities+c-edit-information+-editor.votelastweek+-editor.activelastweek+-vote.perweek+-edit.perweek+-edit.open and http://mbsandbox.org/~nikki/editorstats.html

The former is rather depressing but the latter (assuming my query is correct) suggests it's really just reflecting changes in fingerprinting - people needed to submit TRMs more often than PUIDs and now with AcoustIDs we don't submit them as edits at all.

The query I used there is: select date, count as c from (select distinct editor, date_trunc('day', open_time) as date from edit where type not in (77, 113)) as q group by date order by date;