We've been investigating revert rates of anti-vandalism bots and their communities (T341857), but are grappling with the issue that not all edits that take a page back to an earlier state are vandalism reverts. One way to interrogate this further would be to look at the ORES damaging and good faith scores, to limit our analysis to edits which have a high damaging and/or low good faith score.

Unfortunately we can't do this for most wikis as their anti-vandalism bots use these scores to make their decision, so the analysis would be somewhat circular. ClueBot NG and Salebot, however, do not use ORES scores, so we can investigate for these two bots and see what conclusions we draw.

Questions

For ClueBot NG on the English Wikipedia and Salebot on the French and Portuguese Wikipedias:

• What is the average daily edit revert rate, and bot revert rate, for edits with high ORES damaging/bad faith scores?