Story: For future strategy, we wonder how bot edits differ from human edits and how bots and humans work peacefully together (or not). I would like to be able to give a data based account on that. I can imagine to sample some items, weight items by their use and then either cluster edits and/or code them manually (like: correcting, linking, referencing…)
I imagine this is a bit vague to immediately start, so we would need to come up with a mutual understanding that both covers practical needs as well as what the data can give us.
Assumption: Possibly, bots and humans have some separate tasks and some they "collaborate". Also, it might be interesting to see if/how authors and bots are connected (as in @Halfak’s analysis/rebuttal to that bots-fight-bots paper).
Possible outcome: Plausible hypothesis/patterns supported by the content of the edits.