This is a relatively simple test that RcFilters are having an effect on reversion behavior.
- Create samples of namespace 0 edits by week before and after ORES is deployed for each Wiki that adopted ores
- Scrape the threshholds that are set for each Wiki for each week from Special:ORESModels
- For each wiki plot the probability of reversion as a function of ores scores. Look for jumps at the threshholds when we look after RcFilters changes are released, but no jumps before RcFilters were released.
- Try to measure the jumps by modeling the relationship between reversion and ores score and the threshholds.