The mediawiki_history dataset has a very convenient `is_reverted` field, but mediawiki_history arrives monolithically once a month. If we are trying to track revert rates closer to real time, our current best strategy is querying the API and using the [mw-reverts](https://github.com/mediawiki-utilities/python-mwreverts) package. However, this isn't very performant.
The EventBus data stream of revision creations (see: https://github.com/wikimedia/mediawiki-event-schemas/blob/master/jsonschema/mediawiki/revision/create/3.yaml) includes the revision's hash, which is the same thing that mw-reverts uses for revision detection. It would be useful to create a tool that operates directly on this data (with a to be determined [[https://meta.wikimedia.org/wiki/Research:Revert#Time_to_revert_cutoff|time to revert cutoff]]), which would probably provide much better performance.