Research talk:Exploring systematic bias in ORES/Work log/2019-05-10

From Meta, a Wikimedia project coordination wiki

Friday, May 10, 2019[edit]

First investigations of how ORES affects editing activity[edit]

I used spark to build measures of the time it takes for reverted edits to be reverted and the inequality of the distribution of reverting work for each wiki. Typically, I observe that for the wikis that adopted ORES, identity reverts tended to increase throuought the entire period I'm observing and that the time it takes to revert them increases as well.

I think this is probably not the fault of introducing ORES's to RCFilters. Instead, I think it is more likely that wikis seek to adopt these tools when they are experiencing an increasing difficulting handeling damaging edits.


Plots from a first attempt to estimate the effect of ORES on quality control efficiency