User:HighInBC/find socks by revert rate

• /source

Goal

Create a tool that reads all edits for a page and organizes editors based on what percent of their edits were reverted. This will be helpful catching chronic sock puppets on certain articles. If done right disruptive sock puppet accounts will stand out clearly from productive editors. Naturally final judgment is left to a human.

Notes

What is a revert?

For the purposes of this tool a revert is when the page is returned to the state it was in immediately prior to the edit or string of edits

• Done with checksum, not summary.

• Should catch rollback as well as varied twinkle and undo and manual reverts

Workflow

How will this work

• Using the API it will iterate from the first to the last edit for a given page or set of pages
• Consecutive edits by the same person(or IP) will be treated as a single edit
• A tally for each user will be kept as well as a log of diffs for every revert seen
• A tally for each user will be kept for every edit made
• Using the two tallies the count and revert rate will be computed and displayed in a nice sorted list.

Theory

The same person returning after a block over and over is likely trying the same old thing and getting reverted a lot. Conversely productive editors will likely have a much lower revert rate due to the value of their contributions.

Productive editors that just got there and have a high revert rate due to a low sampling will be obvious because the edit count is also shown.

This tool may not catch productive sock puppets but that is not the intended target.

This tool essentially measures the rejection rate of editors.