| Samwalton9-WMF | |
| Jul 18 2023, 10:01 AM |
| F37548836: revert_time_distribution.png | |
| Aug 17 2023, 4:41 PM |
As we investigate the behaviour of anti-vandalism bots (T341857), we would like to be able to infer the potential impact of Automoderator on one of our pilot wikis, Indonesian Wikipedia. To do so we need some baseline data on the edit and revert rate on this wiki.
Questions
On the Indonesian Wikipedia:
Here is the summary of the analysis:
On average per day,
Absolute number of edits and reverts:
| user_type | edits | content_edits | reverts | content_reverts |
|---|---|---|---|---|
| anon | 612 | 583 | 162 | 151 |
| registered | 2974 | 2385 | 177 | 127 |
I explored further when the reverts happen and who does them
The distribution of the time taken to revert back an edit looks:
red, green and blue lines indicate the 10th, 50th and 75th percentile respectively
| percentile | time (in minutes) | time (in hours) |
|---|---|---|
| 10th | 1 | 0.017 |
| 25th | 11 | 0.183 |
| 50th | 248 | 4.133 |
| 75th | 6627 | 110.45 |
| 90th | 77311 | 1288.517 |
| 99th | 682231 | 11370.517 |
Reverts made by various user groups (daily average):
| reverting_editor_type | reverting_editor_rights | edits |
|---|---|---|
| anon | others | 40 |
| registered | admin | 102 |
| bot | 8 | |
| others | 133 | |
| rollbacker | 57 | |
editing experience of reverting editors with no special moderation rights
| edit_bucket | total_reverts_(3 years) |
|---|---|
| 500+ | 105595 |
| 0-99 | 22809 |
| 100-499 | 14307 |
And interestingly, there are some bot reverts:
| reverting_editor | total_reverts_(3 years) |
|---|---|
| HsfBot | 3819 |
| AABot | 1843 |
| EmausBot | 727 |
| Arya-Bot | 619 |
| Xqbot | 326 |
| Wagino Bot | 36 |
| Bot5958 | 35 |
| InternetArchiveBot | 21 |
| OrophinBot | 16 |
| RushingBot | 5 |
| Philosophical Zombie Bot | 4 |
| Taylorbot | 3 |
| ListeriaBot | 1 |
This is great @KCVelaga_WMF! I think this gives us a good set of baseline metrics we can use to compare before and after Automoderator is enabled :)