Monthly trend in number of edits

Chart shows absolute numbers (y axis scale shows number of edits relative (%) to busiest month)

Monthly trend in revert activity, relative to edit volume

Chart shows ratio reverts:edits per class of editors (registered or anonymous users or bots).
Note: often peaks in revert ratio in early years are not so significant as in later years:
low absolute edit counts easily led to high fluctuations in revert ratio.

Revert ratio is ratio reverts:edits per class of editors (registered or anonymous users or bots).
Only reverts detected by md5 matching are considered here, not (partial) manual reverts only detectable by edit comment.

Reverts is less than reverted edits! Sometimes several edits are undone with one revert.
Reverts broken down by number of reverted edits: 1 edit:77%, 2 edits:11%, 3 edits:5%, 4 edits:1%, 5 edits:1%, >5 edits:4%

When several revisions were reverted in one action, the oldest reverted revision determines editor class (and coloring):
is this a reverted user edit (reg/anon) or a reverted bot edit?

Monthly trend in revert activity towards anonymous edits

Chart shows absolute numbers (y axis scale shows number of edits relative (%) to busiest month)
Note: for many (not all) larger projects the not reverted edits show less seasonal fluctuation than the reverted edits

'All anonymous edits' is same red line from first chart, scaled up vertically
Put succinctly: red - black = blue + brown + green
Only reverts detected by md5 matching are considered here, not (partial) manual reverts only detectable by edit comment.

Percentages are an approximation, as two simplifying assumptions have been made:
1) Anonymous edits are reverted only once (in other words edit wars on anon edits are not taken into account).
2) On each revert only one anonymous edit was undone. A frequency count of multiple revision reverts shows this is a simplification:
Reverts broken down by number of reverted edits: 1 edit:77%, 2 edits:11%, 3 edits:5%, 4 edits:1%, 5 edits:1%, >5 edits:4%

Distribution of reverts

Top Rankings

Generated on Saturday July 31, 2010 from recent database dump files.
Data processed up to Wednesday June 30, 2010
Please note that the lengthy dump process (many weeks) means a delay in publishing these statistics is always to be expected.

Script version:2.5
Author:Erik Zachte (Web site)
Mail:ezachte@### (no spam: ### = wikimedia.org)
For documentation see meta
Scripts: scripts.zip

Charts rendered with R

All data and images on this page are in the public domain.