Editor concentration and reverts on one contested Wikipedia page — an exploratory pilot
This pilot borrows its question — how is a contested history negotiated by its editors? — from the computational study of Wikipedia “edit wars” by Yasseri, Sumi, Rung, Kornai & Kertész (Dynamics of Conflicts in Wikipedia, PLoS ONE, 2012), which built a controversy measure from reverts and editor concentration — the same two signals used here. (Taha Yasseri is now at the Oxford Internet Institute; the 2012 paper predates that move.) I pulled the last 500 revisions of the “2024 Bangladesh quota reform movement” page via the MediaWiki API and computed two simple descriptive indicators: editor concentration (what share of edits come from the most active few) and revert rate (what share of edits were undone). Note the scope: these are behavioural counts on one page in one window. They say nothing about who was right, and Wikipedia editing is human and social — there is no ranking algorithm involved here.
Of 147 unique editors in this window, the 5 most active account for 40.2% of the last 500 edits. On its own, that is unremarkable: editor activity on Wikipedia is almost always highly skewed, with a small core doing most of the work. Without a baseline from comparable pages, a 40% top-5 share is not evidence of “gatekeeping” — it is roughly what you would expect.
The revert rate is 4.2% — fewer than 1 edit in 20 was undone. That is low. It points to a relatively orderly editing process on this page during this window, not the “edit war” the topic’s reputation might lead you to expect. The honest headline is a mild one: on this page, in this window, concentration was typical and open conflict was limited.
Usernames are public on Wikipedia, but naming individuals under a “gatekeeper” label would single out real people on the strength of a raw edit count. For a descriptive pilot that is neither necessary nor fair, so editors are shown here by rank only.
| Editor (by activity) | Total Edits (Last 500) |
|---|---|
| Editor 1 | 73 |
| Editor 2 | 51 |
| Editor 3 | 42 |
| Editor 4 | 22 |
| Editor 5 | 13 |
This is one page, one 500-revision window, with no comparison baseline — so “typical” and “low” above are informed reads, not benchmarked claims. “Reverts” are detected by a simple heuristic that undercounts subtler reversions; the analysis counts editing behaviour and never inspects what was contested or whether a change improved the article. A stronger version would compare several contested pages against calm ones, use Wikipedia’s own revert tags, and sample across time. Treat this as a method probe, not a verdict on this page or its editors.
The raw JSON extraction, CSV data, and analysis scripts are published on the project GitHub repository for transparency. Because a live wiki page keeps changing, rerunning the code returns the current last-500 revisions, not the exact set captured here; the stored data is the archived snapshot.
View Code & Dataset on GitHub →