See how a Wikipedia article changed around an event.

HistoryRewritten measures an article’s edit activity against its own 36-month baseline — edit rate, text churn, reverts, talk-page debate, editor concentration — and shows the exact before/after changes between the last pre-event revision and today, with links to the official revision history.

Activity is not proof of bias or misconduct. An edit spike can mean new sources, cleanup, disputes, or plain attention.

Try: ZionismTheodor HerzlBalfour DeclarationAncient Rome (control)

Add to Chrome

See the activity chip on the Wikipedia article you’re reading. The extension runs only on Wikipedia article pages and sends nothing but an article’s title, language edition, and event date to the HistoryRewritten service — only when the article matches a tracked event with alerts on, or when you open the popup. No browsing history, no accounts, no tracking.

Add to Chromev0.2.1 · open source

Explore by event

Curated sets of articles, each measured against its own 36-month baseline from the same event date, with unrelated control articles measured the same way. Activity is not proof of bias or misconduct.

October 7, 2023 — Hamas attack on Israel

  • 0 very high
  • 3 high
  • 15 elevated
  • 58 typical
  • 3 unscored
  • High activityscore 72
    Hamas: monthly edits, 48 months around the event
    baseline 0.8after the event 5.0 edits/day
  • Typical activityscore 18
    1948 Arab–Israeli War: monthly edits, 48 months around the event
    baseline 0.2after the event 0.5 edits/day
  • Typical activityscore 12
    Ancient Rome: monthly edits, 48 months around the event
    baseline 0.6after the event 0.6 edits/day

Browse all 79

November 30, 2022 — ChatGPT release

  • 0 very high
  • 2 high
  • 0 elevated
  • 11 typical
  • 0 unscored
  • High activityscore 69
    OpenAI: monthly edits, 48 months around the event
    baseline 0.3after the event 3.1 edits/day
  • Typical activityscore 13
    Artificial intelligence: monthly edits, 48 months around the event
    baseline 1.4after the event 2.1 edits/day
  • Typical activityscore 34
    Ancient Rome: monthly edits, 48 months around the event
    baseline 0.5after the event 0.7 edits/day

Browse all 13

February 24, 2022 — Russian invasion of Ukraine

  • 0 very high
  • 0 high
  • 0 elevated
  • 24 typical
  • 0 unscored
  • Typical activityscore 1
    2014 Russian annexation of Crimea: monthly edits, 48 months around the event
    baseline 0.6after the event 1.1 edits/day
  • Typical activityscore 31
    Ancient Rome: monthly edits, 48 months around the event
    baseline 0.5after the event 0.6 edits/day

Browse all 24

March 11, 2020 — COVID-19 declared a pandemic

  • 0 very high
  • 0 high
  • 2 elevated
  • 11 typical
  • 0 unscored
  • Typical activityscore 12
    Coronavirus: monthly edits, 48 months around the event
    baseline 0.7after the event 1.3 edits/day
  • Typical activityscore 17
    Ancient Rome: monthly edits, 48 months around the event
    baseline 0.4after the event 0.2 edits/day

Browse all 13

How the analysis works

A deterministic activity score

Monthly edit rates are compared to the article’s own 36-month pre-event baseline using median/MAD robust statistics — no model involved. Every component and weight is returned with the score, so any result can be reconstructed by hand.

Attention & contest signals

Pageview-normalized editing separates “more readers” from “disproportionate editing.” Revert-like edits, talk-page activity, and aggregate editor concentration flag contested periods — without profiling any individual editor.

Exact before/after evidence

On request, the last pre-event revision is compared against a pinned current revision. Changes are aligned deterministically; a tool-less language model only summarizes the changed excerpts and must cite evidence IDs, which are validated before display.

What this is not

High activity is a screening signal, not proof of bias, coordination, or misconduct. Articles with thin baselines say “insufficient baseline” instead of pretending to a verdict. Every finding links to the official Wikipedia diff so you can check the source.