Revision activity after November 30, 2022

13 articles measured against their own 36-month baselines · computed Sep 3, 2026 · activity-v1

0 very high
2 high
0 elevated
11 typical
0 unscored

Articles about artificial intelligence, the companies and people building it, and the debates around it, measured from the public release of ChatGPT.

Activity is not proof of bias or misconduct. A sustained rise in edit activity can mean new sources, cleanup, disputes, or plain attention; this page shows the measurements, and the conclusion stays yours.

How to read this

Bands are relative to each article’s own past. Every score compares an article’s post-event editing with its own 36-month pre-event baseline. Articles are not ranked against each other; inside a band they are listed alphabetically.
Typical is a result. Most articles about most events keep changing at their usual pace. A typical band is the method working, not an absence of data.
High is not wrongdoing. The score measures how unusual the pace of editing is, not why it changed. Each row links to the full report with the exact before/after text and the official revision history.
Two gates. An article with fewer than 12 months or 20 edits of pre-event history, or one created after the event, is shown as unscored rather than given a low number.
Controls. 8 unrelated, long-stable articles are measured the same way and tagged control. They should land typical; where they do not, that is information about the method, not about them.

High revision activity

2 articles · alphabetical

OpenAI
High activityscore 69
baseline 0.3 → after the event 3.1 edits/day
Sam Altman
High activityscore 70
baseline 0.1 → after the event 1.5 edits/day

Typical revision activity

11 articles · alphabetical

Typical is a result: these articles kept changing at roughly their usual pace after the event.

Ancient Romecontrol
Typical activityscore 34
baseline 0.5 → after the event 0.7 edits/day
Artificial intelligence
Typical activityscore 13
baseline 1.4 → after the event 2.1 edits/day
Existential risk from artificial intelligence
Typical activityscore 13
baseline 0.3 → after the event 0.6 edits/day
Great Barrier Reefcontrol
Typical activityscore 0
baseline 0.1 → after the event 0.1 edits/day
Haikucontrol
Typical activityscore 0
baseline 0.3 → after the event 0.2 edits/day
Johann Sebastian Bachcontrol
Typical activityscore 7
baseline 0.4 → after the event 0.2 edits/day
Jupitercontrol
Typical activityscore 4
baseline 0.7 → after the event 0.5 edits/day
Mitochondriacontrol
Typical activityscore 0
baseline 0.5 → after the event 0.2 edits/day
Photosynthesiscontrol
Typical activityscore 6
baseline 0.1 → after the event 0.2 edits/day
Pythagorean theoremcontrol
Typical activityscore 1
baseline 0.2 → after the event 0.2 edits/day
Siri
Typical activityscore 1
baseline 0.4 → after the event 0.3 edits/day

Why these articles

The set is curated by hand as coarse groups, not chosen by score: it is meant to sample the topic’s neighborhood, including articles expected to be quiet.

Technology1 article: The methods and concepts behind large language models, from long-standing textbook topics to articles that did not exist before the release.
Companies and labs1 article: The organizations building or competing with the new systems.
People1 article: Biographies of researchers and executives named in coverage of the release.
Earlier systems and products1 article: Assistants, image models, and tools that predate the release and were re-described in its light.
Society and policy1 article: Safety, ethics, regulation, education, and copyright — the debates the release reopened.
Controls8 articles: Unrelated, long-stable articles measured with the identical method. They should land typical; they are here so the method can be checked against its own misses.

Data

Download the JSON: data/events/2022-11-30.json — every entry with its status, score, monthly edit counts, edit rates, and the exact revision analyzed. Rebuilt weekly from the public API.

Derived from public Wikipedia revision data via the Wikimedia APIs; article titles and statistics are shared under CC BY-SA 4.0. Method: how the score is computed (activity-v1).