Misinformation and historical revisionism are widespread on some non-English editions of Wikipedia, such as Japanese, which is the most visited after English
During World War II, Unit 731 of the Japanese military undertook horrific medical experimentation in Manchukuo (Northeast China). Tweets: @edfrench , @yumikosatomtbc , @slate , @yumikosatomtbc , @yumikosatomtbc , @cambskeptics , and @martinsfp Tweets: @edfrench : Out of curiosity I compared the English and Arabic wikipedia pages on Mecca- spot the difference! https://en.wikipedia.org/... https://translate.google.com/ ... https://twitter.com/... @yumikosatomtbc : “The notion that ‘the more people continue to edit, the better it gets’ is uniquely American. This doesn't happen to the Japanese version because it's isolated.” @slate : Japanese Wikipedia articles about World War II contain troubling false statements. https://slate.com/... @yumikosatomtbc : Wikipedia has an immeasurable impact on the minds of Japanese readers who stumble upon pages filled with historical revisionism. One frustrated person left a comment: “Japanese Wikipedia is becoming a convenient propaganda tool for those who want to control public opinions.” @yumikosatomtbc : I have a new article in Slate's Future Tense about historical revisionism on Japanese #Wikipedia It's indicative of a wider misinformation and disinformation problem on non-English editions of Wikipedia. Please read: https://slate.com/... Cambridge Skeptics / @cambskeptics : On Wikipedia's 20th birthday it received praise for the accuracy of its articles. But coverage focused on the English version, while there is a misinformation and disinformation problem on non-English editions, reports @YumikoSatoMTBC for @Slate. https://slate.com/... Martin Sfp Bryant / @martinsfp : English-language Wikipedia is largely a haven of accuracy and neutrality. That can't be said of the Japanese version. https://slate.com/...
Context & Ripple Effects
Slate's report lands against a backdrop where English Wikipedia has been actively hardening itself — locking down new-editor access on its main election page and running extended fights over what counts as legitimate framing on contested science pages, including the coronavirus lab-leak debate. The finding here is that those defenses are an English-language phenomenon: on Japanese Wikipedia — the second most visited edition — World War II articles carry revisionist accounts of atrocities like Unit 731, and comparisons such as the one @cambskeptics ran between the English and Arabic pages on Mecca show the same divergence.
The deeper problem the piece surfaces is structural: as @yumikosatomtbc puts it, the assumption that 'the more people continue to edit, the better it gets' is uniquely American — it depends on the large, adversarial editor pool English has, which smaller language editions lack. Earlier governance friction like the year-long ban of a prolific administrator showed how contested even English-language enforcement is.
First-order effects
- Readers of Japanese Wikipedia — a top-two edition by traffic — get revisionist WWII history presented as encyclopedia fact, while Arabic readers see materially different religious-topic pages than English users.
- Wikimedia's volunteer-governance model faces a credibility problem precisely where it claims universality: the self-correcting crowd dynamic does not transfer to editions with thin editor pools.
Second-order effects
- Pressure builds on the Wikimedia Foundation and cross-language administrators to port English-style measures (new-editor restrictions, protected pages, active admin review) to other editions — measures already proven on COVID misinfo and the election page.
- Editor communities in other languages face the same legitimacy disputes seen in the English lab-lealk and administrator-ban fights, but without the depth of experienced reviewers to resolve them.
Third-order effects
- If the pattern holds, Wikipedia stops functioning as a single trusted reference and becomes a set of divergent per-language corpora whose accuracy tracks each edition's editor demographics — with downstream consequences for anything built on top, from search snippets to AI-generated summaries like the kind Wikimedia later piloted and paused after backlash (the AI-summary experiment).
- Trust in 'Wikipedia' as a brand may increasingly depend on specifying the language edition, forcing educators, journalists, and platforms to treat multilingual sourcing as a verification problem rather than a solved one.
The trend: Wikipedia's reliability is diverging by language edition, with the English core's editorial defenses and adversarial editor culture failing to extend to smaller non-English communities.