Wikipedia’s harmful untruths

There was a time when Wikipedia felt like a miracle: a spontaneous, self-governing lexicon arising from the turbid chaos of the web. No editors kept gates, no gilded towers barred entrance, no one had power to impose a worldview, it was all done by thousands of neutral volunteers harvesting and serving the world’s knowledge, onto a digital platter. And their sheer numbers – it was hoped – would preserve accuracy and objectivity. The same way a crowd has more wisdom than the individual.

However, as the years pass, that illusion of noble neutrality has shattered. And a clear and maybe terminal tilt to the left has revealed itself. As Wiki-founder Larry Sanger lamented in a recent interview with Tucker Carlson, the Wikipedia he wanted has long gone. What we have now is, in particular areas, worse than nothing.

I can personally vouch for the way Wikipedia slants left. About 15 years ago, while journalistically researching some horrible crimes in northern England, I came across the case of young Charlene Downes. She was a tragic victim of the vast Muslim rape and grooming scandal in the UK, this time in Blackpool.

Yet one thing marked out her case, amongst the tens of thousands of victims of this crime; a crime which – in terms of state failure – is arguably Britain’s Chernobyl. Charlene Downes was not just raped and tortured, she was murdered, and then – according to lurid witness statements never proved nor disproved – possibly eaten. Literally put into fast-food kebabs. I wish I was joking.

I used Wikipedia to research her appalling fate. But then one day I found the entry on her case had been taken down – disappeared like an Argentinian dissident under the junta. I dug into the entrails of this editorial decision, and discovered that activist editors had deleted it because it “lacked notability.” How could the rape, torture, murder and alleged cannibalization of a young British girl not be “notable”? Especially when much more “everyday” murders had their own entries?

The answer is, I believe, that Woke Wiki editors didn’t like any focus on the ethnic identity of the perpetrators. In the end the page reappeared, and it stands today. But the attempt by Wikipedia editors to conceal the scale and horror of the overall Muslim rape gang scandal goes on. A few months ago the entire entry devoted to this monumental horror got renamed as “Muslim grooming gang moral panic”, like it never existed. Like it was just some fever dream of fascists, rather than the greatest black mark in Britain’s modern history. Again, the original entry – after much angry slanging – has been reinstated, but the fact this attempted erasure even happened is telling.

Deeper data throws light on the problem. A 2024 Manhattan Institute report, Is Wikipedia Politically Biased?, used sentiment analysis on over 1,600 politically charged terms and found a mild-to-moderate tendency for right-of-centre figures to be depicted with more negative language than left-of-centre figures. Another study, Polarization and Reliability of News Sources in Wikipedia, examined nearly 30 million citations and concluded that Wikipedia exhibits a systematic liberal polarization in its choice of news sources – even after controlling for their factual reliability.

The initial gatekeeping of sources is another fierce battleground. A 2025 report by the conservative Media Research Center found that none of the right-leaning media outlets studied earned a “generally reliable” classification in Wikipedia’s source assessments, while 84 percent of left-leaning outlets did. Critics argue these reliability judgments are decided via opaque community consensus among liberal-left editors.

This is of course self-reinforcing. If a conservative editor adds a citation from, say, the New York Post, it is swiftly questioned or deleted; if a progressive editor adds one from Vox, it often slides through. Over time, this creates a lopsided citational ecosystem that makes neutrality impossible. Even if you have the best intentions.

It is worth reflecting on why this tilt persists. Partly it is those demographics: anecdotally, the volunteer army of Wikipedians tends to be young, tech-savvy, university-educated, comfortable with progressive assumptions. Also, this is an example of institutional capture: once a small group of committed editors has embedded a set of norms, they can and do police the boundaries with ferocious diligence.

The consequences are, as we see, grave. First, contested pages simply cannot be trusted. Second, even when a page appears reasonable, the selection of sources ensures that readers encounter a progressive consensus rather than the full spectrum of debate. Third, the instability of controversial pages means that “truth” can swing from week to week, depending on the latest edit war. Fourth, because Wikipedia is mined relentlessly by search engines and artificial intelligence, its bias is then amplified and propagated far beyond its own digital horizon. Elon Musk has complained that his AI, Grok, skews to the left however he tweaks the machine. This must be partly because it is trained on Wikipedia.

It is a melancholy thing: because the Wiki project once seemed like maybe the finest creation of the digital age. An encyclopedia of everything, freely available, self-governing, universal. But noble dreams have a way of dissolving when they meet reality. Wikipedia is just such a case. It has fallen prey to the same “long march through the institutions” that reshaped the universities, many of the arts, much of the media, and the NGOs. It now too often reflects the worldview of its most committed editors, not the messy plurality of reality. And that makes it – how sad, how dispiriting – almost unusable as a source of facts on anything contentious.

The lesson is clear. Use Wikipedia for lists of monarchs, for summaries of chlorophyll or Caravaggio, or deep dives on the moons of Jupiter. For that it remains absolutely marvellous, still an internet miracle. But if you want to understand a political dispute, a culture war, a controversy, you must treat Wikipedia not as the final word, but as a cleverly illustrated propaganda pamphlet. A mirror crack’d. It is also a warning of how even the grandest experiments in collective truth can be bent to one side: producing, instead, mournful and harmful untruths.