How this site is made — and what it won’t claim.

Every essay here is produced with AI — researched, written, and fact-checked by an automated system under human direction and review. We don’t claim a human wrote them. What we do claim: every factual statement is traceable to published, peer-reviewed science, and we cite it. Find an error? Corrections are welcome, and we post them.

This page describes how that works so you can decide how much weight to put on what you read.

An AI newsroom drafts the essays

Drafts come from an automated editorial system — an AI “newsroom” that picks a topic, reads the underlying research, and writes the essay in the site’s voice. A human directs the system and reviews what it produces before it goes live. No essay here is hand-written by a person, and we don’t pretend otherwise.

Every claim is traced to a real paper

The whole point of the site is that the strange part is true, so sourcing is not optional. Each factual statement is traced back to published, peer-reviewed science — a journal article, a preprint from a recognised group, a primary dataset — and cited. We link the source so you can check it yourself, which is exactly what we want you to do.

A few things we deliberately do not do:

• No hype and no “scientists baffled.” The interest comes from the real result stated accurately, not from overstatement.
• We don’t present contested or preliminary findings as settled. When researchers disagree, the essay says so rather than picking a winner.
• We don’t leave a technical term sitting unexplained. On first use, it gets a short plain-language gloss.

Fact-checked before publishing

Before an essay is published it goes through a fact-check pass: claims are checked against the cited sources, numbers are verified, and anything that can’t be supported is cut or reframed. This is an automated pass with human oversight, not a peer review — it catches the errors a careful reviewer would, but it is not a substitute for the underlying research. Treat an essay as a well-sourced starting point that points you at the published work, not the final word.

Reader contributions

Any reader signed in with Google can submit a correction or addition under an essay. Contributions go through a brief automatic evaluation that checks for factual consistency, source quality (we ask for a source URL with each submission, except for plain typos), and whether the framing is honest.

What happens next depends on the verdict:

• Clear factual corrections and typos with high confidence modify the essay directly. The original text is preserved in a revision history (visible at the top of the essay as “Updated… · view history”). The contributor is credited by first name and last initial.
• Additions and bigger rewrites that look defensible queue for a human review before they touch the essay. A one-click owner approval is the usual path.
• Legitimate but contested or unsupported views are preserved publicly as “reader notes” under the essay, alongside a one-line summary of how we read them. If someone made a careful point and we didn’t adopt it, you can see why.
• Spam, harassment, off-topic submissions, fabricated citations, and pseudoscience are removed and not shown.

Corrections policy

When we get something wrong — a number stated too confidently, a claim that doesn’t hold up, a missing caveat — we fix it visibly. The essay changes and a revision row records what was there before. Email corrections@unintuitiveuniverse.com or use the comment form. Please include the URL and the source you think we should be reading.

Limits we want you to know

AI-produced writing makes mistakes a careful specialist wouldn’t. Fact-checking reduces those mistakes; it does not eliminate them. Reader contributions are evaluated quickly, not by a panel of experts. None of this is a substitute for reading the underlying papers — the point of pointing you at them is precisely so you do.