Who decides what counts as misinformation?

## The letter with no evidence

February 19, 2020. Twenty-seven scientists published a Lancet letter declaring lab-origin theories about COVID "do nothing but create fear." The letter cited no genomic evidence. Its organizer, Peter Daszak, ran the nonprofit funneling NIH grants to the Wuhan lab the letter was defending. His conflict of interest went undisclosed for fourteen months. Facebook, Twitter, and YouTube treated the letter as authoritative. Posts mentioning the lab were labeled, throttled, removed.

By May 2021, the Wall Street Journal reported three Wuhan researchers hospitalized in November 2019. The WHO called for investigation. Facebook quietly updated its policies. The hypothesis migrated from "debunked conspiracy" to "plausible origin" without anyone explaining what changed or why the suppression had been wrong.

## The alien anthropologist's question

Describe a fact-check to an alien. An organization funded by the platform whose content it evaluates reviews a claim made by a user with no access to review criteria, no right to confront the reviewer, and no appeal mechanism. The reviewer's political orientation is undisclosed on the label. The label says "false." It does not say "evaluated by an organization receiving 80 percent of its revenue from the company that published the claim." Greg Lukianoff has traced the institutional capture that made this architecture feel inevitable.

The people who built the labeling system and still believe it can be repaired are the platform governance camp. Those who watched the Christchurch shooter livestream for seventeen minutes while AI failed to flag it are the state regulators. Wikipedia's 60 million articles sit behind the open process position. And the people finding John Stuart Mill uncomfortably current are the speech liberalists.

## The machine nobody trusts

Fact-check labels change almost nobody's mind. The suppressed posts find new routes. The apparatus looks less like public protection and more like an expensive machine built to solve a problem it has not demonstrated it understands. Whether the replacement is better platforms, state regulation, community process, or something built on trust-weighted information is the question the lab-leak episode made urgent and five years have not answered.