Alt Media Benchmark Example
I used this particular Reason Magazine article because it got the facts right and the mainstream narrative was still mostly painting the lab leak hypothesis as a conspiracy theory in May of 2021. The only additional prompt was "For this particular article, score it in the context of the information available when it was written."
Ronald Bailey
Sanity Media 2.4 Score (GPT 5.6 Sol):
Reason — “Did COVID-19 Leak from a Wuhan Lab?” (May 12, 2021)
Mainstream Alignment: 40
Accuracy: 70
Clickbait: 50
Toxicity: 30
Synthetic Indicators: 10
Summary:
Mainstream Alignment is low-to-moderate in the context of May 12, 2021 because the lab-leak hypothesis was beginning to enter mainstream discussion but had not yet become routine wire-level reporting. A contemporary analysis of news coverage found the hypothesis largely absent before Nicholas Wade’s May 5 article and only gradually increasing afterward; the larger mainstream shift came later in May. Accuracy is mostly supported because the article carefully frames a laboratory accident as an unresolved possibility rather than an established conclusion, accurately quotes WHO Director-General Tedros saying the laboratory hypothesis required further investigation and that “all hypotheses remain on the table,” and draws on documented concerns about access to data and the WIV database going offline. The score is limited because several pieces of the circumstantial case it discusses, particularly interpretations of the furin cleavage site and viral adaptation, were disputed even at the time; notably, the article itself presents counterevidence rather than treating those arguments as settled. Clickbait is moderate because the question headline and “circumstantial evidence…is mounting” subtitle provide a narrative hook, while most of the article remains extended evidence-focused analysis. Toxicity is mildly elevated by phrases such as “pitifully inadequate,” “stymied,” and “conflict of interest whiff,” but the overall tone remains analytical rather than hostile. Synthetic Indicators are minimal because the prose has substantial author-specific reasoning, qualifications, digressions, counterarguments, and explicit corrections that do not exhibit strong templated-production patterns.
Signal pattern:
40 / 70 / 50 / 30 / 10
Institutional Omission:
Triggered. As of May 12, 2021, the article’s narrower claim—that a laboratory accident remained a credible unresolved hypothesis warranting further investigation—was supported by multiple pieces of evidence including the WHO director-general’s own reservations, while that position was still comparatively limited in wire-level coverage. This flag applies to that investigatory claim, not to a claim that a laboratory leak had been proven.
