About
hypegapcheck scores a paper on the gap between what it claims and what it actually shows — nothing more.
The score
Every scan gets a Hype Gap from −10 (underhyped — a hidden gem whose evidence outruns its claims) to +10 (overhyped — claims writing checks the evidence can't cash). A Hype score and an Evidence score are each computed on a 0–10 scale, and the gap is just the difference between them, clamped to ±10.
Arithmetic, not opinion
Three independent AI passes read the paper — one pulls out hype claims, one pulls out reported statistics, one checks whether the paper justifies its own numbers. None of them scores anything. The score itself comes from a fixed formula run over what they extracted: hype severity plus a claim-count adjustment, evidence looked up against a fixed table of effect-size bands, and the gap is just the difference. Same inputs, same score, every time — never a model's subjective judgment call.
What it reads
Paste a link from arXiv or bioRxiv, or drop in a PDF directly (up to 20MB). No account, no sign-up.
Terms and privacy: see /legal.