Behind the scenes at Rotten Tomatoes, whose Tomatometer is run by 12 “curators” who read just about every review from a pool of about 4,500 approved critics
Humans, not algorithms, determine those ubiquitous scores. Good ingredients, imperfect recipe. Tweets: @heilemann and @wiredculture Tweets: Michael Heilemann / @heilemann : I'd take a Ripe Tomatoes site which only aggregates from the top 50 reviewers. https://twitter.com/... @wiredculture : Humans, not algorithms, determine those ubiquitous scores. Good ingredients, imperfect recipe. https://www.wired.com/...
Context & Ripple Effects
Wired's backstage look lands between two moments that frame why the human layer matters: in 2019 Rotten Tomatoes moved its Audience Score to ratings from Fandango-verified ticket buyers after unverified user scores lost credibility, and three years later Vulture documented how PR firms game the Tomatometer by paying obscure "critics". The 12 curators reading nearly every review from a pool of about 4,500 approved critics are the site's answer on the professional-critic side — a manual gate where rivals like IMDb lean on crowdsourced volume.
First-order effects
- Admission to the approved-critic pool is now the scarce resource: whether a review counts toward the Tomatometer depends on 12 curators' judgment, giving them de facto power over which voices can move a score studios market on posters and trailers.
Second-order effects
- The same manual gate creates the attack surface Vulture later exposed — once approval is the currency, PR firms have an incentive to get marginal outlets into the pool, forcing Rotten Tomatoes to keep tightening vetting just as it did with verified purchases on the audience side.
Third-order effects
- If the pattern holds, review aggregation stops being a counting exercise and becomes a credibility-ranking business — human editorial gates layered over verified-identity signals — the same direction credibility-scoring projects like NewsGuard have taken for news sites.
The trend: Review aggregators are shifting from raw vote-counting toward layered trust systems — human curation, critic vetting, and purchase verification — because open aggregation is too easily gamed.