Draft v3 · · What changed
Gary Marcus
Marcus's contrarian record holds because he has paired his architecture critique with operational specifics his rivals will not match — settlement-priced bets none have accepted, peer-reviewed evidence on every healthcare claim, named reactions to every frontier release, and a selective-credit pattern that sorts technical artifacts from capability narrative.
Positions
Pattern matching, not reasoning
Marcus produced named, dated reactions to seven AI systems, from DeepSeek R1 to Claude Code — sorting pure-scale launches as "the wall" and tool-integrated launches as "vindication" — under a single diagnostic phrase…
The bet no one will take
Marcus has put a settlement price on the AGI-by-2027 forecast. Amodei, to whom he offered it, never responded, by Marcus's account; the only taker of either wager is a former OpenAI policy researcher.
Where Marcus gives credit, and where he doesn't
Marcus runs a falsifiable two-axis rule — credit the technical artifact when it integrates symbolic tooling, dismiss the same lab's capability narrative — and the pattern holds across Anthropic, DeepMind, OpenAI, and…
Frequently wrong, never in doubt
Marcus's deployment-caution slogan, "frequently wrong, never in doubt," is anchored by four peer-reviewed studies, including a 52% emergency under-triage finding and a 60-point gap between standalone-model accuracy…
·Issue 1 · The Scaling Bet
deslop.media editorial. Issue 1, Draft v3, first published October 10, 2026. Reporting current through October 2026. Use the flags on this page to report an error or a disagreement. Changes to the issue are listed on the changes page.
Part of Issue 1 · The Scaling Bet