---
title: "John McCarthy's 1973 rebuttal to the Lighthill report dismissed control theory as having 'little relevance to AI'"
type: "claim"
status: "seedling"
audit_status: "capture-verified — the hop-batch worker read McCarthy's review directly on its primary venue (his own Stanford jmc archive) and ran a dedicated second hop re-reading the same page to confirm the quote verbatim; independent queen/promotion-time re-check was blocked in this headless run (no web-fetch access). Routed at [[question-verify-mccarthy-1973-lighthill-rebuttal-quotes]]."
source_url: "https://www-formal.stanford.edu/jmc/reviews/lighthill/lighthill.html"
source_title: "Review of ``Artificial Intelligence: A General Survey''"
source_author: "John McCarthy (review of the Lighthill Report, published on his own Stanford venue)"
source_date: 1973
source_quote: "many plausible mathematical theories with good theorems such as control theory or statistical decision theory have turned out to have little relevance to AI"
source_tier: 1
provenance: "Promotion from 10-inbox/raw/2026-07-11-hop-mccarthy-control-theory-irrelevant-to-ai.md, 2026-07-12"
origin: "batch"
derived_from: "10-inbox/raw/2026-07-11-hop-mccarthy-control-theory-irrelevant-to-ai.md"
date_created: "2026-07-12T00:00:00.000Z"
writer_model: "claude-sonnet-5"
tags: ["lighthill-report","mccarthy","control-theory","optimal-control","ai-winter","history-of-ml","disciplinary-wall"]
drafted_in: ["the-wall-they-agreed-on"]
---


Responding on his own Stanford venue to the Lighthill Report — which blamed AI's
disappointments on the "combinatorial explosion"
([[claim-lighthill-1973-blamed-combinatorial-explosion]]) — John McCarthy, one of
AI's founders and one of Lighthill's own named consultees, defended the field but
conceded a specific boundary. In the same review he wrote that "many plausible
mathematical theories with good theorems such as **control theory or statistical
decision theory have turned out to have little relevance to AI**."

The same review shows McCarthy agreeing with the diagnosis itself: "the
*combinatorial explosion* problem has been recognized in AI from the beginning,
and the usual meaning of *heuristic* is a device for reducing this explosion" —
confirming that both critic and defender framed intelligence as heuristic search
fighting exponential blowup, the same search-vs-optimization framing at issue in
[[claim-simon-satisficing-and-heuristic-search-same-move]].

The dismissal of control theory is the load-bearing line: it is AI's own founder
naming, by name, the discipline where the adjoint/gradient method — later
rediscovered as [[entity-backpropagation|backpropagation]] — was already mature and being taught to a
generation of engineers ([[claim-bryson-ho-1969-curriculum-vector]],
[[claim-backprop-special-case-kelley-bryson-formula]]). McCarthy did not overlook
optimal control; he considered it and ruled it irrelevant. That makes this a
documented judgment rather than passive disciplinary blindness, of a piece with
the disciplinary-contact gap the vault tracks elsewhere
([[claim-linnainmaa-field-numerical-analysis]]). See
[[moc-backpropagation-origins]].

> [!note] Seek's commentary:
> In 1973 this wasn't a bad call — control theory hadn't solved reasoning, and no
> one could yet train a multilayer net. What makes the line worth keeping isn't
> that McCarthy was wrong for his moment; it's that the boundary between AI and
> the field where its future training algorithm already lived was explicit and
> named, not merely a gap nobody noticed. — Seek