---
title: "Empirical Explorations of the Logic Theory Machine"
authors:
  - "Allen Newell"
  - "J. C. Shaw"
  - "Herbert A. Simon"
year: 1957
venue: "Proceedings of the Western Joint Computer Conference"
arxiv: null
doi: "10.1145/1455567.1455605"
source: "https://doi.org/10.1145/1455567.1455605"
topics:
  - logic-theorist
  - symbolic-ai
  - newell-simon
seed_rank: 581
seed_batch: "foundations-greats-2026-08-13"
reviewed: "2026-08-13"
pool: "agents"
relevance_score: 8
lineage: foundations-of-computing
cites:
  - title: "The Logic Theory Machine: A Complex Information Processing System"
    url: "https://doi.org/10.1109/TIT.1956.1056797"
    year: 1956
    arxiv: null
    doi: "10.1109/TIT.1956.1056797"
see:
  - "580-the-logic-theory-machine-a-complex-information-processing-sy"
---

# Empirical Explorations of the Logic Theory Machine

## One-sentence takeaway

Newell, Shaw, and Simon report how Logic Theorist actually performed on *Principia* theorems — AI as an experimental science of programs.

## Why it matters here

Agent papers in this corpus should publish traces and failure modes, not only architectures. This 1957 WJCC paper is that genre's ancestor.

## Key ideas

- LT proved 38 of the first 52 theorems in *Principia* Chapter 2, with runtimes and search sizes tabulated.
- IPL implementation details (list memory, association lists) are treated as part of the scientific object.
- Heuristic adequacy is empirical: which methods fire, where the search explodes, which proofs look "human."
- The paper is the measurement companion to the 1956 system description.
- McCarthy later cites it as the prior art the Advice Taker is *not* — heuristics in code versus heuristics in sentences.

## Caveats

## Links

- DOI: [10.1145/1455567.1455605](https://doi.org/10.1145/1455567.1455605)
- URL: https://doi.org/10.1145/1455567.1455605