GoKawiilThe Aaronson Oracle: a decades-old game that predicts human 'random' choices
The piece traces a matching-pennies experiment back to engineer David Hagelbarger's 1950s Bell Labs machine SEER, which tracked only a handful of facts about recent rounds to guess a human opponent's next move better than chance. Claude Shannon built a similar device shortly after, crediting Hagelbarger's design and describing it as a simplified model of the same idea. The modern 'Aaronson Oracle' is a descendant of these relay-based machines, now implemented as software.
GoKawiil's interpretation of the reporting above, not reported fact.
The story illustrates that humans are much worse at generating true randomness than they assume, a finding with implications for psychology, game theory and security systems that rely on unpredictable human input. It also shows how a simple statistical memory—tracking just a few bits of recent history—can exploit these patterns, suggesting that even minimal algorithms can outperform people at tasks requiring genuine randomness. The lineage from physical relay machines to modern software demonstrates how foundational computing-era ideas persist in today's tools.
- SEER, built by David Hagelbarger at Bell Labs in the early 1950s, predicted human choices in matching pennies using only three recent-history facts.
- Claude Shannon built a related machine in 1953, explicitly crediting Hagelbarger's design as the basis for his own.
- The experiment reveals that humans struggle to act truly randomly, a weakness later formalized as the 'Aaronson Oracle'.
Source: aaronsonoracle.com — Spencer Stanton, 2026-10-14
Published there as: “The Aaronson Oracle”
Read the original report → The summary and analysis above are GoKawiil's own, written from reporting by the source above. Facts and quotes belong to the original publisher.