Unbounded rationality fails due to combinatorial explosion of decision spaces
Evidence
As was quickly discovered, the ‘brute force computing’ tactic, as the unbounded rationality assumption was known, was a failure because the decision space to be searched for any moderately realistic problem was of unmanageable proportions. For example, even contemporary machines take an inordinate amount of time to examine all possible two-move sequences in the opening stages of a chess game, because the number of branches in the relevant decision tree undergoes a combinatorial explosion. Not surprisingly, this tactic was quickly dropped, and much effort has subsequently been devoted to the development of more sophisticated and realistic alternatives. If we are to understood the development and maintenance of cooperative and trusting relations, similar unrealistic assumptions will have to be abandoned too. — Diego Gambetta, Trust - Making and Breaking Cooperative…, 737
Categorization
- QUESTION: The chess-opening combinatorics claim is a nice concrete anchor for the broader argument that full-rationality models collapse under computational cost.
- XREF: Connects to the decision-theory foundation of Milgrom’s law merchant and game-theoretic treatments of cooperation. — connects to: Rational preferences operate over whole states, not single proxies; Iterated distillation and amplification can build judgment beyond our own — Brian Christian, The Alignment Problem, 1069; Rule-based machine ethics produce unintended emergent consequences — , The Braided Reading EPUB: 5-Year…, 1591 (+1 more)
- corroborated: 2 independent passages support this claim — Diego Gambetta, Trust - Making and Breaking Cooperative…, 737
Broader Connections
- Iterated distillation and amplification can build judgment beyond our own
#responds-to(0.82) — Brian Christian, The Alignment Problem, 1069 - Rule-based machine ethics produce unintended emergent consequences
#exemplifies(0.75) — , The Braided Reading EPUB, 1591