CONCEPT
Satisficing
Simon's 1956 neologism for the decision procedure bounded agents actually use — searching sequentially through alternatives and accepting the first that clears a threshold of acceptability, rather than optimizing across all options.
Satisficing is the search procedure that bounded minds perform in place of optimization. The agent faces alternatives she cannot evaluate simultaneously and evaluates them one at a time, comparing each to a threshold of acceptability rather than to every other alternative. When an alternative meets the threshold, the search terminates. The threshold is not fixed — it adjusts to the cost of continued search. In rich environments where alternatives are easy to find, the threshold rises; in sparse environments, it falls. This dynamic, which Simon formalized in 1956, connects the theory to every major technological transition in the history of tools: each reduction in the cost of generating alternatives has shifted satisficing thresholds upward, producing both better outcomes and greater cognitive demand on the decision-maker. AI represents a phase transition in this dynamic — when the cost of generating the next alternative approaches zero, the satisficing calculus enters a regime where the threshold rises faster than the bounded judgment that polices it.
