Source
Simon (1973), LIT-1634, §§2–2.4 and §3.6 (pp. 182–187, 197); read in NOTE-1232.
What was actually shown
The criteria are relational. Six conditions are listed (p. 183): a mechanizable solution test; a problem space holding all states; legal and “considerable” moves; all acquirable knowledge representable in problem spaces; exact correspondence with the laws of the external world; and all of it “in the strong sense that the basic processes postulated require only practicable amounts of computation”. The criteria “generally express a relation between characteristics of a problem domain, on the one hand, and the characteristics and power of an implicit or explicit problem solving mechanism, on the other” (p. 182), and their vagueness reflects “the continuum of degrees of definiteness” (p. 183). Footnote 1 (p. 185) fixes the standard: “effective or practicable solvability”, not undecidability.
The standard cases fail. Theorem proving fails condition 4 “for all problem solvers” once the solver may use metalinguistic, model-space or analogical knowledge (pp. 184–185). Chess replaces “best move” with an approximate evaluation, and its real moves are irrevocable, so playing a game “involves continually redefining what the problem is … a WSP in the small … an ISP in the large” (p. 186). This could have come out otherwise: the two were chosen as the “extreme” and “best” examples of well structured problems (pp. 184–185).
The generalization. “Definiteness of problem structure is largely an illusion that arises when we systematically confound the idealized problem that is presented to an idealized (and unlimitedly powerful) problem solver with the actual problem” (p. 186). “It is not exaggerating much to say that there are no WSPs, only ISPs that have been formalized for problem solvers” (p. 186). The shift from computability in principle to practicable computability is what moves the boundary (p. 197).
What this does not say
- That structuring is the real problem solving. Simon calls that allegation refuted (by the Newell–Simon argument that a sequence of tasks from one environment would otherwise pose only one problem) and grants only that “much problem solving effort is directed at structuring problems” (p. 187).
- That a problem is ill structured for every solver. The relativity cuts both ways: a standard design on the architect’s shelf makes house design not ill structured (p. 187), and the “no WSPs” sentence is hedged and about problems “presented … by the world” (p. 186).
- That ill-structuredness needs new mechanisms. That is the paper’s further and weaker thesis (THEORY-1084).
- Anything about questions or inquiry. The paper’s terms are problems, problem spaces and solvers; carrying it to what an agent can ask is the record’s own step (CLAIM-475).