Multi-Factor Decision Making

Most real decisions weigh many factors at once — price, safety, size, color — drawn from different dimensions. The standard approach seeks an overall evaluation by scoring each factor, multiplying by a weight, and summing; CF argues this is mathematically broken as a general method.

The core problem is arithmetic. Factors in different dimensions are unlike terms: you cannot add 3 acres + 8 hours + 5 grams. Addition only combines like terms, so summing requires converting every factor into a shared “goodness” dimension — and those conversion weights are made up, uncheckable, and usually reverse-engineered to fit a conclusion the chooser already held. Multiplication does combine unlike terms, but yields useless multi-dimensional units (a “gram-second-meter-dollar”). So pro/con lists and weighted factor analysis fail as self-contained methods.

CF’s solution is binary factor multiplication. Convert each non-binary factor into a related yes/no question — “is the price good enough?” — which has an objective answer and loses information one-way (breakpoints mark the cutoffs). Each factor becomes a sub-goal scored 1 or 0. Multiplying binaries acts as logical and: any single 0 fails the whole option, and no number of 1s compensates. The multi-dimensional units stay meaningful — they list which sub-goals passed.

This is satisficing, not maximizing: pass/fail mappings are resilient to small data changes, against the perfectionist update-on-everything stance of Bayesian weighing. It ties to decisive criticism (a passing option is non-refuted) and to excess capacity — optimize the constraint, not every local factor.