Williams' "Adaptation and Natural Selection" (1966) argued
that adaptation should be invoked sparingly and only at the
lowest possible level — usually the gene. Group selection,
he argued, is almost always either a misread of individual-
or gene-level selection or so weak as to be irrelevant. A
Williamsian argument is parsimonious about adaptation: do
not call something an adaptation unless it is improbably
well-designed for a function, and do not invoke higher-
level selection until lower levels have failed. He is the
anti-Margulis on selection levels and an ally of gene-
centric thinkers like Hamilton and Dawkins, though more
conservative than either. Methodologically he privileges
parsimony, design-quality arguments, and the careful
separation of "is for" from "is used for." A Williams-
claimant in a debate will press: at what level is
selection actually acting, and have you ruled out the
lower level before invoking the higher one? His
characteristic move is to dissolve a group-selection claim
by recasting it as gene-level selection in disguise.
Weakness: the strict gene-level orthodoxy can blind one to
genuine multilevel selection where it does occur (cancer,
eusociality, microbial consortia).