Innovation and Population 255
and a most harmonious and comprehensive language; so that every circumstance among that people
seemed to favour the rise of the arts and sciences.... Europe is at present a copy, at large, of what
Greece was formerly a pattern in miniature. (Ibid.: 65)
And he fi nishes by telling us what the Chinese got wrong:
In China, there seems to be a pretty considerable stock of politeness and science, which, in the course
of so many centuries, might naturally be expected to ripen into something more perfect and fi nished,
than what has yet arisen from them.... But China is one vast empire, speaking one language,
governed by one law, and sympathising in the same manners. The authority of any teacher, such as
Confucius, was propagated easily from one corner of the empire to the other. None had courage to
resist the torrent of popular opinion. (Ibid.: 66)
14.10 Population Thinking and Evolution
We should not be disheartened by our exploration of the Needham Question. It is true that
on some occasions the conjectures we reach through population thinking could just as well
be reached by informal refl ection. But work by the likes of Richerson and Boyd (2005),
Nelson and Winter (1982), and Philip Kitcher (1993), in relation to other aspects of innova-
tion, suggests that the intuitive sketches of population explanations we arrive at informally
(such as Hume’s explanation for the success of innovation in Europe) can be tested and
corrected by more formal populational modeling. This is the lesson of Nelson and Winter’s
skeptical evaluation of Friedman’s defense of the assumption of profi t-maximization. It is
also the lesson of many of their more complex models, which try to ascertain, for example,
under what circumstances imitation and innovation can coexist as research and develop-
ment strategies, and under what circumstances imitators will drive out more effortful
innovators, in a population of competing fi rms (Nelson and Winter 1982: pt. V).
It also seems plausible that formal modeling of this sort can throw up new hypotheses
for empirical testing. Boyd and Richerson draw usefully on the statistical thinking that,
although largely absent from Darwin’s work, was central to the establishment of selection
as an important factor in evolutionary change during the modern synthesis. Consider, for
example, Fisher’s remarks on particulate inheritance. He claimed that if blending inheri-
tance were the dominant mode, then selection could only lead to permanent evolutionary
change if mutation rates were very high. Otherwise the population would always tend to
regress to the mean, regardless of how well individuals with advantageous mutations might
do in virtue of them. Observed mutation rates are, as a matter of fact, comparatively low.
Hence in the organic world, Fisher’s population thinking shows that for selection to be
effi cacious, inheritance must be particulate. Now this does not show (as memeticists may
be inclined to assume) that for selection to lead to permanent changes to artifacts, the
resources that underpin technological inheritance must be particulate also; rather it prompts
Boyd and Richerson to ask whether mutation rates are high enough in this domain for