C. Understanding the Evolution of Science
If knowledge accumulation is an intrinsic feature of scientific advance, then the two
dynamics in science, documented above, follow naturally. As knowledge accumulates,
innovators both extend their training phases and achieve narrower expertise, changing the life-
cycle of innovation and the value of teamwork. The empirical value of the knowledge
accumulation hypothesis is partly in its generality: it can explain science-wide patterns,
ubiquitous across fields and research institutions, both in time series and in cross-section, and
aggregate a diverse range of underlying facts under a simple, unifying framework. Knowledge
accumulation can thus provide a foundational reason for shifting norms that have been perceived
within individual fields, but typically without an appreciation for their generality or an
underlying theory of change.8
III. Core Roles of Science Policy
Getting science policy right is a key role of government and, arguably, the preeminent
role of government in terms of fostering increasing economic prosperity. This claim can be
motivated by three mainstream observations in economics. First, a defining feature of the
modern age is that certain economies grow and keep growing: the United States has repeatedly
doubled its income per person since the Industrial Revolution, leading to unprecedented levels of
income and associated increases in health and longevity. Second, this growth comes largely
through technological advance - the collective impact of an enormous array of novel ideas.
Third, while markets are good at many things, markets face critical failures in the production of
new ideas.
The first two observations emphasize the importance of idea creation for economic
prosperity. The last observation suggests that government policy can play a critical and even
necessary role in encouraging ideas. Indeed, central features of ideas themselves may lead
8 Of course, when looking at any particular trend or pattern, other forces may play substantive roles. For focus, this
summary leaves aside alternative specific explanations for specific subsets of the facts. Jones (2009), Jones (2010),
Jones and Weinberg (2010), Wuchty et al. (2007), and Jones et al. (2008) discuss alternative explanations, and the
reader is pointed there for further discussion.
15