Toward an ecology of ideas: a new approach to the science of science
The study of ideas is ripe for a revolution. The growing electronic availability of scientific text allows us to identify large-scale patterns in the dynamic network of knowledge claims. These patterns, in turn, can be used to expose the biases, strategies, and social processes that give knowledge its shape and govern its evolution. I illustrate this approach by describing the first quantitative exploration of what Thomas Kuhn called the "essential tension" between safe, career-preserving "tradition" and risky, history-making "innovation." By analyzing over 6 million abstracts from PubMed, I show that scientists persistently focus on digging deeper into established knowledge (tradition) and largely ignore opportunities to make new connections (innovation). Probabilistic models of research choice allow us to quantify scientists' preferences, while careful analysis of incentives (including a unique database of thousands of prize-winners in biomedicine) suggests mechanisms for forming and reinforcing these preferences. I show that such preferences are efficient for initial exploration of a knowledge network but highly inefficient for further exploration. My approach builds on the fundamental insight that science is a complex system, a rich and evolving ecology of ideas, people, and practices. I close by outlining the implications of the emerging science of ideas for our understanding of the wider social world.