University of California, Los Angeles, USA
Made to Know: Science as the Social Production of Knowledge (and its Limits)
In this talk, I develop a view of science as the social production of collective knowledge by a complex adaptive system. Using data from millions of scientific papers, I illustrate that scientists’ research choices are shaped by the tension between tradition and innovation, which generates a distributed algorithm for directing scientists’ collective attention. I then show how this distributed algorithm leads to more (and less) efficient collective discovery. Such distributed algorithms are “programmed“ by scientific institutions. To clarify our understanding of these institutions, I describe a simple formal model of scientific problem choice and use it to show that taken-for-granted features of scientific institutions can have unexpected consequences on the pace of knowledge production. I draw together these results using ideas from computational learning theory to suggest how scientists’ strategies, though objectively adapted to social goals, can nonetheless support robust collective creation of knowledge about the natural world. In other words, the production of collective knowledge is made possible by the distinctive cultural technologies of science—which also produce limits to that same knowledge. I conclude by briefly considering the ominous possibility that the participation of (even quite modest) “machine knowers” in science could produce insurmountable limits to human understanding.