i picked this up together with some other books on sociology of science at the behest of liam bright in his #noheroes reading list. simonton's project in this book is to develop a probabilistic model for scientific creativity that incorporates attributes of individual scientists and social context, with some light mechanistic undertones. more specifically, he proposes that each scientist is exposed to a set of scientific items (say, ideas), and that scientific output results from the combining these ideas together until a scientist finds a novel and interesting combination, which are taken to be rare. as the basis for something like an agent-based model, i would find this compelling, but simonton never actually develops the model formally; indeed the whole book amounts to little more than an extended verbal thought experiment together with some underreported regressions. to simonton a process is poisson distributed if the outcome is rare, citation counts are direct measurements of the creativity of a scientific product, and causal identification is unnecessary, which is pretty bold given that he's essentially trying to make a bunch of complex moderation claims. this style of research is straight out of the 60s and 70s and it was sloppy and fun when price was doing it, but it's pretty hard to justify for work in the early aughts