A study suggesting papers and patents that change the course of science are becoming less dominant is prompting soul-searching — and lively debate about why, and what to do about it.