Michael D. Gordin is a professor of history at Princeton University. He is the author of The Pseudoscience Wars: Immanuel Velikovsky and the Birth of the Modern Fringe, due out in October from the University of Chicago Press.

The term "pseudoscience" gets thrown around quite a bit these days, most notably in debates about the dominant consensus on anthropogenic climate change. Say "pseudoscience," and immediately a bunch of doctrines leap to mind: astrology, phrenology, eugenics, ufology, and so on. Do they have anything in common? Some posit unknown forces of nature, others don't. Some are advocated by outsiders to the scientific community, while others have been backed by the elite. And the status of each can fluctuate over time. (Astrology, for example, was considered an exemplary field of natural knowledge from antiquity through the Renaissance.)

For millennia, philosophers have attempted to erect a boundary between those domains of knowledge that are legitimate and those that are anything but­—from Hippocrates' essay on "the sacred disease" (epilepsy) to editorials decrying creationism. The renowned philosopher Karl Popper coined the term "demarcation problem" to describe the quest to distinguish science from pseudoscience. He also proposed a solution. As Popper argued in a 1953 lecture, "The criterion of the scientific status of a theory is its falsifiability." In other words, if a theory articulates which empirical conditions would invalidate it, then the theory is scientific; if it doesn't, it's pseudoscience.

That seems clear enough. Unfortunately, it doesn't work. Epistemologists present several challenges to Popper's argument. First, how would you know when a theory has been falsified? Suppose you are testing a particular claim using a mass spectrometer, and you get a disagreeing result. The theory might be falsified, or your mass spectrometer could be on the fritz. Scientists do not actually troll the literature with a falsifiability detector, knocking out erroneous claims right and left. Rather, they consider their instruments, other possible explanations, alternative data sets, and so on. Rendering a theory false is a lot more complicated than Popper imagined—and thus determining what is, in principle, falsifiable is fairly muddled....