This is a book I've long made good use of, even prior to actually reading the whole thing all the way through, I would often lean on Suvin's definition of science fiction, a definition that (like the best ones surely) is more about what science fiction does than what it looks like.

According to Suvin, “SF is, then, a literary genre whose necessary and sufficient conditions are the presence and interaction of estrangement and cognition, and whose main formal device is an imaginative framework alternative to the author’s empirical environment” (7-8). Key to this idea is the novum, the “strange newness.” What makes science fiction different from fantasy is that the novum is based on “primarily the political, psychological, and anthropological use and effect of knowledge, of philosophy of science, and the becoming of new realities as a result of it” (15). This is, essentially, what Suvin means when he refers to “cognition”: something that includes science, but also encapsulates “rationality” more broadly, I think. Though of course there are different levels of “science.” There are probably clearer ways to put this; no one would ever read Metamorphoses and then accuse Suvin of overwhelming clarity. When I taught this definition to one of my classes, they reformulated it and threw it back at me, which I appreciated, but did not think to write down!

Suvin's concept is maybe best explain through contrast: “…[L]ess congenial to SF is the fantasy (ghost, horror, Gothic, weird) tale, a genre committed to the interpretation of anti-cognitive laws into the empirical environment. …[T]he fantasy is inimical to the empirical world and its laws” (8). Or, he has a summation of a formula coined by Robert Philmus, which also does a nice job: “naturalistic fiction does not require scientific explanation, fantasy does not allow it, and SF both requires and allows it” (65). What I would add here (and maybe Suvin says this somewhere, I don't remember), is that the explanation often does not actually appear; science fiction just implies that it could offer you an explanation if it wanted to, but it's holding back. Star Trek is perhaps a good example of this; really, its science is meaningless on most counts, but everyone conspires to act as though it is science, and so the novum is maintained.

Despite the subtitle giving them equal weight, and despite the pages giving “history” more weight (it receives about 200 pages, whereas “poetics” gets only 85 or so), I would say that Suvin's discussion of history is not as interesting. It's a little idiosyncratic, and not quite as insightful. Suvin's one of those writers who works a little too hard to claim sf predecessors as actual sf, which I think obscures what those texts are actually doing, and the history only goes up to Wells, with the only 20th-century discussions being of Russian sf and of Karl Čapek. I mean, sure they're important, but why discuss the 20th century at all if you're going to ignore everything else significant that happened in it? As far as histories went, I preferred Brian Aldiss's Trillion Year Spree, though Aldiss's discussion of the poetics is much weaker. Which is why if you want a feeling for the foundation of sf criticism as it existed in the 1980s, you read both.