Quentin Meillassoux lays out his concept of extro-science fiction. Science fiction can be designed as a genre that introduces a departure from our world that can be articulated through the principles of science (basically, the novum, as sci fi scholar Dark Suvin puts it). And one of the fundamental principles of science is that a given concept or feature can be proved scientifically, through experimental means. Extro-science fiction, then, is fiction where experimental science as a concept or principle exists, but conducting experimental science is impossible. Essentially, the idea of a universe made through the laws of science is known and accepted, but within the story, those laws aren't just suspended, but fundamentally fail to hold. It's an interesting idea, and Meillassoux does a really impressive job of explaining this nonintuitive concept in an approachable, reasonable manner.
However, I have two basic issues with the book. First, I'm not totally convinced on the necessity or the value of this newly coined genre, as anything beyond a thought experiment. A world where scientific experiment is for some reason impossible as a foundation for knowledge is a world that it's kind of hard to tell compelling stories about. That's an interesting thing too; to what degree are the rules for making a narrative co-concurrent with the understanding of a scientifically consistent world? But that's not totally something Meillassoux is interested in, at least not here. Second, even if you do find this approach really interesting, it's hard to make the case that this is a book worth its price tag. About half the book is a reprinting of one of the texts Meillassoux references, the 1966 Asimov short story "The Billiard Ball." That means there's about 50-60 pages of actual argument here, which is about an extended chapter's worth of actual content. I like "The Billiard Ball," but at the end of the day, it's a50+ year old sci fi short story, and doesn't add a lot to the overall value proposition here.
It's a good, well-argued argument, and well translated too. But it more left me with a desire to read his other book, After Finitude: An Essay on the Necessity of Contingency, than appreciation for the book I just read.
(EDIT: Also, I have a lot of sympathy with the argument that comes up in a lot of the other reviews, that Weird Fiction and other speculative fiction addresses a lot of Meillassoux's proposed genre here. You can make a little distinction that these stories don't fully count as the same because they're set in worlds where the scientific principles still hold in some form, or never held, but that feels like quibbling.)