Still thinking a lot about Meillassoux's position and his route to get there in the book. I think it is a great book full of inspiring ideas. Indeed very much routed in what I take to be Badiou's ideas. Today I thought of listing its bright and its shady points. I start with the insights:
1. The idea that correlationism is followed and yet betrayed when the correlation is made absolute (or there is an attempt to build a metaphysics out of correlation). The important element of the idea, to me, is that correlationism carries a considerable measure of contingency. This is lost if the correlation is thought of as a necessary starting point for our relation with the world - the relation ceases to be factual, becomes absolute. I think this is a powerful idea and it is debatable whether it carries its force when we consider that there are not one but many correlations. I guess when one thinks of correlations as absolute, the second requirement of correlationism (the facticity of correlation) has been left behind.
2. The argument against frequentism using Cantorian transfinite arithmetic. This is brilliant. I had a similar discussion with Mariano many years ago in the context of the no-miracles argument for scientific realism. He was indeed pointing out that the no-miracle argument (science's capacity to predict is either based on its (approximate) truth or in a miracle) assumes that there is a totality of all possibilities so that it would be a miracle that things turn out often according to the (mature etc.) scientific predictions. Once this frequentist assumption is no longer granted, the argument looses its force. I guess the argument shows very well the Humean message that there could be no necessary connection entailed by the frequency of events.
My two misgivings has to do, each with one of the good points:
1. The presentation of correlationism and the cartography of positions that follow - compelling as it is - is sometimes vague and its vagueness makes the whole argument less persuasive. At one point, for instance, Meillassoux writes: "no correlationism, however insisting on its antisubjectivist rhetorics, is capable of thinking a dia-chronic statement without destroying its veritable meaning" (p. 121 of the English edition). This is the case, perhaps, if we take all correlationisms that would assume that the human subject is part of the correlation. But for other kinds of correlationism (or of their respective subjective metaphysics) such as the position according to which living organism, dispositions or desiring machines are correlata of the correlations, it is not clear at all that dia-chronic statements cannot be adequately understood. Yes, dinosaurs where in a correlation, therefore they did exist (without our thinking about them).
2. The endorsement of the Galilean thesis that what is nature is mathematizable and what is described in mathematical terms is independent of our thought is also vague and, at face value, means very little. It is unclear that the limits of the mathematizable could be established and, moreover, the Copernican-Galilean thesis as Meillassoux takes it to be would have to be defended against, say, the arguments that Wittgensteing presents in the Remarks, for example. We cannot ignore that much has happened in the philosophy of mathematics since Galileo. Plus, the defence of realism concerning science based on the mathematical characters of scientific laws seems strange and out of place: how can one be a scientific realist and argue that laws are not true and no prediction is in any sense reliable? Maybe this part of the book, however, was just sketched and much work needed to be done in this direction. And maybe, only maybe, that doesn't affect the standing of speculative materialism itself.
More about all that when I get back my lectures on correlationism in Madras next Thurs (or earlier...)
Douaillier just confirmed the dates of my cours libre in Paris 8, from 12 of March to 9 of April. More on this soon.