> When reading Capital I saw Marx mention he was starting out with an
> analysis of an ideal capitalism, and mentioning that of course supply
> and demand lead to all sorts of temporary fluctuations in exchange value
> and so on, and I think I sort of filled in the standard list of unrealistic-but-
> analytically-useful assumptions I had drilled into me from neoclassical.
As a start, that must have given you more mileage than those of us initially unfamiliar with neoclassical economics got. Personally, I think it's wrong to say that people have to "unlearn" this to truly understand that (or vice versa, to "unlearn" that to understand this).
People can start from different points, with different experiential backgrounds and factual referents, yet be able to converge to similar conclusions if they are intellectually honest in their dialectics, if (as Joanna suggested in a recent post) they have a commitment to engage in a common struggle. Just like the reproduction of a commodity producing society (enabling a division and reintegration of total social labor via exchange) presupposes the *practical* homogenization of labor (a process captured by the concept of "abstract labor"), similarly social practice (labor cooperation among humans) presupposes the ability of humans to *communicate* in spite of all obstacles, i.e. the ability to share or make *common* ideas that are at the outset different.
Two things:
1. As Mike Beggs noted, the neoclassical notion of demand (namely, the relationship between the amount of a good consumed and the factors that influence that amount) is inherent to Marx's notion of value as socially *necessary* labor time. What happens is that Marx, in volume 1 at least, assumed a constant "effectual demand" (to use Adam Smith's term). As far as I'm aware, the only place in volume 1 where Marx removed that assumption (to clarify how value regulated production in a society of private independent producers) is in that first long paragraph of ch. 3, section 2, devoted to: "C — M. First metamorphosis, or sale." (http://www.marxists.org/archive/marx/works/1867-c1/ch03.htm#S2a) If you assume that the "effectual demand" is constant, then the social labor necessary to produce a commodity is determined only by productive requirements. But that's because the size of "the stomach of the market" is assumed fixed.
2. As Mike also suggested, the neoclassical notion of supply is inclusive of the factors Marxists usually emphasize as arguments of the value "function" (or, its reciprocal, productivity) -- namely, technology, environmental conditions, etc. Much insight into value theory can be gained by studying the mathematical notion of *duality*.
With duality math, the most powerful theorems in conventional price theory (e.g. quantity vs. shadow price, Marshallian vs. Hicksian demands, utility vs. indirect utility, budget vs. expenditure, Slutsky or substitution vs. income, Roy's vs. Shephard's, etc.) can be derived from a bare-minimal conceptual framework. For people who enjoy the aesthetics of mathematical reasoning, this is truly appealing. All advanced microeconomics textbooks have a chapter on duality covering basically the same stuff. My personal favorites are Mas-Colell and Varian. One semester course on real analysis (taught in most colleges) gives one what's needed to tackle this material, which should be mandatory to any meaningful discussion of value theory. To repeat myself: Working through duality is one of the most illuminating experiences to anybody engaged in reflecting on value theory.