Skip to the end of all that, though, cut away the business about life being the ultimate value and the vague appeals to `metaphysics', and what you have is basically an eudaimonist account -- an ethics based on the notion of a good life.* Rand's eudaimonia (which, for the purposes of this post, I'll treat as synonymous with both `happiness' and `good life') is characterised by individualist rationality - the exercise, by the individual, of her capacities for `conceptual judgement'. The best human life, then, is one in which the individual acts according to (or possibly just in accordance with) the judgements produced by the exercise of her own reason, without either forcible interference from others or herself forcibly interfering with the actions of others.
But is this coherent? Notice the addition of the qualifications in the last sentence of the previous paragraph. Rand doesn't want to allow `parasites' to go around stealing the products of other people's hard work. Everything they have they should acquire by hard work and fair exchange. So eudaimonia, for X, consists not just in
(A) X acting according to X's reasoned judgements,
(B) Y not forcibly interfering with X performing A, for all Y, and
(C) X not forcibly interfering with Y performing A, for all Y.
For A-C to be coherent, it must be the case that
(1) there can never be a X and Y such that X's reasoned judgement is to forcibly interfere with Y performing A.
(1) doesn't seem at all plausible. Suppose, first, that X's considered judgement is to assassinate Y. Certainly this would interfere with Y acting according to his reasoned judgements. In this case, A is incompatible with C, and (1) fails. Second, suppose that Z now gets involved, and, acting on her considered judgement, stops X. This case is even more problematic for Randian eudaimonia than the first, because now Z's forcible interference in X performing A, by preventing X from forcibly interfering with Y performing A, brings about C. That is, one violation of C is necessary for another instance of C. For a third example, identify Z with Y: X is attempting to assassinate Y, based on his considered judgements, but Y is attempting to stop X from assassinate her, based on her considered judgements. The considered judgements of these two individuals have lead to an impasse; it is impossible for either of them to lead a good life. (Compare these three scenarios to the following premiss in the Objectivists' response to Nozick: `In a social context, the initiation of physical force ... by one man against another serves to destroy the precondition for living the life of a rational animal, since acting upon one's judgment becomes impossible.')
How can Rand avoid these dilemmas? She could say that some people simply can't lead a good life -- X and Y are simply doomed. because of their conflicting interests. But Rand actively denies this. In `The ``conflicts'' of men's interests' (in The virtue of selfishness), Rand argues that such conflicts can never exist (though she completely fails to consider these sorts of possibilities, instead worrying about whether to save a drowning stranger and two people who are both competing for the same job).
The only other option I can see is to introduce qualifiers into (A). So eudaimonia, for X, will consist, in part, in X acting according to X's reasoned judgements, except when G. But, for G to function as an additional constraint on X's reasoned judgements, it must be something independent of those judgements. For other eudaimonists, these are community standards, or something similar -- if my reasoned judgements tell me to do something wildly out of line with the standards of my community, then I shouldn't follow through on them. Rand, of course, can't make this move, as it amounts to subordinating the individual to the community.
The only constraint Rand seems to be willing to countenance on an individual's actions is rational human nature. But this is not independent of X's reasoned judgements -- after all, they're the product of her rational human nature -- and hence cannot provide additional constraints on them, just as modus ponens is no additional constraint on an argument that is assumed to be logically valid. In order to solve the dilemmas, then, Rand needs an argument for (1). But this is where we started. Rand's options seem exhausted, and no progress has been made.
* If this is right, then Rand can be added to the list of twentieth-century female ethical thinkers that challenged the prevailing utilitarian and deontological approaches to ethics and political philosophy using an eudaimonist approach. Other thinkers on this list include GEM Anscombe, Philippa Foot, Rosalind Hursthouse, Carol Gilligan, and Martha Nussbaum. Note that, as I understand things, of these six, only Gilligan and Nussbaum would be considered feminists.