In Chapter 4 Joyce continues to argue that rational agents as such have no particular ends, not even the end of being or remaining rational. It seems to me that this is pretty much insisting on one particular conception of rationality, and a pretty narrow and odd one at that. Perhaps we can call it a technical sense of rationality and reason. In a more normal sense of the word, it is not rational to choose to have a lobotomy (in normal circumstances) or to do something that is predicted to make you insane. Nor is it rational to harm yourself physically or financially. Nor to do things that have no point beyond satisfying some bizarre whim, such as putting all your green books on the roof, to use one of Anscombe's examples. To do something on a whim, I would think, is to do it for no reason. That doesn't make it irrational, if the whim is fairly normal. If I see a flat stone near a pond and try to get it to skip across the surface then I might have little to no reason for doing so, but I'm not being irrational. Here, I think, doing something for no reason and one's reason being a mere desire or whim are pretty much the same thing. We could explain the action either way and get the same point across. But having a sudden urge to walk into traffic or throw myself off a cliff does not in any sense give me a reason to do those things, surely. Not every whim is a reason to act on that whim, and those that are, or that give you a reason to do what you have a whim to do, are only reasons in a weak sense. "I just felt like it" is, at best, not much of a reason for doing anything (and by this I don't mean that it isn't a good reason).
Near the end of the chapter we get some more food for thought on philosophical uses of Nazis. There is an interesting contrast between Joyce's views and Anscombe's on this too. Anscombe's paradigm of injustice is the judicial killing of people known to be innocent. Another might be the deliberate ordering of the killing of noncombatants in war. Deliberately killing people you know to be innocent is murder, in Anscombe's book, and murder is about as bad as crimes get. So she was no fan of the Nazis. On the other hand, she didn't seem too impressed by the Nuremberg trials, which I think she saw as a bit of victor's justice. At least some of the Nazis put on trial were there, not because they had broken any existing laws, but for what they should have known were crimes against humanity. Joyce, though, treats the 1946 trials as a paradigm case of moral discourse. This discourse, he says (see pp. 98-99), demands that the Nazi leaders be censured in the strongest terms. He adds that they were hanged. He does not say that this was a good or a bad thing to do, but he seems to regard it as an example of the moral discourse that he wants to both critique (when taken literally) and defend (when viewed in a fictionalist light).
He also compares the question "Ought Hitler not to have ordered genocide?" to "Is Beethoven's Fifth Symphony a piece of music?" This suggests to me that he regards Hitler's order as paradigmatically immoral in the way that Beethoven's Fifth is something like the most famous piece of music in the world. We are in the same ballpark as G. E. Moore's "I know that I have two hands." Joyce quotes Gilbert Harman saying that "it sounds odd to say that Hitler should not have ordered the extermination of the Jews," and agrees that it sounds odd, but adds that most people asked whether Hitler ought not to have done so would be perplexed by the question but then answer "Of course!" I think this underestimates the oddity of the question. I find it hard to imagine anyone answering it straight, i.e. without some sarcasm. Many people, I suspect, would not get beyond perplexed questioning about the question. Forced to answer yes or no, of course most people would say yes. But I don't think this proves Joyce's understanding of ordinary moral thought correct. More on this later.
On p. 99 he writes that "the neo-Nazi's moral praise of Hitler is as absolutist as is our condemnation of him." Isn't this false? Don't neo-Nazis typically deny the Holocaust, rather than insisting that it was a good thing? If they say it was good, isn't this generally to insult and annoy the people they don't like, rather than a straightforward report on what they believe? And if anyone really does believe in genocide, do they really do so as absolutely as its opponents oppose it? Does anyone think "Ever again," for instance, in the way that many people think "Never again"? I don't think so. What people believe seems to be getting flattened or reduced in a way that strikes me as similar to the way that what it means to be rational was earlier shrunk to something almost unrecognizable as a conception of rationality.
In picking on stray thoughts that I find particularly wrong or telling, I'm not giving much of a sense of what Joyce takes himself to be doing. So let me outline his strategy. (Better late than never.) Moral discourse, he thinks, commits us to the idea that we have reasons to behave morally. If you ought to do x then you have a reason to do x. But this reason is supposed to be independent of what we happen to desire, value, or care about, etc. The only reasons like this come from practical rationality, Joyce thinks. That is, the rules of various practices require us to behave in certain ways, but if you don't care about those practices then you have no reason to follow those rules. If you ask "why should I care about practical rationality?", though, then you have undermined yourself. The question itself presupposes that you do care, that you want reasons. So moral reasons could be saved if they turned out to be part of practical rationality. But it turns out they aren't. Practical rationality depends on what each agent happens to value or desire. So it is relative in a way that moral discourse is not meant to be. So moral discourse, with its contradictory requirements, is misconceived.
Or rather, Joyce thinks that it looks that way. He concedes that he has not proved his point, but he thinks the burden of proof lies with those who claim that rational agents as such must have the same normative reasons. And he thinks they have not proved this.