Interestingly, though, Alastair is a value realist. He thinks there is intrinsic value (and disvalue), and seems to accepts a traditional hedonistic account of these (the wrong view, IMO, but not our topic for today). Such Value Realism may naturally lead one to a broader Normative Realism, I think, in a couple of ways. So I'll address the rest of my post to any readers who share Alastair's starting point of Value Realism without Normative Realism, and see whether either of these arguments is persuasive.
First, we can ask whether you'd like to give up your Value Realism in favour of a relativistic view on which there's "hedonistic value", "desire-fulfilment value", and "Nazi value", all metaphysically on a par. If not -- if there's really just one correct view of value, regardless of what subjective standards anyone might arbitrarily endorse -- then we can raise the question of why normative reasons don't move in parallel. Surely an account of reasons for action that is grounded in facts about what's genuinely valuable is superior to an alternative account that bears no connection to the true value facts?
This leaves open a Sidgwickian dualism of practical reason between impartial and prudential reasons, depending on whether you think it appropriate to pursue goodness generally or only as it appears in your own life (and perhaps others that you care about). But it at least seems to rule out "Nazi reasons" and other candidate frameworks that have no essential connection to what's truly good. So that's something! (We might even secure some rational pressure towards impartialism via the idea that responding to all values in proportion to their objective weights is more principled than any more restricted focus.)
My second move is to ask about the significance of value attributions. When you call pain "bad", to a certain degree, are you just describing another qualitative feature of the world -- like shape or size -- or is there something more to it? One natural and (I think) appealing answer adverts to the analytic connections between "good" and "desirable", "bad" and "undesirable". As I previously put the point:
[I]t's a major advantage of the fittingness view that it makes explicit what the normative significance of value facts is to us. A value primitivist might say, "I see that this is good, but why should I care?" The fittingness view closes off such questions: to be of value just is to be something that merits concern (that it's rational to desire, pursue, etc.). The fittingness view makes explicit how abstract normative facts call for certain psychological responses.
This then yields a straightforward connection between the value facts and normative constraints on agents. It is fitting, justified, or rationally warranted to desire just what is truly desirable (good). To fail to pursue the good is thus constitutive of a kind of (substantive) rational error, or failure to respond to the genuinely normative reasons that apply to us.
You may be antecedently suspicious that our ultimate ends are rationally evaluable in this way. But you shouldn't be. Firstly, you need this implication in order for your claims about intrinsic value to have any normative bite. (Seriously, what do you even mean by calling one thing 'good' and another 'bad' if not that the first thing is more desirable, or worth pursuing, than the second? What kind of real value or "worth" would have no implications for what's worth pursuing?) Secondly, following Parfit, there are intuitive examples of irrational non-instrumental desires, such as those found in the Future-Tuesday Indifferent agent. If you're on board with those, there's no principled barrier to simply extending the range of ends that you're willing to judge as misguided, to cover all failures to respond appropriately to the objective values.