I think ‘meet’ is a bit too strong. Imagine that we incontrovertibly, axiomatically know – perhaps because God’s told us – that there is a Platonic ideal of Justice, and that no human understanding of justice encompasses it or ever will. Meanwhile back in the 21st-century philosophy-reading world, people are still talking about justice and putting forward alternative conceptions of it, all of which are more or less comprehensible and more or less capable of guiding actions. I don’t have a problem with calling both of these things justice, or with teleological arguments which assume a progress towards a closer approximation of the ideal of Justice – even if our conception of what that ideal is, and what constitutes progress towards it, changes over time. I think our conception of justice as something that has a coherent ideal form is a large part of what makes justice worth talking about, even if we know we’ll never have a final answer to what that ideal form is. In Nigel Simmonds’s words, “law is the process of its own becoming”; I think there’s a similar, if not quite so tight, linkage between the Justice behind the curtain and the debate between versions of ‘justice’.

To put it another way, what’s the difference between being told “you’ll never know what justice really looks like” and being told “you’ll never know what the stars really look like”, other than that a few dozen astronauts can actually say the latter?

This was something I was trying to get at before. (I’m not sure who said this- I pulled it as a quotation from Dsquared, so maybe it’s wrong to attribute an idea like this to Cohen, though he does seem committed to something like this to me.) But, I don’t think that morality or justice could be like this, that is, that we could in principle fail to meet the standards, at least some of the time. (We might fail to meet them much of the time for completely banal reasons, of course.) This is what I meant by saying that ideas like justice or morality are essentially action-guiding. If it turns out that on one’s account of justice or morality, we might always fail to meet it because we can’t figure it out, that shows you’ve got the wrong concept. In this way justice and morality are not like numbers or the like- we might think it makes sense to say there’s an answer to the continuum problem but that we are not smart enough to work it out, but I don’t think that, or something like it, can be right about justice or morality. But that seems to be an essential part of any Platonic view. And even if that view can be made coherent (I think it cannot), the answer would just be “who cares?” It wouldn’t at all be clear why we’d care about such concepts. The two reasonable answers would be “it turns out that justice or morality are not something we do or should care about” or “that’s not justice or morality, as those have to be something that could be action-guiding. ” The second answer seems better to me.

I think that’s right. Or at least, whilst it is difficult for people to fix the quantum with precision, this isn’t really necessary, and we can pretty easily identify egregious violations.

1) If we’re aiming to give an analysis of justice, one of the things about justice that our analysis had better retain is the fact that we can be fallible about what justice requires. That there are individuals and groups who are systematically misguided about what justice requires must be true if there is anything properly called justice at all. But once you accept that, you also have to accept that the requirements of justice do not depend in any straightforward manner on what people do or can know. (As Chris notes, saying this certainly doesn’t commit you to thinking that justice has some physical or spiritual existence.)

2) One point that’s floating around is whether the concept of justice will turn out to be coherent at all. Perhaps this is fuelled by a worry that the above thought – that justice is supposed to be independent of what we think about it – is somehow incompatible with other features that justice is supposed to have (e.g. that it is supposed to be action-guiding) . This is very roughly the John Mackie-style error theory, only applied to justice. This kind of view is certainly academically respectable. But one obvious and salient problem is that if you go down this route it’s not clear that you can come out with any political views at all. If nothing is just, are you going to say that systematic sexism (for example) is only bad, if at all, for pragmatic reasons?

If I read Cohen correctly, he says that in thinking about justice in the situation described in my post, we shouldn’t take as given that people will work as much as they want to and no more because, in acting this way, the talented are violating principles of justice. It seems to me that, for this argument to go the way he wants, this must be a deliberate violation of justice by the individuals concerned, not merely an unfortunate outcome of ignorance. (cf his discussion of paying ransom to kidnappers).

So, it seems to me to matter whether talented individuals can determine their own obligations. If, for argument’s sake, you agree that no individual can determine how much they ought to work, then it seems to me that Cohen’s argument fails. As I hinted in my post, I thought the response was going to be (following the charity/restaurant comparison in the intro) that people can determine this pretty easily (see also Daniel @ 25 & 41, which came up as I was writing this) and that those who suggest it’s difficult are being self-serving.

That seems plausible to me, but in view of my difficulties with the initial intuition, not as convincing as I would like.

This does at least look consistent (although we’re assuming here that “the effects of their actions” has a referent – actually, me and John have swapped arguments in the past over whether there exists a suitable probability measure over events of the sort that would be needed for an expectation to exist), but surely it is too bad; if we end up in that position, surely we need a different theory of morality (and even more so if we want a theory which is going to underpin the institutions of society).

I think that there is clearly a “typical economist” problem here though – I think me and John both would tend to regard the transformation problem, the socialist calculation problem and the problem of interpersonal aggregation of utilities (and a bunch of other aggregation problems) as deeply, intrinsically insoluble in a way in which people coming from other backgrounds might see them as merely epistemologically very difficult. But I do think there’s an issue here; to take John’s example, if we had settled on a concept of justice which depended very much on there being a fact of the matter about length, then surely General Relativity would have been a serious blow to that concept.

My guess is that this is only really of a problem for someone who thinks that the meaning of justice has to do with specific rules about the institutions of society. As I said above, if John’s bricklayer makes a good faith effort to contribute a bit toward the welfare of others, then there’s a sense in which he’s behaving in a “just” manner, and if he purely responds to his personal incentive schedule then there’s a sense in which he’s being “unjust”, and it seems to me to be perfectly clear that Cohen is in tune with the normal meaning of justice in disagreeing at a fundamental level with Rawls here. And I don’t think there’s much of a problem created by the fact that whether a good faith effort has been made is private subjective information to the hypothetical bricklayer – I don’t think that kind of epistemic problem is important.

I’m still thinking through the incoherence point which seems to go deeper than the suggestion that it might just be difficult or even impossible to find out whether a conception of justice was being realised in practice. (I took your point to be the latter, and I see that I may have missed something there.)

John queries whether justice has an existence pretty much independent of what we know about it. Well, yeah, obviously – though it’s a jump to say so in a Platonic sense. This reminds me of a post-modernist argument: we can’t know for sure anything about the external world, so surely we cannot say it exists? The problem is that the epistemic problem of knowing about the external world does not entail an ontological problem about the external world’s existence. Likewise with our justice problem here, epistemic problems with achieving the distribution do not entail that it is not just to achieve it. It is simply a non sequitur to say otherwise, and that is why Chris can dismiss from the start any epistemic concerns with the principle of justice. That is not to say those concerns are unimportant, they had pretty dire consequences in the moral utilitarianism example above, but simply to say that they alone cannot invalidate a particular justice principle.

More importantly, rather than arguing that the particular epistemic problems I’ve raised aren’t serious difficulties, or that there is an adequate response available (which I’m perfectly prepared to believe), you’ve just ruled them out of court from the start. As the examples dsquared and I have given suggested, there are lots of concepts that turn out to be incoherent (greatest good of the greatest number and aggregate capital stock are neat examples) or to be misdefined because they rest on assumptions about the world that aren’t true (such as that the length of an object is a well-defined property independent of the observer).

Compare “the greatest good of the greatest number”. Lots of people used to think that this was the goal of society, and it took some fairly interesting developments in economics and mathematics to get the formalisations to a point where we could realise that it’s actually not a coherent concept – I take it that you wouldn’t say that utilitarianism is actually all about promoting “the greatest good of the greatest number” despite a load of practical problems in doing so.

Is this a reference to Arrow? If so, then the claim is that there’s no non-arbitrary way of aggregating individual preferences into a SWF. But it doesn’t follow from that that the very idea of maximizing the good is incoherent, because maximizing the good might not be a matter if individua preference aggregation.

(If you were making a different point, then ignore me.)

Compare “the greatest good of the greatest number”. Lots of people used to think that this was the goal of society, and it took some fairly interesting developments in economics and mathematics to get the formalisations to a point where we could realise that it’s actually not a coherent concept – I take it that you wouldn’t say that utilitarianism is actually all about promoting “the greatest good of the greatest number” despite a load of practical problems in doing so.

I think it’s entirely open to John, me, etc, to say that “maximise the welfare of the worst-off”, which looks like it’s a coherent goal for society to have, might also have serious conceptual problems. Some kinds of epistemic problems aren’t just “practical difficulties”.

[1] And in fact a lot of economists who have forgot or never learned the CCC do in fact claim that it’s merely an epistemic problem or can be ignored because “reswitching” doesn’t happen very often in the real world.