Link raised an interesting concern for me regarding utilitarianism, which I'm currently doing a lot of contemplation on - is utility fungible?
Let's assume for the purposes of writing something useful that my last post doesn't apply; we'll assume utility exists in a quantifiable measure. Should I, as a utilitarian, choose a universe in which I and everyone I love is tormented for all eternity, in exchange for a billion people living in perpetual bliss? Is this a fair utilitarian exchange? Should I choose a minute of extraordinary pain over a hundred years of minor inconvenience?
Utilitarianism, in point of fact, -depends- upon the idea that utility is to some extent fungible.
An argument in favor of the fungibility of utility is that, if some circumstance forced me to choose between two things (abstract or real, it doesn't matter), I would choose one.
An argument against the fungibility of utility is that, absent circumstances forcing me to choose, there are things which I would not exchange for anything that does not include the thing itself or a means of recreating that thing.
I am forced to conclude that utility is incompletely fungible, something which in fact already implied by marginal utility. There are flavours of utility which cannot be freely converted. No amount of sleep, no matter how good, can make up for a lack of food. No amount of food can make up for a lack of sleep.
The utility function is no such thing; there's no one value which can represent how well-off you are, nor how well-off the universe is in terms of your values. I can envisage a function which could -approximate- this value, but in extreme situations it would cease to present any meaningful information; should I prefer an existence in which I'm going to starve together in .0001 seconds, but that fraction of an instant will be filled with such utility through other means as to dwarf my life utility as it stands today? What does that even -mean-? The utility, whatever it is, isn't fungible with the utility of not starving to death, unless maybe it is - maybe I am strapped into a machine that gives me subjective eons in that .0001 seconds - but that's just it: There's some utility which -can- be exchanged with other utility, and some utility which can't. There are, as previously mentioned, different flavours of utility, and they don't map to a single value representing how desirable anything is.
Another thing that suggests the non-fungibility of utility flavours is the existence of cyclic preferences - where I prefer universe A to universe B to universe C to universe A. A>B, B>C, C>A - which do you choose?
A moment of thought permits me to construct a cyclic preference list for myself:
A: Restaurant with bad-tasting food, plenty of drink
B: Restaurant with salty (but delicious) meals, no drinks
C: Restaurant with boring meals, limited drinks
Maybe you don't find this preference sequence cyclic; I do. Perhaps you can construct a cyclic list of preferences in your daily life, perhaps you can't - personally, I can, on a number of things. In utilitarian logic, this means my preferences are irrational. So I suppose it's a good thing I don't use utilitarian logic!
(Note, incidentally, that I wouldn't actually choose any of those options, had I any other choices. A necessary ingredient in cyclic preference is a trade-off between different values, different flavours of utility. In practice, I'd find somewhere else to eat. Those are based on real restaurants, actually, and no I won't tell you which ones. The actual cyclic preference list of restaurants with flaws is really long; the entire time I lived in that region I found exactly one restaurant that didn't have a flaw)
Picking the ideas of the last post back up again:
The issue ultimately comes down to this: The idea of "Utility" is a -very- crude and clumsy way of representing "desirability" of a state of affairs, desirability being both multivariate and indeterminate. The "utility function" is an abstraction which serves to permit utilitarians to pretend that their philosophy can account for everything while not actually having to account for everything. Can it account for love? "Yes, it's utility input #17 in our list of known utility inputs." Okay, how does love compare to having enough food to eat? "Well..." Okay, that's pretty hard, how about this: At which point should we resort to cannibalism if we're trapped somewhere with our loved ones? Who should be eaten first - should it be one of the parents, or one of the children? Should we wait for somebody to starve first, and eat the dead, or eat somebody sooner that that? "Uh..." You have no idea how to even begin answering these questions in terms of utility, do you? I mean, it's a bit of an extreme circumstance, sure, but you're over there trying to decide whether theoretical universe A is superior to theoretical universe B, can't you spend some time trying to figure out what your philosophy has to say about the real world and the circumstances people occasionally find themselves dealing with?
I don't necessarily take marks away from utilitarianism for not having a prepackaged answer to the question of cannibalism, mind. I don't think any nontrivial moral philosophy has a good answer to those questions. I -do- take issue with utilitarianism's implicit claim that there is a simple answer if you can just substitute in these utility values for your loved one's lives and their odds of survival and the odds of rescue and this and that and that. It -sounds- simple, when you put it in those terms, but if you actually try to do the math, you start having to account for the unknowns and the unknown unknowns and you come up with an answer that, if you're lucky, has some bearing on the actual universe, but is probably wrong anyways. In the end utilitarianism doesn't actually help you make any decisions; it doesn't provide any kind of tangible framework in which to evaluate anything. It's like having the C specification when all you have to work in is assembly; you can sort of make your assembly code look the way you imagine C specifies it should look, but ultimately the specification has no bearing on how you actually code anything. You're not a C programmer because your assembly was written while your were imagining how C code would compile; you're not a utilitarian because you think about utility while you figure things out for yourself.
Because nobody actually performs the math. The idea of utility is an illusion, a handwave, a massive blank space in the -middle- of the map on which is written "There be mathematics here."