A (bad) solution to the inter-personal utility comparison problem

Utility is a measure of how well-off a person is. But, notoriously, it is very difficult to provide any precise measure of this, in a single case, or worse still, across people. How do we decide if the resulting benefits of an action for one person outweigh the associated losses for someone else?

The following is a tentative partial solution, of sorts. A lecturer once put me on this track, so the credit(/blame) is unfortunately not all mine. I’m not sure about it, but here goes anyway. Note that the solution provides little to no guidance for practical purposes – the point is to show that utility comparisons are meaningful, not that they are easy to make in practice.

First, let’s get a well defined measure of utility for one person. Let us assume, that the normatively significant kind of utility is preference satisfaction (weak assumption #1). The following is possible, at a theoretical level. People can, under pressure, prioritize their preferences. Let’s imagine that after some thought, we get Bloggs to decide that his preference ordering is: A>B>C ….>X>Y>Z.

Next, we get him to sharpen up those preferences so that we know how much he prefers A to B, B to C, and so on. Without this, we don’t know if he will prefer B&C to A. We can do this by offering him probabilistic options for A-Z. If he prefers a 6/10 chance of A over a certain chance of B, but not a 4/10 chance of A instead of a certain chance of B, we can assume that he desires A twice as much as B. Using this method, we can get a total list of Blogg's desires, and how much each is preferred to the last.

All that remains now is a method to compare these lists inter-personally. Plausibly, all human beings who wish to continue living will desire a glass of water when they are dehydrated to degree X (where X is some scientifically measurable physical body state). And, it seems that we could also assume that all such people desire that water equally (weak assumption #2). Such desires are not a function of individual personality, but instead a function of hardwired biological necessity. If that’s true, then it’s at least somewhat plausible that we can assume that everyone desires a glass of water under conditions X equally, and can then use their pre-established preference orderings to establish how much each person desires the rest of A-Z in comparison to the water.

And there you have it. If I desire F twice as much as water in circumstance X, and you desire G three times as much as water in conditions X, then G is impersonally better than F.

It’s not perfect, but it is interesting.