i also think that many of the intuitions for egalitarianism are actually desert intuitions in disguise. not to say that desert doesn’t have its own issues, but people can capture these intuitions without giving up eg ex ante pareto
I think the transitivity argument works (but not necessarily as you present it).
One might argue that 1 util is such a low amount that the mere existence of 1M people with only 1 util over their entire life is bad. But an easy fix for this is to say that A+ has 1M people with 49 utils.
But another worry for you is that many egalitarians are fully willing to bite the bullet on levelling down. Larry Temkin, I think, is one such egalitarian. But also see Brett Doran's reply
I'm not sure that the expected utility argument works. People who are risk averse (and a lot of people are risk averse) will find choosing a guaranteed 50 utils more intuitive than a 50% chance of 101 utils.
Now, there might always be some level of tradeoff in risk vs expected utility that people will be willing to accept and you might sorites your way to 50% chance of a much larger amount of well-being. But, I don't think that gets you quite the result that you want.
Importantly, there seems nothing ipso-facto irrational about caring more about the first n units of well-being than the subsequent n+1 units.
To fix intuitions, I think it is helpful to imagine assigning nontradeable experiences (falling in love, getting divorced, death in the family) to either a billionaire or a janitor. Here we have abstracted away both from the diminishing value of wealth, and the possibility of a positive-sum trade**. I think most people would choose the world in which the janitor gets the good experience / avoids the bad experience. Even if we stipulate, for example, that the billionaire would suffer more from the loss of her mother than the janitor, such that helping the janitor is strictly utility decreasing, I think many people would still choose to help the janitor.
I think non-egalitarians have to account for this intuition in some way, which I suspect has something to do with a vaguely Rawlsian notion of cosmic fairness.
**If the good is tradeable, giving to the janitor will always be utility-maximizing since she can sell it to the billionaire if that is more efficient. We want to consider worlds in which giving it to the billionaire is utility-neutral or utility-maximizing.
I coincidentally posted something about Huemer's argument earlier today but will paraphrase it here:
I think it commits equivocation on the criterion used to compare each world. Principles 1 and 2 are switched out for each other arbitrarily in order to support the arguments.
When comparing A+ and A the principle is self-interest of the occupants of each world, in which case A+ seems better. When comparing A and B, the principle used is overall or average utility, in which case A seems better.
The trouble is that if one wants to use transitivity to infer A+ > A > B, the principle for each comparison needs to be the same, or else no comparison is made between A+ and B.
If we are consistent and use self-interest for example, then for one million people in B who do not exist in A, B > A because 50 wellbeing level is better than none.
If we compare everything using utility, that does nothing to debunk egalitarianism, since the argument for ~(A+ > B) is based off inequality, not anything to do with overall utility. Thus they would object to saying A+ > A on this basis.
"In this case, the prioritarian cares more about the first 50 utils than the next 51. So they have every incentive to take option A, even though it’s worse for people in expectation."
Do you mean "option B", which is worse than A in expectation (B increases welfare by 50 utils whereas A increases welfare by 50.5 utils)?
I just found your blog and don't know a ton about this subject, so IDK if this point has been discussed, but the claim in 2.1:
> If one world is better than another world for everyone who exists in either world, that world is better. E.g. our world would get better if everyone was made better off, or if everyone was made better off and a bunch of other happy people would be created.
Has to be false, or at least not well-defined. Consider world A having a constant amount of "better-ness," call it x, and world B has an amount x - epsilon. But tomorrow, world B has an amount "x + epsilon." So A > B today, and B > A tomorrow.
But that's a problem, because as you mention you can't generally include hypothetical future utils in a way that makes sense, even though in special cases you can, eg, a well-known deterministic process adds that extra epsilon.
Fundamentally, this problem is not about time-dependence or path-dependence. It's about the landscape of some potential in an optimization problem, and generally there is no well-ordering on the space of all such landscapes. (Even if the optimization metric itself is fixed and well-defined, which it may not be--there's no reason that can't be dynamic too!)
This also implies points 2 and 3 are also false, as there is never any ordering to begin with.
When I'm asked whether something has positive value in isolation, I get the feeling that I'm being set up. It's like if your teenage self heard your Mom ask "Do you think cleanliness is an intrinsic good?" Well, cleaning up takes time, Mom. What's the trade-off? Playing video games? Doing my homework?
Likewise here, I feel like the tradeoff is going to be imposed AFTER I commit. And that's not fair, because as any economist will tell you, it's all about the trade-offs. Those of the right say the trade-off to egalitarianism is growth. Those on the left will say the trade-off to not having it is social justice.
And sure enough, the trade-off I didn't sign up for does come: poor Bob gets punched in the nose, obviously hurting general welfare, which I was told was going to stay equal. All the more intuitive cases that wouldn't break the welfare-stays-the-same – the wealthy dude taking 4 vacations a year he REALLY enjoys, and the poor dude working every day of the year – somehow go unconsidered.
> The first big objection to egalitarianism is that it implies, counterintuitively, that it is in some way good to simply make a person worse off. Suppose that Bob is very well off. You can break his nose and benefit no one. That doesn’t seem good in any way. But on egalitarianism it is good in some way—it reduces inequality. It might not be good all things considered, but it’s at least good in some respect.
I think this is one of the intuitions that *justifies* it. It's the intuition behind destroying billionaires' property, which I suspect I could find a decent amount of support for.
I think these are interesting arguments, though I would be really interested in an in-depth analysis of the extent to which the philosophical nuances around these views could affect actual policy discussions. I would think that in practice, egalitarianism and prioritarianism are not so much simple and abstract modifications of the utility calculus, but rather empirically motivated methodological attitudes that are grounded in economics and social science. Another way to put this is: you end the piece by explaining the relevant intuitions using diminishing marginal utility. Well, to me, that *just is* an aspect of prioritarianism. And high inequality exacerbating social instability *just is* an aspect of egalitarianism. Then one has to hash out the details of the size of these effects and the levels at which they kick in, but that will be more involved than coming up with two hyper-idealized "worlds" A and A+ and observing that A+ is obviously better. Maybe this will be a bit snarky, but oh well: has something like this ever helped with settling a policy question?
i also think that many of the intuitions for egalitarianism are actually desert intuitions in disguise. not to say that desert doesn’t have its own issues, but people can capture these intuitions without giving up eg ex ante pareto
Yeah makes sense.
"Egalitarianism is false"
-- Ooh this is going to be good
"So is prioritarianism"
-- Wow, rude
(Great post though.)
I was confused for a moment until i realized this was abstract philosophy stuff, not the political theory of everyone being equal under the law.
I think the transitivity argument works (but not necessarily as you present it).
One might argue that 1 util is such a low amount that the mere existence of 1M people with only 1 util over their entire life is bad. But an easy fix for this is to say that A+ has 1M people with 49 utils.
But another worry for you is that many egalitarians are fully willing to bite the bullet on levelling down. Larry Temkin, I think, is one such egalitarian. But also see Brett Doran's reply
I'm not sure that the expected utility argument works. People who are risk averse (and a lot of people are risk averse) will find choosing a guaranteed 50 utils more intuitive than a 50% chance of 101 utils.
Now, there might always be some level of tradeoff in risk vs expected utility that people will be willing to accept and you might sorites your way to 50% chance of a much larger amount of well-being. But, I don't think that gets you quite the result that you want.
Importantly, there seems nothing ipso-facto irrational about caring more about the first n units of well-being than the subsequent n+1 units.
You can also just fix 1 util to be above the threshold.
To fix intuitions, I think it is helpful to imagine assigning nontradeable experiences (falling in love, getting divorced, death in the family) to either a billionaire or a janitor. Here we have abstracted away both from the diminishing value of wealth, and the possibility of a positive-sum trade**. I think most people would choose the world in which the janitor gets the good experience / avoids the bad experience. Even if we stipulate, for example, that the billionaire would suffer more from the loss of her mother than the janitor, such that helping the janitor is strictly utility decreasing, I think many people would still choose to help the janitor.
I think non-egalitarians have to account for this intuition in some way, which I suspect has something to do with a vaguely Rawlsian notion of cosmic fairness.
**If the good is tradeable, giving to the janitor will always be utility-maximizing since she can sell it to the billionaire if that is more efficient. We want to consider worlds in which giving it to the billionaire is utility-neutral or utility-maximizing.
I coincidentally posted something about Huemer's argument earlier today but will paraphrase it here:
I think it commits equivocation on the criterion used to compare each world. Principles 1 and 2 are switched out for each other arbitrarily in order to support the arguments.
When comparing A+ and A the principle is self-interest of the occupants of each world, in which case A+ seems better. When comparing A and B, the principle used is overall or average utility, in which case A seems better.
The trouble is that if one wants to use transitivity to infer A+ > A > B, the principle for each comparison needs to be the same, or else no comparison is made between A+ and B.
If we are consistent and use self-interest for example, then for one million people in B who do not exist in A, B > A because 50 wellbeing level is better than none.
If we compare everything using utility, that does nothing to debunk egalitarianism, since the argument for ~(A+ > B) is based off inequality, not anything to do with overall utility. Thus they would object to saying A+ > A on this basis.
What do you think about sufficientarianism?
I think most of what I say in the article applies to it.
yep
Hi Matthew. Nice post.
"In this case, the prioritarian cares more about the first 50 utils than the next 51. So they have every incentive to take option A, even though it’s worse for people in expectation."
Do you mean "option B", which is worse than A in expectation (B increases welfare by 50 utils whereas A increases welfare by 50.5 utils)?
I just found your blog and don't know a ton about this subject, so IDK if this point has been discussed, but the claim in 2.1:
> If one world is better than another world for everyone who exists in either world, that world is better. E.g. our world would get better if everyone was made better off, or if everyone was made better off and a bunch of other happy people would be created.
Has to be false, or at least not well-defined. Consider world A having a constant amount of "better-ness," call it x, and world B has an amount x - epsilon. But tomorrow, world B has an amount "x + epsilon." So A > B today, and B > A tomorrow.
But that's a problem, because as you mention you can't generally include hypothetical future utils in a way that makes sense, even though in special cases you can, eg, a well-known deterministic process adds that extra epsilon.
Fundamentally, this problem is not about time-dependence or path-dependence. It's about the landscape of some potential in an optimization problem, and generally there is no well-ordering on the space of all such landscapes. (Even if the optimization metric itself is fixed and well-defined, which it may not be--there's no reason that can't be dynamic too!)
This also implies points 2 and 3 are also false, as there is never any ordering to begin with.
So you're saying it shouldn't matter what others think if it gives me more pleasure to bring about the end of the world?
So you're saying that a concept that is impossible to implement is absurd?
I don't need to read through such rubbish.
When I'm asked whether something has positive value in isolation, I get the feeling that I'm being set up. It's like if your teenage self heard your Mom ask "Do you think cleanliness is an intrinsic good?" Well, cleaning up takes time, Mom. What's the trade-off? Playing video games? Doing my homework?
Likewise here, I feel like the tradeoff is going to be imposed AFTER I commit. And that's not fair, because as any economist will tell you, it's all about the trade-offs. Those of the right say the trade-off to egalitarianism is growth. Those on the left will say the trade-off to not having it is social justice.
And sure enough, the trade-off I didn't sign up for does come: poor Bob gets punched in the nose, obviously hurting general welfare, which I was told was going to stay equal. All the more intuitive cases that wouldn't break the welfare-stays-the-same – the wealthy dude taking 4 vacations a year he REALLY enjoys, and the poor dude working every day of the year – somehow go unconsidered.
Is that so?
https://substack.com/home/post/p-192659760
I wrote an article responding to this post (and Huemer's):
https://substack.com/home/post/p-192659760
> The first big objection to egalitarianism is that it implies, counterintuitively, that it is in some way good to simply make a person worse off. Suppose that Bob is very well off. You can break his nose and benefit no one. That doesn’t seem good in any way. But on egalitarianism it is good in some way—it reduces inequality. It might not be good all things considered, but it’s at least good in some respect.
I think this is one of the intuitions that *justifies* it. It's the intuition behind destroying billionaires' property, which I suspect I could find a decent amount of support for.
I think these are interesting arguments, though I would be really interested in an in-depth analysis of the extent to which the philosophical nuances around these views could affect actual policy discussions. I would think that in practice, egalitarianism and prioritarianism are not so much simple and abstract modifications of the utility calculus, but rather empirically motivated methodological attitudes that are grounded in economics and social science. Another way to put this is: you end the piece by explaining the relevant intuitions using diminishing marginal utility. Well, to me, that *just is* an aspect of prioritarianism. And high inequality exacerbating social instability *just is* an aspect of egalitarianism. Then one has to hash out the details of the size of these effects and the levels at which they kick in, but that will be more involved than coming up with two hyper-idealized "worlds" A and A+ and observing that A+ is obviously better. Maybe this will be a bit snarky, but oh well: has something like this ever helped with settling a policy question?