Summary

All classical utilitarians need to choose how much they care about various experiences. Negative-leaning utilitarians are actually just classical utilitarians using a set of parameter values that tend to take extreme suffering more seriously.

Introduction

Several people have asked me about my negative-leaning utilitarianism, so I figured it was time to write a summary. There's not a lot to say, actually. The main difference between myself and other classical utilitarians is that I consider severely painful experiences to be much more serious than they do relative to other things. I don't consider severe pain to be infinitely more important than any pleasure, but it's a very large finite amount more important.

Exchange rates

One way to say this is that when trading off suffering against happiness, my suffering:happiness exchange rate is higher than what some other people use. This is roughly accurate, although I think my exchange rate is pretty typical for mundane kinds of suffering -- stubbing your toe, etc. Rather, I begin to diverge with other people when it comes to the really atrocious forms of suffering, like torture, suffocation, or being eaten alive.

The choice is up to us

In fact, speaking about "exchange rates" is really just an approximation to a deeper story. No, Virginia, there really is no such thing as a utilon in any non-arbitrary sense. Happiness and suffering are not actually cardinal numbers that live in the physics of the universe that we can measure. Rather, we use numbers to express how much we care about an experience. I might say a minute of sunshine is +1, a minute of listening to wonderful music is +3, a minute of seeing my best friend after a long separation is +25, stubbing my toe is -3, waking up to my alarm clock is -15, etc. These numbers are all things I make up; they're free parameters in utilitarianism. In fact, utilitarianism has an unbounded number of free parameters: What numerical value do we want to assign to each experience?

Of course, we can choose to make our assignments more uniform. We can say that very "similar" experiences (e.g., stubbing my big toe vs. stubbing my little toe) shouldn't have vastly different numbers attached. We can try to impose other regularity conditions that seem plausible and jibe with our intuitions. But at the end of the day, it's up to us what numbers we pick.

My negative-leaning utilitarianism is as simple as saying that I pick pretty big negative numbers for torture-level suffering, and I do this because I find it intuitive that torture-level suffering is really serious relative to other things. That's it.

So yes, I'm really just a classical utilitarian, using one of the unboundedly many possible parameterizations of classical utilitarianism.

Here's a reason to be skeptical about the intuition that motivates your "non-classical" exchange rate between positive and negative experiences. Elsewhere you write that you have "toyed with non-pinprick negative utilitarianism when the experience is as bad as, say, 2 minutes of burning at the stake." But you acknowledge that there are decisive objections against negative utilitarianism. So you embrace instead a form of "semi-negative" utilitarianism where intense pains are weighed more heavily in the calculus than are on classical utilitarianism. The mental process by which you arrive at your current position thus seems to me something like this:

1. "It seems intuitively that, for pains of sufficient intensity, there is no amount of pleasure that could outweigh them."
2. "So I should become a negative utilitarian."
3. "But wait! When I try to express this intuition formally, I can see that the resulting theory is unsustainable."
4. "Okay, then I should embrace semi-negative utilitarianism instead: a form of utilitarianism that attaches much more--but finite--weight to pains relative to pleasures. Such a theory approximates my original intuition tolerably well, without facing the problems of strict negative utilitarianism."

Now, if it turns out that the theory that best represents the intuition can be shown to face decisive objections, doesn't this provide reasons to be skeptical about the intuition that motivated that theory? When modern physics tells me that my naive theory about the world is wrong, I don't react to this by coming up with a new, unobjectionable theory that tries to preserve the intuition that made me believe the original theory. Instead, I react by concluding that my physical intuition was unreliable, and shouldn't be trusted.

Moreover, we do have some plausible "debunking" psychological mechanisms accounting for why it seems to you (and many others, including many classical utilitarians like myself) that intense pleasures are so serious, relative to pleasures. Carl Shulman, for instance, has speculated that, due to evolutionary pressures, we should expect creatures like us to be capable of experiencing much more intense pains than pleasures. We could add that, because this difference in intensity is so great, it might appear to be infinite, or "incommensurable", due to our inability to intuitively process large numbers (cf. Broome).

Thus, we have independent reasons to distrust the intuitions that, if taken literally, would result in a theory that can shown to be false. On the whole, then, I think it's far more reasonable to conclude that the intuition that motivates semi-negative utilitarianism is unreliable, and that as such shouldn't be accorded evidential weight in our moral theorizing, than to try to preserve the spirit of the intuition by tweaking the theory until it no longer faces the original objections.

Brian wrote:No, Virginia, there really is no such thing as a utilon in any non-arbitrary sense

This seems to me a very weird claim, at least in the very strong sense (‘any’) in which you’ve phrased it here. Sure, there might not be any identifiable Planck-esque unit of welfare, but you seem to be saying that if I consented, with full information, to a suffering–pleasure trade that only impacted my own welfare, you might overrule it, not on the grounds that you think I’ve made a factual error, but because you just think your caring trumps mine. This seems intuitively bizarre.

It also seems recursive. Do you calibrate your intuitions, for eg? If so, what are you calibrating them against? If there’s some standard, some sense in which your intuitions can be improved, then there must be some sense in which they’re non-arbitrary (in which case we can posit a non-arbitrary utilon even if the unit itself is only a measurement of convenience, like those of the metric system, for eg). If there’s no sense in which they can be improved, why bother informing them at all?

How can it be wrong to go with your gut reaction to any ethical question, if your gut reaction is the only ethically relevant factor in the question?

I'll now address a different aspect of Brian's defense of semi-negative utilitarianism. He writes:

Of course, we can choose to make our assignments more uniform. We can say that very "similar" experiences (e.g., stubbing my big toe vs. stubbing my little toe) shouldn't have vastly different numbers attached. We can try to impose other regularity conditions that seem plausible and jive with our intuitions. But at the end of the day, it's up to us what numbers we pick.

I'm not sure that this is the case. As I wrote in a Facebook comment a while ago:

[I believe] we can make comparisons between levels of suffering and happiness independently of our values. When I introspect, I find that I can say things like "the suffering I experienced this morning was at least two times more intense than the happiness I experienced this afternoon" without making any assumptions about the relative value of these experiences. But [...] semi-negative utilitarians [need not] agree that these comparisons are possible. [...] One way the semi-negative utilitarian could formulate his or her position is by saying that a pain of a given intensity instantiates more negative value than a pleasure of that same intensity instantiates positive value (I believe Jamie Mayerfeld formulates his view in this way). But there is an alternative formulation: the semi-negative utilitarian could simply say that more pleasure is needed to fully cancel the negative value of a given pain than a classical utilitarian would require. The first formulation assumes that there is an "objective phenomenal exchange rate" between pleasure and pain, but the second formulation makes no such assumption.

In case the paragraph above is not sufficiently clear, let me try to make the same point in a different way. Consider this claim:

(1) Positive and negative experiences are "polar opposites" of each other, in the sense that every positive experience has an intensity that is matched by the intensity of a given negative experience. This is not an arbitrary artifact of the way we choose to measure the respective intensities of positive and negative experiences, but is rooted in the nature of the experiences themselves. Just as a certain amount of positive charge is needed to cancel the negative charge of a subatomic particle, so similarly there is a certain intensity of positive experience that will "cancel" the intensity of a negative experience.

Brian denies this claim, and argues that semi-negative utilitarianism should not therefore be defined as the view that

(2) negative experiences are more seriously bad than the corresponding positive experiences are good,

because there is no objective matching between positive and negative experiences, but claims that this position should instead be defined as the view that

(3) negative experiences are more seriously bad, relative to positive experiences, than classical utilitarians claim they are.

As defined in (3), semi-negative utilitarianism doesn't presuppose any objective matching between experiences of different a sign value; rather, it defines itself in opposition to the subjective evaluations of other people (specifically, classical utilitarians).

I agree that, if (1) is false, semi-negative utilitarianism, as defined in (3), would not be more arbitrary than classical utilitarianism. However, if (1) is true and semi-utilitarianism is defined as in (2), then it seems to me that classical utilitarianism has an advantage over semi-negative utilitarianism, since the null hypothesis should be that positive and negative experiences of the same objective intensity are equally valuable, and hence unless an additional, positive reason is provided for deviating from that hypothesis, that is the position we should take.

To sum up the contents of both this and my previous comment, I think there are a number of considerations against semi-negative utilitarianism:

(I) Semi-negative utilitarianism is an attempt to do justice to the spirit of the intuition that originally motivated negative utilitarianism. Since NU can be decisively shown to be false, we should be skeptical of that intuition.

(II) There is a plausible debunking explanation of why people believe in negative and semi-negative utilitarianism.

(III) Semi-negative utilitarianism is more arbitrary than classical utilitarianism.

Brian, I remember that you once mentioned that you would only trade two minutes burning at the stake (say at 1548 degrees C) for at minimum an additional one million years of the best year of your life. Did I recall that accurately and is it still the case?

If so, you're exchange rate is about 20K to 200K times larger than mine. There's nothing meta-ethically wrong with that, of course, but it seems shocking that the differences in our preferences could be so large.

Pablo Stafforini wrote:1. "It seems intuitively that, for pains of sufficient intensity, there is no amount of pleasure that could outweigh them."
2. "So I should become a negative utilitarian."
3. "But wait! When I try to express this intuition formally, I can see that the resulting theory is unsustainable."
4. "Okay, then I should embrace semi-negative utilitarianism instead: a form of utilitarianism that attaches much more--but finite--weight to pains relative to pleasures. Such a theory approximates my original intuition tolerably well, without facing the problems of strict negative utilitarianism."

FWIW, in the old days, I had a somewhat more positive-leaning exchange rate between pleasure and severe pain. So my past intuitions have been in multiple places.

I think the reasoning steps 1-4 above are fine, because why would I choose something more counterintuitive than it needs to be? What I tried to suggest in my post was that there isn't a unique classical utilitarianism at all, so you can't argue for classical utilitarianism on Occam grounds. Ok, maybe you could say that assignments of goodness/badness parameter values that use fewer bits to specify are better on Occam grounds, but there has to be a countervailing factor, or else we're going to end up assigning 0 to everything because it takes the fewest bits to specify. In any event, maybe I don't care about Occam here. As the song says, "It's my party, and I'll use the numbers I want to."

Pablo Stafforini wrote:(3) negative experiences are more seriously bad, relative to positive experiences, than classical utilitarians claim they are.

Yes. Your whole discussion here was an accurate representation of the debate. Of course, I think (1) and (2) are wrong.

Arepo wrote:but you seem to be saying that if I consented, with full information, to a suffering–pleasure trade that only impacted my own welfare, you might overrule it, not on the grounds that you think I’ve made a factual error, but because you just think your caring trumps mine.

First of all, Arepo_now might not care enough about Arepo_later. We already have tons of examples where people do stupid things they later regret. But in any case, even if you were in a calm state of mind and less prone to such things, I would still reserve the right to override your decision.

Arepo wrote:If there’s no sense in which they can be improved, why bother informing them at all?

They can be improved relative to my own (ultimately arbitrary) ideas of what improvement might mean. I think intuitions adopted with more understanding of the world are generally better, so it's good to gain more insight to better inform my intuitions.

Arepo wrote:How can it be wrong to go with your gut reaction to any ethical question, if your gut reaction is the only ethically relevant factor in the question?

There are different kinds of guts. Some intuitions say that others don't matter, and I listen to some over others.

peterhurford wrote:Brian, I remember that you once mentioned that you would only trade two minutes burning at the stake (say at 1548 degrees C) for at minimum an additional one million years of the best year of your life. Did I recall that accurately and is it still the case?

I waffle on the exact amount. Some days it's 1 million, some days it's at least 1000, some days it's more than 1 million. Maybe as low as 100 if I'm feeling brave, but usually not.

Now you baited me to log on again just to respond to this.

In any event, maybe I don't care about Occam here. As the song says, "It's my party, and I'll use the numbers I want to."

The problem is, it's not your party. It would be your party if your actions affected only you, and it might be a more open question if the affected beings couldn't consent one way or the other.

Arepo's trade-offs are not your party. They are Arepo's party.
Peter Hurford's wish not to be wireheaded is not your or Arepo's party.
The consensual sexuality of other humans is not Elijah's party.
The right to die of strangers is not Robert Wiblin's party.
I would add: The painlessness of pigs is not McDonald's customers' party.

Brian Tomasik wrote:First of all, Arepo_now might not care enough about Arepo_later. We already have tons of examples where people do stupid things they later regret.

While it is technically true that Arepo_now is not identical to Arepo_later, note that this is a huge divergence from the way humans psychologically work and the Schelling points that come out of that. Also note the derogatory language ("stupid") which is typical for defenders of aggressive paternalism. Even worse, Brian_now is not Brian_later. Why not call Brian_now's decision stupid and override it?

But in any case, even if you were in a calm state of mind and less prone to such things, I would still reserve the right to override your decision.

The right to override his decision? Is this rightism? No, you're asserting power over him, and as you yourself admit, not only are your own trade-offs fluctuating, but there is no meta-ethical correctness about your judgment. You're both saying that your chosen values are ultimately arbitrary and that might makes right. Ted Bundy, of course, could say the same.

Now I happen to believe that Ted Bundy would not be incorrect. But you would never see me give money, political capital, or practical support to Ted Bundy. Your values are kinder, but this doesn't change the structure - you're openly attacking the autonomy of others, and in practical homo hypocritus reality, this translates to a direct erosion of trustworthiness and subsequent levels of trust. This is one reason why I don't support utilitarianism anymore. (I am also not a moral realist, but I am a hedonistic realist, i.e. I do think that pleasure-pain-tradeoffs are objectively real, and intensities are objectively real things in the brain - however, I don't think openly attacking other people's autonomy is a good idea, ever.)

Brian Tomasik wrote:First of all, Arepo_now might not care enough about Arepo_later.

Assume I’m structuring a trade such that the benefit follows the cost. Then you can’t invoke undervaluing my future self (conceptually you can, since I might arbitrarily value further-future-me more than near-future me, but this is hard to believe) as a reason to overrule me. So then, if you think I’m not actually wrong about anything, you basically have to accept that you’re simply forcing your arbitrary desires on me. Are you happy to bite that bullet?

I think intuitions adopted with more understanding of the world are generally better, so it's good to gain more insight to better inform my intuitions.

This seems pretty weird. On your worldview, value is set by/defined as what you think it is. So understanding the world can’t possibly improve your intuitions, since they’re arbitrary anyway and we reject any non-arbitrary criteria against which we could evaluate them. In short what you’re arguing for isn’t really utilitarianism, but an intuitionism that happens to map very closely onto utilitarianism.

But in that case it doesn’t make sense to appeal to the expected value of information, the way utilitarians normally would – your intuitions will still be there, still the source of all value, whether or not you learn what qualia insects actually experience.

So if it’s good to get more insight, it must be a fundamental good, one which we can weigh against the others in your value system, like pain and pleasure. That is, even where your knowledge has no instrumental value, it has some positive value such that it would be worth removing some amount of pleasure/adding some amount of pain to the world for you to understand it better.

HT! If you haven't already, read our replies to your goodbye post! I want to seduce you into helping us with some project! Negative leaning utilitarian stuff may really take off in the coming months!

Hedonic Treader wrote:The problem is, it's not your party. It would be your party if your actions affected only you, and it might be a more open question if the affected beings couldn't consent one way or the other.

Arepo's trade-offs are not your party. They are Arepo's party.
Peter Hurford's wish not to be wireheaded is not your or Arepo's party.
The consensual sexuality of other humans is not Elijah's party.
The right to die of strangers is not Robert Wiblin's party.
I would add: The painlessness of pigs is not McDonald's customers' party.

This seems like a reasonable kind of preference utilitarianism to me, which I would be very tempted to endorse. The interesting question, however, is when does it become your party? Certainly utilitarianism generally, and even preference utilitarianism, does allow some degree of paternalism. And also some degree of "I won't let you drink from this water despite you being really thirsty and wanting it really badly, because I know it contains a deadly poison and you do not.

Hedonic Treader wrote:Even worse, Brian_now is not Brian_later. Why not call Brian_now's decision stupid and override it?

If Brian_now expects Brian_later to think Brian_now's decision was stupid, Brian_now might override it. But ultimately, by necessity, it's Brian_now's party because Brian_later can't stop him.

Hedonic Treader wrote:however, I don't think openly attacking other people's autonomy is a good idea, ever.)

You make a good point. There may be a big difference between theory and practice. At the end of the day, what I said is honestly the way I see things, but in real life, it may often be better to let people make their own choices. In any event, people are a rounding error compared with animals, and we have to be more paternalistic with regard to animals.

Arepo wrote:Are you happy to bite that bullet?

In theory yes, though maybe less in practice given rule-utilitarian / social-repercussions kinds of considerations.

Arepo wrote:So understanding the world can’t possibly improve your intuitions

I have an intuition that I would like my intuitions better after learning more. So I learn more and like my intuitions more. Nothing magical going on.

Arepo wrote:But in that case it doesn’t make sense to appeal to the expected value of information, the way utilitarians normally would – your intuitions will still be there, still the source of all value, whether or not you learn what qualia insects actually experience.

Intuitions are about values. Information is about facts. Say I value green ping-pong balls. If I learn that the balls I've been hoarding are in fact red, that will displease me. There's high expected value of information in figuring out what color my ping-pong balls are.

Similarly: I value certain kinds of brain operations as "sentience." If I learn that insects don't actually run those operations, I don't care about insects. If they do, then I care about them.

Toby Ord has a new essay, "Why I'm Not a Negative Utilitarian," which Pablo shared on Facebook.

Toby and I chatted about his article last week when he was collecting feedback. At the end of the discussion, Toby said that I was basically a classical utilitarian, although I approximate weak NU. Thus, most of this essay doesn't apply to my views. (Pablo earlier this week shared my own thoughts on negative-leaning utilitarianism.)

I'm fine with Toby's points up through "The worse-for-everyone argument." In that argument, Toby says:

Indeed, suppose that there was a situation in which all individuals want to accept 5 wellbeing units of suffering in order to gain 10 wellbeing units of happiness. This would be in everyone's interest. However, Weak NU would say that it was impermissible

My response is, no, there is no such thing as a wellbeing unit apart from our utilitarian assessment of the situation. There are no utilons except the ones that we project onto what we observe.

Now maybe Toby is thinking of something like von-Neumann-Morgenstern utility derived from preference orderings. But I'm a hedonistic utilitarian, and I don't care intrinsically what people's choices are. People make a lot of choices they later regret. In any event, organisms take many actions that aren't purely hedonically driven. The brain is built from dozens/hundreds of motivational hacks that developed at specific times for specific uses -- like, say, the drive to escape predators even if your life is miserable or the drive to have unprotected sex in the heat of the moment despite later consequences. Animals seem to have a "will to live" independent of hedonic considerations. Post-humans with advanced knowledge of neuroscience could build a mind that would choose to consciously suffer. We don't normally see this because in nature, liking and wanting usually align pretty closely, but in principle, there's no reason a robot couldn't choose to cause itself agonizing conscious experiences.

So it's not the case that everything people choose aligns with a hedonic calculation. And even deciding how much different parts of the hedonic calculation matter requires exchange rates between types of experiences. So I deny these "wellbeing units" of which Toby speaks if they're not just our own degree of caring about the experiences in the first place.

For example, it implies it is generally good if people get murdered, and there is no reason for police or society to do anything to stop this. Indeed it would typically be immoral for them to stop the murderers. Similarly, it is great news if your mother dies. Furthermore, it also implies that much healthcare and lifesaving is of enormous negative value. It says that the best healthcare system is typically the one that saves as few lives as possible, eliminating all the suffering at once. This turns healthcare policy debates on their heads and means we shouldn't be emulating France or Germany, but should instead look to copy failed states such as North Korea.

No, in practice, even a full negative utilitarian would not support such things, because societies of this type would probably have more total suffering. In any event, focusing on present-day humans is not a priority. Real negative utilitarians will be focused on wild animals and future sentients, and doing this means getting along in society without making lots of enemies.

In the "Alternatives" section, Toby lists some intuitions that can tempt people toward NU that are actually non-utilitarian intuitions. I do not share any of these intuitions on reflection; they are not the reason I give severe suffering more weight, and they do not appease that intuition. The only paragraph in that section with which I do identify is the last one, which is a utilitarian argument: "adding happiness to those who are already happy seems to typically lead to less Utilitarian value than using the same resources to alleviate suffering."

----

Finally, I want to point out that I don't lean very negative except for extreme suffering. For ordinary pains vs. ordinary pleasures, I probably agree with most people's assessments about the tradeoffs. I suspect that many people in rich countries do have net-worthwhile lives, because most of them are never tortured, eaten alive, drowned, thrown into a brazen bull, etc. My negative lean only starts to show itself for these really extreme kinds of suffering, which I think are more serious than people assume.

In theory, torture can be outweighed by enough pleasure, but it's important not to confuse thought experiments with real life. The amount of pleasure needed to outweigh torture is immense, and we almost certainly can't create ratios of pleasure to severe pain that imbalanced in the real world. Even with regard to the future, there's no way we can get the probability of massive torture low enough for our utilitronium-shockwave dreams to compensate in expected value.

Hi Brian,

Thanks for you response!

Brian Tomasik wrote:At the end of the discussion, Toby said that I was basically a classical utilitarian, although I approximate weak NU. Thus, most of this essay doesn't apply to my views.

I'm still not quite sure whether you currently believe what I call Weak NU, or what I call Practically-Negative Utilitarianism (if the latter, you seem to be between what I outlined as 'Strong' and 'Weak' versions of it). It would be helpful if you could clarify.

My response is, no, there is no such thing as a wellbeing unit apart from our utilitarian assessment of the situation. There are no utilons except the ones that we project onto what we observe.

If you are just a Practically-Negative Utilitarian (i.e. a Classical Utilitarian with certain empirical views) then it can make sense to deny the concept of wellbeing. Many Classical Utilitarians make use of wellbeing as a concept, arguing first that happiness and the absence of suffering is what makes an individual's life good for them and then aggregating this. But you might be able to skip this intermediate step if you wanted. However, if you believe Weak NU, then I don't see how you could have a coherent view about how a unit of happiness can count for less than a unit of suffering without some independent yardstick of what these 'units' are, and I don't see any other plausible candidates. If you are a Classical Utilitarian, then you don't believe that a unit of happiness counts less than a unit of suffering.

Even if you deny the concept of wellbeing, there is a question of whether the way you match empirically described levels of suffering with empirically described levels of happiness are morally plausible. For example, if you say a pinprick outweighs a year of happy life I would think this is a crazy view. Even if you say that a day of torture outweighs ten happy years of life, I'd find this pretty implausible too, as I'd certainly make that trade for myself.

It would suggest to me that you are illicitly using your tastes as to how bad things are for people instead of theirs (by tastes I mean how much they would enjoy or suffer from different stimulus, not what they believe before they try it). This would be analogous to me saying that feeding people mushrooms is bad for them (and the world) just because I don't like mushrooms. We should use their scales of how much having a particular thing done to them makes them suffer, not ours. I think you are trying to use their scale for what makes them suffer and how much (e.g. do they find mushrooms great, or unpleasant, or vilely unpleasant), then a universal moral scale (which is in your head given your metaethical views) to rate that. If so, I think you might be making a mistake about the intermediate step. If you really think that the only facts of the matter about how much suffering mushrooms cause me compared to how much happiness love causes me are in Brian Tomasik's head, then I find that a very implausible theory and I think most others will too!

No, in practice, even a full negative utilitarian would not support such things, because societies of this type would probably have more total suffering.

I really don't think you have dodged this bullet here. Even with those empirical views about the side effects of murder, you would have to believe, for example, that it is *great* that your mother got murdered and that other people get murdered -- it is one of the best things that can happen to them (for their own sake) -- but that it is a real pity it comes with additional side effects which will outweigh this great blessing. This is a really implausible thing! You may not believe it is true in your country, which might make the bullet easier to bite for you, but it sounds like you would have to believe it of most people in the world, which is still pretty shocking.

In the "Alternatives" section, Toby lists some intuitions that can tempt people toward NU that are actually non-utilitarian intuitions. I do not share any of these intuitions on reflection; they are not the reason I give severe suffering more weight, and they do not appease that intuition. The only paragraph in that section with which I do identify is the last one, which is a utilitarian argument: "adding happiness to those who are already happy seems to typically lead to less Utilitarian value than using the same resources to alleviate suffering."

That's very interesting. Thanks for letting me know.

TobyOrd wrote:Even if you say that a day of torture outweighs ten happy years of life, I'd find this pretty implausible too, as I'd certainly make that trade for myself.

I certainly wouldn't. The question is, are you underestimating how bad one day of torture is, or am I underestimating how good ten happy years of life are, or are our brains really that different from each other that the difference in hedons is actually a real quantitative difference between our phenotypes?

And here is what really scares me about utilitarianism of this sort: The next logical step is to say, if someone believes that they would make this trade-off, they should endorse governments who force this trade-off onto others. And then suddenly, you have a complete game-theoretic paradigm shift where it's no longer about personal choice but about men in uniforms wielding tasers and batons kicking down the doors of unarmed civilians "for their own good" - this seems to invite serious risks of authoritarian negative externalities.

(Btw, if you think this is fiction, look at some of the "suicide intervention" footage on youtube. These are serious human rights violations.)

I really don't think you have dodged this bullet here. Even with those empirical views about the side effects of murder, you would have to believe, for example, that it is *great* that your mother got murdered and that other people get murdered -- it is one of the best things that can happen to them (for their own sake) -- but that it is a real pity it comes with additional side effects which will outweigh this great blessing. This is a really implausible thing!

This interpretation shows the insensitivity of superficial utilitarian thinking to the game theory of individual rights and autonomy: The disutility of tolerating and/or expecting murder is not the prevented life-years, it's the erosion of trust in society. What happens to a society where you have to expect that strangers will stab a knife in your gut when you just want to buy groceries? The externalities of this are considerably higher than the bit of prevented happiness or suffering from a shortened life.

You may not believe it is true in your country, which might make the bullet easier to bite for you, but it sounds like you would have to believe it of most people in the world, which is still pretty shocking.

I've worked with elderly people who openly wished for death, and weren't allowed to die. When they finally did die from natural causes, we (the staff) all inofficially agreed that it was good they finally died, and there was nothing shocking about it. The question is what exact aspects lead you to the conclusion that it is shocking if a person dies and we judge it to be good rather than bad. This is probably not a purely utilitarian intuition.

Brian wrote:You make a good point. There may be a big difference between theory and practice. At the end of the day, what I said is honestly the way I see things, but in real life, it may often be better to let people make their own choices. In any event, people are a rounding error compared with animals, and we have to be more paternalistic with regard to animals.

Yes, inevitably. I also agree there's the risk that people underestimate the suffering of animals or overestimate their happiness. As for the theory vs. practice aspect, you could commit to a simple heuristic never to override people's choices if they seem well-informed and non-whimsical unless the externalities are comparatively huge (and never endorse such overrides). Combined with your heuristic to be honest about your positions, this could avoid a lot of alienation (and prevent interpretation of your position as justifying authoritarian power abuse, which I think causes predictably more harm than good).

peterhurford wrote:Certainly utilitarianism generally, and even preference utilitarianism, does allow some degree of paternalism. And also some degree of "I won't let you drink from this water despite you being really thirsty and wanting it really badly, because I know it contains a deadly poison and you do not.

Yes, clearly. I would have experienced a lot less personal resentment if paternalists were only trying to overcome epistemic deficits rather than forcing their value judgments into the lives of others.

Let me also reiterate that a lot of my resentment comes from the authentic experience that (fake) paternalism is often just a smokescreen for much less benevolent motivations.

Thanks, Toby! It's nice to have you back for a visit on Felicifia.

By the way, I didn't say so in my first reply, but I appreciate the earnestness of your essay. You're aiming to do what you think is a good thing by encouraging some with NU intuitions to second-guess their position. In turn, I think I'm doing a good thing by defending a certain flavor of negative-leaning utilitarianism (even if its resemblance to NU is more practical than theoretical).

TobyOrd wrote:I'm still not quite sure whether you currently believe what I call Weak NU, or what I call Practically-Negative Utilitarianism (if the latter, you seem to be between what I outlined as 'Strong' and 'Weak' versions of it).

Yes, it's a subtle issue. I've refined my own thinking over the years about exactly what I mean by my position.

Weak NU is sometimes the way I talk about my position, but it's not exactly right. Speaking of an "exchange rate" on top of suffering and happiness presupposes there is such a thing as amounts of suffering and happiness independent of the exchange rate. But in fact, suffering and happiness have the magnitude that we decide they have. We assign a suffering value to experiences X, Y, and Z; there's not an independent measure of them, which is the point I made when denying "The worse-for-everyone argument."

On the other hand, it's empirically the case that the negative values I assign to extreme suffering are larger than those assigned by most people. So if I do talk about an exchange rate of E, what I actually mean is that, "relative to the human population's average magnitude of suffering for torture, my parameter assignment is E times bigger." (Of course, the average magnitude of suffering due to torture for the population may be infinite due to some negative utilitarians among us, but we could talk about the median instead.)

So I'm a classical utilitarian who believes that torture-level suffering is actually as bad as the numbers I assign to it, but I don't claim that other people will necessarily agree with my assessment of how bad it is. Empirically it seems I care about it more than most people.

You could almost call this "Strong Practically-Negative Utilitarianism," but I don't agree with the description of that in the article: "Classical Utilitarianism with the empirical belief that suffering outweighs happiness in all or most human lives." I think many people in rich countries probably have net positive lives because they don't typically experience torture-level suffering. If a person lived 75 years of a happy life and then died after being conscious for 5 minutes in a brazen bull, I would say that life was net negative. But most wealthy people never experience anything like a brazen bull in their lifetimes (for which we can be grateful).

TobyOrd wrote:If you are a Classical Utilitarian, then you don't believe that a unit of happiness counts less than a unit of suffering.

Yes, this is my position, but with the proviso from above that I realize that my assessments don't map onto the median assessments, and I'm not troubled by this either, because ultimately, all the numerical hedonic assignments in classical utilitarianism are up to us to decide; they're not facts "out there" in the world.

TobyOrd wrote:For example, if you say a pinprick outweighs a year of happy life I would think this is a crazy view.

No, I'm close to the population median in my views of how bad a pinprick is relative to happiness.

TobyOrd wrote:Even if you say that a day of torture outweighs ten happy years of life, I'd find this pretty implausible too, as I'd certainly make that trade for myself.

A day of torture!?!!!
I wouldn't take even 10 seconds of torture for 10 years of extra happy life.
Maybe we're not being specific enough about what kind of torture. How many minutes in a brazen bull for 10 happy years?

TobyOrd wrote:It would suggest to me that you are illicitly using your tastes as to how bad things are for people instead of theirs (by tastes I mean how much they would enjoy or suffer from different stimulus, not what they believe before they try it).

A few things to say:

1. Your statement assumes that people's hedonic welfare is deducible from their choices. As noted in my first reply, I don't trust people's choices to necessarily reflect their hedonic self-interest. There are evolutionary drives pushing people to prefer existence irrespective of its average hedonic value.
2. There is no unique person making a choice about whether torture is acceptable. If you torture someone hard enough, he will give in. From 1984:
   

   ‘By itself,’ he said, ‘pain is not always enough. There are occasions when a human being will stand out against pain, even to the point of death. But for everyone there is something unendurable—something that cannot be contemplated. Courage and cowardice are not involved. If you are falling from a height it is not cowardly to clutch at a rope. If you have come up from deep water it is not cowardly to fill your lungs with air. It is merely an instinct which cannot be destroyed. It is the same with the rats. For you, they are unendurable. They are a form of pressure that you cannot withstand. Even if you wished to. You will do what is required of you.
   

   
   In a calm, euthymic state, you might decide to go for the torture in return for the happy years. In the throes of torture, you would change your mind and wish you had never made the bargain. Which is the Toby whose preferences we should listen to?
3. We can't get away from making some judgment calls based on our own intuitions. Consider the problem of interpersonal comparisons of utility. Say David Pearce would not accept any amount of happy life in exchange for one minute of a brazen bull. Say a year of happy life is +1. Then does that mean that 1 minute of a brazen bull is -infinity for David? Does this swamp all the happiness any finite number of Tobys could ever experience? No -- it means we need some third party (i.e., me) to decide how to compare the experiences of the two people. Now, you could insist that the utilities be normalized in such a way as to respect each person's individual choices, so that you would only need this third-party arbitration among people. But hey, Toby_now and Toby_under_torture are two different people, and Toby_under_torture would say that no amount of happiness could outweigh what he's experiencing now. So even here, we need a third party arbiter to decide. In other words, there's no getting away from the kind of arbitration that I'm doing. People do not have unique, stable preferences about these tradeoffs. (This is seen even in the fact that I change my own third-party-arbiter sentiments depending on my mood.)

TobyOrd wrote:you would have to believe, for example, that it is *great* that your mother got murdered and that other people get murdered -- it is one of the best things that can happen to them (for their own sake) -- but that it is a real pity it comes with additional side effects which will outweigh this great blessing.

No, because (1) as noted, many people in rich countries have net positive lives and (2) even if the person's life weren't positive, murder sounds a lot more painful than death in a hospital with pain medicine, and considering that the pain of death is one of the worst things anyone experiences, it's much better to wait a few years for a more humane death.

TobyOrd wrote:You may not believe it is true in your country, which might make the bullet easier to bite for you, but it sounds like you would have to believe it of most people in the world, which is still pretty shocking.

I haven't studied the average quality of life of people in other countries. My guess based on the hedonic treadmill is that their lives aren't too much worse, but it might depend a lot on the details (e.g., more likely that life is net positive for independent farmers than for slave laborers who are beaten, etc.).

In any event, it's clear that your argument about murder is unfair because of the emotions it arouses without justification. For example, you surely must think there are people who would be better off if they didn't exist (e.g., people with severe mental illnessess causing their lives to be absolutely agonizing). I could then argue that you think it would be *great* if those people got murdered, ignoring the side-effects of the murder. This gives an unfair tone to what is otherwise a reasonable position because it's hard to shake off the "side-effects of murder" from the way we're imagining it. Those people getting murdered without the side-effects of murder basically means those people being allowed to euthanize themselves in peace, which is a far different thing to imagine.

TobyOrd wrote:That's very interesting. Thanks for letting me know.

Hedonic Treader wrote:The next logical step is to say, if someone believes that they would make this trade-off, they should endorse governments who force this trade-off onto others.

In some sense, Toby is forcing David Pearce to be tortured against his will. This follows from Toby's choice of a non-extreme exchange rate, leading him to support expansion of humanity into the universe, even though the result will be decillions of David Pearce-like sentients enduring torment that they would not have chosen even for the corresponding pleasure that would be created. You can criticize me for my authoritarian exchange rate, but Toby is authoritarian as well. He's implicitly forcing David Pearce to be tortured against his will because of Toby's own preferences. (Sorry, Toby! I mean this in the sweetest way possible. )

Hedonic Treader wrote:What happens to a society where you have to expect that strangers will stab a knife in your gut when you just want to buy groceries? The externalities of this are considerably higher than the bit of prevented happiness or suffering from a shortened life.

Yeah, I think Toby was trying to minimize this by saying "it is a real pity it comes with additional side effects which will outweigh this great blessing," but the analogy is too distant to be fair.

Hedonic Treader wrote:I've worked with elderly people who openly wished for death, and weren't allowed to die. When they finally did die from natural causes, we (the staff) all inofficially agreed that it was good they finally died, and there was nothing shocking about it.

It's great you have firsthand experience! I've watched some documentaries on dying, and I definitely want euthanasia when the time comes.

Hedonic Treader wrote:As for the theory vs. practice aspect, you could commit to a simple heuristic never to override people's choices if they seem well-informed and non-whimsical unless the externalities are comparatively huge (and never endorse such overrides). Combined with your heuristic to be honest about your positions, this could avoid a lot of alienation (and prevent interpretation of your position as justifying authoritarian power abuse, which I think causes predictably more harm than good).

You make a very interesting suggestion. I'll take this seriously. In practice, most questions about how much suffering negates how much pleasure matter with respect to not-yet-created future minds, and in that context, authoritarian implications aren't so salient. We have to decide for them because they literally cannot decide for themselves. In any event, we're not talking about whether people can make their own choices. We're talking about whether we make a macro-scale decision about computational power in the future. This is not about personal choice; it's about social choice.

Brian Tomasik wrote:In some sense, Toby is forcing David Pearce to be tortured against his will. This follows from Toby's choice of a non-extreme exchange rate, leading him to support expansion of humanity into the universe, even though the result will be decillions of David Pearce-like sentients enduring torment that they would not have chosen even for the corresponding pleasure that would be created.

Perhaps Toby would reply that David is misinformed about his true preferences. If David thought more clearly about the situation, he would realize that a day of torture could be outweighed by 10 years of happy life. Well, in this case, Toby has gotten into the business of making judgment calls about people's hedonic balances that go against their own firmly held, seemingly cool-headed beliefs about what they themselves would want. This is what Toby criticized me for doing.

Toby has two choices:

1. Admit that he's "torturing David Pearce against his will" in the sense that he's supporting policies that will create people like David Pearce who will be tortured without an infinite amount of compensating happiness for themselves or other organisms. (I'm including the "other organisms" caveat because utilitarianism often makes tradeoffs that hurt some in order to benefit others. But my point is that according to David, the amount of benefit to others is not sufficient, in the sense that even if all the benefit went to himself, he would not accept it.)
2. Deny that David Pearce's true will is actually to give up arbitrary amounts of happiness to avoid torture, in which case Toby is claiming he knows better than the informed choice of his wise friend.

Of course, I'm in the same boat. We all are. I'm just showing that the corners that Toby has put me into are corners that he is pushed into as well.

Final remark from me:

Brian Tomasik wrote:ultimately, all the numerical hedonic assignments in classical utilitarianism are up to us to decide; they're not facts "out there" in the world.

Well, something has to be "out there" in the world. Pleasantness and unpleasantness are clearly real and their intensities are clearly encoded somehow. Maybe you are doubting that they can be coherently abstracted as cardinal numbers - maybe because they are too complex patterns, or different modes of experience that don't have "the same unit", if you will. But we use cardinal numbers to express other complex patterns (e.g. GDP, which expresses wealth that is clearly "out there" in the world). And it seems that at least for simultaneous stimulation, activations of certain pains and pleasures can cancel each other partially or fully (measured by the behavoir, including self-assessment).

Let's say you're right and there's nothing "out there" in the world that could be expressed in hedons. But then why bother at all? Why not just say, whatever happens, happens. I could surely act indifferently to whether I enter a Brazen Bull and then if I do, just resolve the situation by screaming a lot. If the intensity encodings aren't "out there", i.e. real things in the brain, why should I care about them cognitively?

Brian Tomasik wrote:My response is, no, there is no such thing as a wellbeing unit apart from our utilitarian assessment of the situation. There are no utilons except the ones that we project onto what we observe.

Now maybe Toby is thinking of something like von-Neumann-Morgenstern utility derived from preference orderings. But I'm a hedonistic utilitarian, and I don't care intrinsically what people's choices are.

There's something funny going on here. You think utilons are an arbitrary construction determined by how much we care about the experiences in question. You also think there's nothing fundamentally important about people's preferences, and even that they can be mistaken, as judged by how much welfare they bring the person - so it seems as though the value of a choice would boil down to the welfare(utilons) it brings, and by transitivity, to how much we care about the experiences.

But now we seem to have gone in a circle. What is this 'caring' thing you speak of, if not something resembling a preference or the utilon values of generating possible outcomes, and why might anyone - you included - heed it? The only fundamental difference I can see between it and a regular preference is that on your ethic it refers only to *your* preference.

Obviously (almost by definition) you'll struggle to persuade others to heed such a personal ethic, though I think that probably doesn't matter to you, since you're only considering your marginal decisions?

What seems more odd is that it being your ethic doesn't really seem relevant to any of the objections you yourself raise, and if not it's clearly circular. If it's basically a preference, then just because it's your preference you presumably wouldn't claim that the possibility of it being misguided falls to nil. So does it relate to the amount of utilons the ideas of other people's suffering cause you? Presumably not, since utilons don't exist in your view. So by transitivity it boils down to how much you care about how much you care matters.

I don't think circularity is instant disproof of an epistemological(ish) principle - at some level we have to either assert or self-justify (if there's any difference between the two). But what makes this axiom of Brian's caring (aBc) seem dubious to me is that once you exclude from its recipe either utilons or preference it seems totally incommunicable even in principle.

If someone says to me 'I'm experiencing some hedons right now', I might not be able to recreate their experience well, but I can run some approximate simulation of it in my own head, experiencing, or remembering experiencing hedons. If someone says 'I'd prefer coffee to icecream' I might think both that their taste is repugnant and that this 'preference' thing they're talking about is reducible to something like hedons and anticipation of hedons, but I can still just about parse it well enough to imagine coffee preference. If someone modus ponens to me I can't exactly empathise with it, but I can at least simulate the process it describes in my head an observe that it tends to produce the predicted results.

aBc has no such properties. If Brian Tomasik tells me X is important because he cares about it, I can't simulate hedonic anticipation, or take Y and run any kind of function that I've been informed about to feed out X.

In a calm, euthymic state, you might decide to go for the torture in return for the happy years. In the throes of torture, you would change your mind and wish you had never made the bargain. Which is the Toby whose preferences we should listen to?

From a pragmatic perspective, why not the one who's experienced both? It seems like a lot is riding here on you and Toby's guesstimation of the intensity of a feeling neither of you have ever experienced. Also, how much utility difference is there between 1) 10 minutes in a brazen bull with the assumption that either you'll die from this or survive it and be dumped in the uncaring world along with your injuries to fend for yourself, and 2) 10 minutes in a brazen bull when you've been reliably informed by a source you have near complete confidence in that any damage up to and including brain death will be perfectly reversed on your exit, and you'll then be guaranteed 10 years of the happiest life you can imagine?

I get the impression you're envisaging 1, while Toby is thinking about 2 (and perhaps still will continue to do so even after contemplating it), but the difference in these subjective experiences could be huge (cf Camus 'The Myth of Sisyphus', claiming that Sisyphus' fate, which the original authors presumably picked as one of the worst physical outcomes they could imagine, would actually leave him happy after proper contemplation).

Hedonic Treader wrote:The next logical step is to say, if someone believes that they would make this trade-off, they should endorse governments who force this trade-off onto others. And then suddenly, you have a complete game-theoretic paradigm shift where it's no longer about personal choice but about men in uniforms wielding tasers and batons kicking down the doors of unarmed civilians "for their own good" - this seems to invite serious risks of authoritarian negative externalities.

I think you're ignoring your own logic from further down in this post here; the primary hedonic impact in the real world of stormtroopers kicking down doors and forcing people to what they say isn't the number of hedons that switch values at the scene where the SA actually wirehead/torture you for a day then give you the happiest life possible for 10 years, it's the social impact caused by living in such a society in fear of something you perceive as very bad happening to you (not to mention the long-term economic impact of making everyone very happy for 10 years and ensuring they're tortured first - imagine just the bureaucracy!).

I don't know why you've suddenly decided sane utilitarians can't look at these situations and judge that actually, in practice, they're worth avoiding. Yes, you or I can construct a thought experiment in which I will think the best outcome is for me to hold Peter down and plug him in, but if you flesh it out, you'll see it's so obviously removed from the real world as to be basically irrelevant. I seriously doubt Peter feels threatened by reading this, so why would you feel threatened on his behalf?

In the future, as separateness of people becomes less embedded a concept in society it's conceivable that some form of coercive happiness might become more appealing, but this will be a very different world, in which a) our intuitions from modern society will be ill-adjusted to imagine what this will be like, b) the more separateness of people breaks down, the less it can make sense to be concerned with one person making decisions for another (and the less anxiety society will feel from it) and c) the more separateness of people is replaced by deflationary personal identity, the more evident it will become that present-me is as capable as anyone of unkindly treating future-me.

Arepo wrote:I don't know why you've suddenly decided sane utilitarians can't look at these situations and judge that actually, in practice, they're worth avoiding.

Honest answer to an honest question: I don't believe most people are sane.

Put less crudely, I have already observed this kind of "let's harrass them for their own good" logic more than once in explicitly self-identifying utilitarians such as Robert Wiblin (e.g. gun control so that people no longer have suicide methods), and I have observed more than once that clearly non-benevolent people use fake pseudo-utilitarian justifications to sanctify almost arbitrary aggression. Should governments steal wealth from people and use it for symbolistic useless projects, destroying incentives to invest? A sane utilitarian would say no, but this requires an honest analysis with actual benevolent intentions by the relevant power players. That is not what causally happens in the actual world when you increase social acceptance of utilitarian-type justifications.

In the future, as separateness of people becomes less embedded a concept in society it's conceivable that some form of coercive happiness might become more appealing, but this will be a very different world, in which a) our intuitions from modern society will be ill-adjusted to imagine what this will be like, b) the more separateness of people breaks down, the less it can make sense to be concerned with one person making decisions for another (and the less anxiety society will feel from it) and c) the more separateness of people is replaced by deflationary personal identity, the more evident it will become that present-me is as capable as anyone of unkindly treating future-me.

Of course, it is completely open to conjecture whether the dynamic replicator-entities in this future will maximize hedons at all, or something else entirely. But let's not enter this speculative discussion here.

Arepo wrote:But now we seem to have gone in a circle. What is this 'caring' thing you speak of, if not something resembling a preference or the utilon values of generating possible outcomes, and why might anyone - you included - heed it? The only fundamental difference I can see between it and a regular preference is that on your ethic it refers only to *your* preference.

Yes, the difference is that it's just my preference.
But as I tried to show, this will always be the case. There has to be a third-party arbitrator to do interpersonal comparisons between David's and Toby's happiness/suffering. Similarly, you make all kinds of arbitrary decisions that are just your opinion, like whether you want to care about thermostats' preferences. I don't, and you don't, but maybe the thermostats would prefer that. It's you personally who is making this judgment call. If you think thermostats don't have preferences like this, remember that any preference we assign to an agent is part of the intentional stance and doesn't actually exist in a Platonic way. Preferences are what we think they are.

Arepo wrote:So does it relate to the amount of utilons the ideas of other people's suffering cause you?

No, although as you say, I don't believe utilons exist objectively, but even if I were to assign them for my own thinking about others' suffering, the utilons there wouldn't align with my caring. My caring is influenced by this somewhat, but it's subject to other factors, like my desire for linear scaling with number of organisms, etc.

Arepo wrote:So by transitivity it boils down to how much you care about how much you care matters.

I have to admit, I'm kind of lost by your argument here. : P

Arepo wrote:aBc has no such properties. If Brian Tomasik tells me X is important because he cares about it, I can't simulate hedonic anticipation

You mostly can, because as an empirical fact, my caring about tends to correlate with what people generally think of as hedonics. Hedons don't objectively exist, but we can at least compare what I care about with what most people mean when they use that word.

Arepo wrote:From a pragmatic perspective, why not the one who's experienced both? It seems like a lot is riding here on you and Toby's guesstimation of the intensity of a feeling neither of you have ever experienced.

Yes, that's actually a great suggestion. It's not perfect because when euthymia returns, it can give a rosy-retrospection gloss over the past suffering, but indeed, this would still be a very good source of data. Can you think of someone who has been tortured and would be able to make such an assessment? It would be best if it's a hedonist who doesn't have ancillary ideas about the value of life, religion, etc.

Arepo wrote:Also, how much utility difference is there between 1) 10 minutes in a brazen bull with the assumption that either you'll die from this or survive it and be dumped in the uncaring world along with your injuries to fend for yourself, and 2) 10 minutes in a brazen bull when you've been reliably informed by a source you have near complete confidence in that any damage up to and including brain death will be perfectly reversed on your exit, and you'll then be guaranteed 10 years of the happiest life you can imagine?

I would guess almost no difference between them. These higher-level thoughts are swamped when the pain gets bad enough.
If you want to make the thought experiment precise, pretend that you're bargaining to experience #1, but then also, you forget about that during the 10 years of happy life so as to avoid PTSD, etc.

Hedonic Treader wrote:

Brian Tomasik wrote:ultimately, all the numerical hedonic assignments in classical utilitarianism are up to us to decide; they're not facts "out there" in the world.

Well, something has to be "out there" in the world. Pleasantness and unpleasantness are clearly real and their intensities are clearly encoded somehow. Maybe you are doubting that they can be coherently abstracted as cardinal numbers - maybe because they are too complex patterns, or different modes of experience that don't have "the same unit", if you will. But we use cardinal numbers to express other complex patterns (e.g. GDP, which expresses wealth that is clearly "out there" in the world).

Sure, GDP is a good example: It's based on real things, but the measure itself is one that we make up, and we can modify its definitions to accord better with our intuitions. Maybe we want to start putting Wikipedia into GDP. Maybe we don't want to count paid housework as GDP but unpaid housework as not. And so on. We could even come up with more radical revisions, like Gross Happiness Index measures or whatever.

Hedonic Treader wrote:And it seems that at least for simultaneous stimulation, activations of certain pains and pleasures can cancel each other partially or fully (measured by the behavoir, including self-assessment).

Yes, but we only consider behavior / self-assessment to the degree we want to. In any event, when the torture is bad enough, there's nothing that cancels it out at that moment.

Hedonic Treader wrote:Let's say you're right and there's nothing "out there" in the world that could be expressed in hedons. But then why bother at all? Why not just say, whatever happens, happens.

I choose to.
Put it this way: Suppose hedons were real. Why would you care about them then? Iron and copper atoms are real too, but you're not trying to maximize iron atoms and minimize copper atoms. Why not?

I pointed a friend to my discussion about how in the throes of torture, nothing else seems to matter. She replied:

Interesting. Something I've thought about myself as well. Last night I dreamed that I was watching an old woman be tortured to death and knew I was next. In the dream I reflected that pleasure was meaningless compared to the profound horror of pain. Interestingly it was one of the rare dreams in which I could feel. As I was tortured (my bones were broken), I actually felt pain, though not equal to the true pain of those experiences.

Brian Tomasik' wrote:There is no unique person making a choice about whether torture is acceptable. If you torture someone hard enough, he will give in.

Is this so? I was under the impression that there are known cases of people who refused to incriminate others in spite of being brutally tortured. A while ago, this issue came up in a private conversation I had with Dave, and my recollection is that he admitted that there were such cases.

Pablo Stafforini wrote:Is this so? I was under the impression that there are known cases of people who refused to incriminate others in spite of being brutally tortured. A while ago, this issue came up in a private conversation I had with Dave, and my recollection is that he admitted that there were such cases.

Thanks, Pablo. I don't know for sure about the details, but even if this were the case, it doesn't negate the larger point. The only reason we're wondering whether Toby would give in after enough torture is because we want to assess how bad torture is relative to other things for organisms as a whole. Even if Toby is not one, there are many organisms that would say they'd give up all the pleasure in the universe to make the pain stop during bad enough torture. There are some like David who maintain such a position even when not tortured. We are thus forced into the position of arbitrating between inconsistent preferences both between people and even between the same person at different times.

So the objection to my position that my suffering-to-happiness weighting violates people's preferences may be true, but every suffering-to-happiness weighting violates someone's preferences about his own wellbeing -- indeed, violates most people's preferences when they're being tortured.

Another friend wrote to me in response to my post, saying that really extreme suffering can change your outlook on this question. In reply, I said:

Yeah, I sometimes wonder if one reason more people don't take suffering as seriously as I do is because they haven't experienced the same intensity of suffering as I have at times. I won't claim mine has been Brazen Bull-level, but I did have some pretty extreme pain during my high-school years when I had esophagitis (see the second half of this post if you're curious). Combined with a few other bad periods, I think I've suffered more than most people in my socio-economic and age demographic. (I might still be on the low end relative to the world population, though.) In the last few years, I've had basically no intense suffering at all, and if I had gone through my whole life like this, I might not even know what was possible in terms of how bad it could get. (I still probably don't, but I have some adumbrations of it. For a few fleeting seconds during the esophagitis, I literally felt like I was in a torture chamber.)

Hedonic Treader wrote:this kind of "let's harrass them for their own good" logic more than once … e.g. gun control so that people no longer have suicide methods … governments steal wealth from people and use it for symbolistic useless projects, destroying incentives to invest? A sane utilitarian would say no

Hold on, you’ve just conflated a position that almost everyone, utilitarian or otherwise, would disagree with (gun control to prevent suicide) with one that loads of people, utilitarian and otherwise would agree with, assuming from this post and your leaving reference to libertarianism that by ‘[symbolic] useless projects’ you’re referring to core aspects of the welfare state like state healthcare, education etc. If you’re asserting the value of minarchy so vociferously that you consider anyone who doesn’t think peeling back the state solves everything ‘insane’ then a) the educated public at large would probably agree more with utilitarians than your approach and b) on probabilistic grounds given that you’re convinced you’ve basically solved politics, it seems more likely that your views are informed by strong personal bias than a piercing non-quantitative insight.

Brian wrote: Yes, the difference is that it's just my preference.
But as I tried to show, this will always be the case. There has to be a third-party arbitrator to do interpersonal comparisons between David's and Toby's happiness/suffering.

I think we can simplify this claim to say that what we’re really talking about is motivation, rather than some mysterious realm of ethics. I agree that discussing ethics as something irrelevant to motivation is spurious, but your claim seems to entail two parts: a1) nothing about ethics is worth pursuing except (by definition) what motivates us, and b1) there are no non-arbitrary lines to be drawn or even comparisons to make about experience.

So I agree with a1), but think b) is clearly false. As an analogy consider a2) nothing about physics is worth pursuing except (by definition) what motivates us, and b2) there are no non-arbitrary lines to be drawn or even comparisons to make about gravitational effects. a2 might be more controversial than a1, but I think it’s still basically true, while b2 is clearly false.

So far as I can see, your only argument for b1 is the assertion that it follows from a1. Perhaps the analogy will convince you otherwise (or at least make it clearer why I think your apparently parsimonious view invokes more conceptual entities than it seems).

But more fundamentally than trying to change your intuition, what I’m trying to argue is that your own view entails b) being false, and so is effectively contradictory.

the utilons there wouldn't align with my caring. My caring is influenced by this somewhat, but it's subject to other factors, like my desire for linear scaling with number of organisms, etc.

My challenge to you, to show that I’m wrong about you tacitly treating b) as false is not to show what your caring is influenced by, but to show what it fundamentally is. If it’s not another name for (some set of concepts including and possibly comprising) hedons or preferences, what qualities does it have to show the distinction?

Here’s some things I understand about your caring:

i) it entails a mental state (yours)
ii) it comes in different intensities (representable as both positive and negative?), such that you can care about two things and be able to prioritise between them based on expected outcomes
iii) it is something which is visible to other people only via its effect on your behaviour

That’s about it. Assuming that’s not a wild misrepresentation of the traits of ‘you caring’, it seems that b1 above would entail the claim that at least ii) here holds only for you, which goes beyond metaethical scepticism into solipsism.

I think it also implies that wireheading you against your will to falsely believe that no-one was suffering would be as good an outcome on your own account as preventing their suffering. Sure, beforehand you might object (though I think doing so would contradict your views), but we’d be trading pre-wirehead-Brian’s caring-thwartation for much less post-wirehead-Brian’s caring-thwartation.

Can you think of someone who has been tortured and would be able to make such an assessment? It would be best if it's a hedonist who doesn't have ancillary ideas about the value of life, religion, etc.

I do know one of the institutes in Oxford conducts pain research, obviously seeking ways to cause it that don’t inflict any long-term harm (something along the lines of injecting capsaicin just under the skin IIRC), and presumably there are others like them. I could find out the details if you wanted to research this further?

Arepo wrote:b1) there are no non-arbitrary lines to be drawn or even comparisons to make about experience. [...] As an analogy consider [...] b2) there are no non-arbitrary lines to be drawn or even comparisons to make about gravitational effects.

No, I wouldn't claim b1. There are plenty of objective facts about "experience." When Alice hits her hand with a hammer, it triggers more nociceptive neural firings than when a feather falls on her finger. Right now, Bob would choose to accept a stubbed toe in return for an extra 10 minutes of listening to his favorite symphony. Charles is feeling lousy and would rather not exist for these moments.

What I would claim is that there's not an objective fact about how much I value an experience in my motivational system -- except for the trivial sense in which I am a machine embedded in physics whose thoughts are themselves facts about the universe. I mean there are no external requirements for me to choose -8 instead of -12 for the severity of a bruised arm, even though there are external requirements for me to choose acceleration as 9.8 m/s^2 if I want to get the right answers. There are no "right answers" for our motivations.

Arepo wrote:My challenge to you, to show that I’m wrong about you tacitly treating b) as false is not to show what your caring is influenced by, but to show what it fundamentally is. If it’s not another name for (some set of concepts including and possibly comprising) hedons or preferences, what qualities does it have to show the distinction?

My caring is preferences.

Arepo wrote:ii) it comes in different intensities (representable as both positive and negative?), such that you can care about two things and be able to prioritise between them based on expected outcomes

Well, if we're going to assign numerical intensities, they have to come from either (a) a von-Neumann-Morgenstern-type utility function (if it exists at all) -- something that tells what the choices would be or (b) someone's subjective gut feelings about what the numbers should be. The intensities don't exist "out there" as numbers. There are just algorithms my brain follows, and someone external has to attach numbers to that.

Arepo wrote:I think it also implies that wireheading you against your will to falsely believe that no-one was suffering would be as good an outcome on your own account as preventing their suffering. Sure, beforehand you might object (though I think doing so would contradict your views), but we’d be trading pre-wirehead-Brian’s caring-thwartation for much less post-wirehead-Brian’s caring-thwartation.

No, because it's not my preference to be wireheaded, and my caring is an instantaneous assessment -- it's not the sum-total of caring over the future. It's the caring right now. It's the value of the function, not the integral over future time.

Arepo wrote:I could find out the details if you wanted to research this further?

Interesting. That specifically doesn't sound like the pain would be severe enough. I'm thinking of suffering at the level of extreme torture. Researchers wouldn't be allowed to induce this on subjects because the emotional pain alone (PTSD or whatever) could cause permanent damage.

I have to admit, I'm getting confused about what we're discussing/disagreeing on.

You seem to be discussing what motivates you to act.

I think we agree that 'ethics' in any metaphysical sense is irrelevant to this. So far so good.

You've now identified what motivates you to act as your preferences. I think this is probably false on any definition of 'preference' which doesn't entail 'that which is revealed by your actions' - I'm fairly sure I've seen you admit to akrasia. This feels like a quibble, but I wonder if it's not relevant to our broader disagreement. It seems like what I'm asserting exists might lie in the middle ground between your actual behaviour and the preferences you believe you have.

Brian Tomasik wrote:I mean there are no external requirements for me to choose -8 instead of -12 for the severity of a bruised arm

Suppose you'd bruised your arm, and assigned it -8, then when the pain disappeared you assigned it -12 (and imagine that you gave eating an icecream a constant +5 throughout this thought experiment). On bruising your arm again, you reverted to assigning it -8. So during the pain, if I gave you two icecreams, you'd feel happy on balance, but if I offered to trade you a bruised arm for two icecreams when you weren't in pain, you'd turn me down. But then if I just bruised your arm and gave you two icecreams (in a universe where this has no further effects natch), you'd feel better for it.

You would claim that pain-free Brian hasn't made a mistake here, I think?

Too coldy to parse this well at the moment, but I wonder if a source of our disagreement is over whether/how a mind is reducible. You claimed somewhere that you think a brain split into two minds would double in utility value, presumably implying that 'a mind' is a fundamental unit of experience.

This seems totally alien to me - I think of the mind as something like a vast collection of units of experience who've agreed to trick me into thinking they're unitary because doing so has survival benefits On this account, experiences are very unlikely to *ever* be wholly good or bad, only aggregate to one or the other. In which case there's nothing fundamentally different between aggregating all the experiences in one brain and aggregating all the experiences across several brains, except that the former process has evolved to provide more direct feedback, which might be self-reinforcing (believing I'm suffering might in turn make me suffer more).

If I try to simulate discarding my view and taking up yours, inasmuch as I'm able to do so, your view does seem to make more sense - if minds, then it's not obvious why experiences should be, and it becomes harder or even impossible to draw analogies between the composition of one brain and the composition of a set of brains.

Maybe you can try to simulate the reverse? Then if you find your simulation agreeing with me, I think we'll have refined our point of discussion.

Interesting. That specifically doesn't sound like the pain would be severe enough. I'm thinking of suffering at the level of extreme torture. Researchers wouldn't be allowed to induce this on subjects because the emotional pain alone (PTSD or whatever) could cause permanent damage.

I think you're begging the question here. They're not allowed to cause permanent damage, but they often don't know in advance how painful an experience is going to be. Since they don't have any brazen bull survivors to compare against, it seems quite plausible to me that they are doing as much harm, for much shorter periods, with the mental safeguards I discussed above modifying the mental state that the pain provokes. I can anecdotally recount that one of the test subjects, who was renowned for his tolerance to pain (or at least to the same pain stimuli as others) described one experience as '10/10'.

Arepo wrote:You've now identified what motivates you to act as your preferences. I think this is probably false on any definition of 'preference' which doesn't entail 'that which is revealed by your actions' - I'm fairly sure I've seen you admit to akrasia.

Yes. My preferences are what I care about, but they're not necessarily what I do.

Very interesting example with the bruised arm and ice cream. I would say that, while Brian when not experiencing the bruising and ice cream is allowed to assign whatever numbers he wants, I would encourage him to assign numbers closer to representing how he feels during the experiences (namely, he prefers that over unconsciousness for those moments). This is intuitive because Brian during the experiences seems to be in a privileged position in the sense that he's experiencing both things rather than neither. On the other hand, usually these tradeoffs don't involve a privileged position: For example, if in a peaceful state you sign yourself up for torture in return for happiness, but then during torture, you wish you hadn't done so, it's not clear which self to trust more, because neither one is experiencing the other emotion at the same time.

Arepo wrote:If I try to simulate discarding my view and taking up yours, inasmuch as I'm able to do so, your view does seem to make more sense - if minds, then it's not obvious why experiences should be, and it becomes harder or even impossible to draw analogies between the composition of one brain and the composition of a set of brains.

Yeah. Fundamentally, my position says that feelings are not discrete units but are pictures that we paint when we squint at what a brain does. If I held your view that emotions could be associated with micro-level units that add together linearly, then I would see emotional intensities as being basically absolute, because they'd be fixed by how many of those micro units there were.

Arepo wrote:I think we'll have refined our point of discussion.

What a gentleman you are, Arepo.

Arepo wrote:I can anecdotally recount that one of the test subjects, who was renowned for his tolerance to pain (or at least to the same pain stimuli as others) described one experience as '10/10'.

Yikes! Who would ever sign up for that research?

Elijah wrote:What about scaphism?

It's not as bad per minute as brazen bull. Because my aversion increases really steeply with the per-second intensity of the experience, I might prefer a whole sequence of scaphism over a whole sequence of brazen bull. Of course, I'm not signing up for either.

A case for optimism: Let's say, conservatively, the future will see a technology that allows the safe induction of non-addictive pleasure of at least 10% the intensity of our current worst known agony. Let's say this happens with only 10% probability, but if it happens, it will probably be wide-spread.

Assume that people will use this pleasure-inducing technology as often as they now watch TV, listen to mp3s, or have non-reproductive sex. That's a lot of seconds of pleasure per second of agony, assuming a distribution like on our planet. Even wild animals have far more seconds of sex and play than seconds of intense agony, on average.

Furthermore, agony will be much rarer or go away altogether in scenarios in which even the mildest pain-reducing hedonic augmentation technologies are widely applied. Given the above numbers, you would only need a ratio of 100 seconds of this pleasure to 1 second of agony, which seems entirely reasonable. (This is an argument in addition to the higher expected quantity of hedonium than dolorium)

If you think the probability is too optimistic, or if you value suffering intrinsically more than pleasure, you can use other numbers. But the case may still stand, since 100:1 of "efficient entertainment" seconds per 1 agony second is really very conservative.

If you care about intensity non-linearly, you could consider only the probability of scenarios with artificial pleasures that are as intense, or even more intense, than our current agony, and consider their relative distribution.

I did write a brief response to Toby on Facebook....
___________

Toby, a few thoughts...

1) World destruction? You write, "a thorough going Negative Utilitarian would support the destruction of the world (even by violent means") No, a thoroughgoing _classical_ utilitarian is obliged to convert your matter and energy into pure utilitronium, erasing you, your memories and indeed human civilisation. By contrast, the negative utilitarian believes that all our ethical duties will have been discharged when we have phased out suffering. Thus a negative utilitarian can support creating a posthuman civilisation animated by gradients of intelligent bliss where all your dreams come true. By contrast, the classical utilitarian is obliged to erase such a rich posthuman civilisation with a utilitronium shockwave. In practice, I don't think it's ethically fruitful to contemplate destroying human civilisation, whether by nuclear Doomsday devices or utilitronium shockwaves. Until we understand the bounds of intelligent agency. the ultimate sphere of responsibility of posthuman superintelligence is unknown. Quite possibly, this ultimate sphere of responsibility will entail stewardship of our entire Hubble volume across multiple quasi-classical Everett branches, maybe extending even into what we naively call the past (cf. http://arxiv.org/pdf/quant-ph/0105101v2.pdf). In short, we need to create full-spectrum superintelligence.

2) Negative utilitarians can (and do!) argue for creating immense joy and happiness. Indeed, other things being equal, negative utilitarians are ethically bound to do so, For the thought of a painless but joyless world strikes most people as _depressing_. Negative utilitarians are committed to phasing out even the faintest hint of disappointment! The prospect of a painfree life without peak expediences - mere muzak and eating potatoes - sounds bleak. If a thought or deed causes distress, then other things being equal, that thought or deed is _not_ negative utilitarian.

3) You write, "Absolute NU is a devastatingly callous theory". No: NU, is a ridiculously compassionate theory. You are "callous" only if you are indifferent to someone's suffering, not if you don't act to amplify the well-being of the already happy or act to create happiness de novo.- although in practice, negative utilitarians should promote intelligent superhappiness too. Inducing sadness or disappointment is not NU.

[I think the force of your example depends on an untenable metaphysics of personal identity. If instead we use a more empirically supportable ontology of here-and-nows strung together in different sequences thanks to natural selection, no one is _harmed_ by waking up happy in the morning rather than superhappy like their namesake the night before. This is really an issue of population ethics, normally reckoned a different topic.)

Let us compare the callousness / compassion of classical utilitarianism and NU.
Since we're doing thought-experiments, imagine if a magic genie offers me superexponential growth in my bliss at the price of exponential growth in your agony and despair. If I'm a classical utilitarian, then I am ethically bound to accept the genie's offer. Each year, your torment gets unspeakably worse as my bliss becomes ever more wonderful. Indeed, the thought I'm ethically doing the right thing increases my bliss even further! By generating so much net bliss, I'm the most saintly person who ever existed! If you knew how incredibly superhumanly, wonderful I'm feeling, then you'd realise that my superbliss easily offsets your tortured despair. Your tortured despair is a trivial _pinprick_ in comparison to my superexponentially growing bliss!

Of course, as a real-life negative utilitarian, I'd politely decline the genie's offer.
But if you win me over to classical utilitarianism, I'll accept.
Which is the callous choice?
_____

davidpearce wrote:Thus a negative utilitarian can support creating a posthuman civilisation animated by gradients of intelligent bliss where all your dreams come true.

The problem is that such a world could be much bigger and still contain some pleasure-pain tradeoff. There always is some probability that suffering occurs, either as an error or because it is locally reintroduced in a simulation, or whatever. If one actually had to choose between destroying the world painfully now, and creating a posthuman civilization colonizing the Virgo Supercluster, destroying the world would cause the least suffering. No matter how good one thinks one's bioengineering and civilization-planning skills are, there's almost certainly some error margin, and that means the raw quantity of sentience in the latter case implies more suffering.

Since we're doing thought-experiments, imagine if a magic genie offers me superexponential growth in my bliss at the price of exponential growth in your agony and despair. If I'm a classical utilitarian, then I am ethically bound to accept the genie's offer.

Note that this uses the same personal identity references that should be excluded if one really believes personal identity to be ethically irrelevant (not that anyone really makes all practical choices as if they did). Also note that it is hard for humans to intuit exponential or superexponential growth in intensity of experiences.

If we reframe the thought experiment, it could look like this: I can make the choice to create a loop of some finite n iterations, which does the following:

It creates 2^n seconds of pain for me, then it creates 2^n^n seconds of pleasure of the same intensity for me.

Do I want to take this offer? Yes, with n as high as possible. Do I want you to take this offer away from me against my will? No, it would indeed be callous.

Some of the posts in this thread are really getting to some specific conceptions of measurement and the big picture of utilitarianism. Given that we are getting very detailed, I'd like to scrutinize this statement:

Pablo Stafforini wrote:(1) Positive and negative experiences are "polar opposites" of each other, in the sense that every positive experience has an intensity that is matched by the intensity of a given negative experience. This is not an arbitrary artifact of the way we choose to measure the respective intensities of positive and negative experiences, but is rooted in the nature of the experiences themselves. Just as a certain amount of positive charge is needed to cancel the negative charge of a subatomic particle, so similarly there is a certain intensity of positive experience that will "cancel" the intensity of a negative experience.

Brian denies this claim, and argues that semi-negative utilitarianism should not therefore be defined as the view that...

I would disagree. What is consistent and objectively defined is the number line and the relationship between numbers on that line (or positive versus negative numbers of the same absolute value). The arbitrary part is the assignment of particular events or experiences to particular values. There is no objective theory that says that the spectrum must be balanced such that there is an equal number of items near -10 as there is near +10. Perhaps the set of things near +10 is empty but the list of things near -10 is populated. If you want to define your system such that "every positive experience has an intensity that is matched by the intensity of a given negative experience" then that is your choice--not an observation of a natural law of the universe. I also doubt that there is even a rough estimate for some things: if one person is tortured and has his arm cut off, is that balanced by growing a third arm on some unsuspecting person? If not, what is the exact opposite of cutting off a person's arm during torture?

My other comment pertains to how perceptions of one's own experiences manifest. The process does not inherently pass through an assignment engine that maps experiences to a perfect number line. There are collections of neurons firing--perhaps some of which lean toward pleasure and some that lean the other way, then (perhaps) a decision process in which a person makes a choice and perhaps sticks to the decision to avoid being called a flip-flopper (i.e., possible irrational processes or external influences). If Person A gives more attention to negative experiences than Person B does, I cannot say that one is objectively wrong and the other is right because of how things map onto a number line. When I consider things that affect me, I am closer to the negative utilitarian view than traditional utilitarian view (in part for different reasons not stated here).