How do most utilitarians feel about "replacement" thought experiments?

post by richard_ngo · 2019-09-06T11:14:20.764Z · score: 13 (13 votes) · EA · GW · 38 comments

This is a question post.

Contents

  Answers
    22 Carl_Shulman
    5 Halstead
    3 Ramiro
    3 RomeoStevens
    2 MichaelStJules
    2 MichaelStJules
    2 Simon_Knutsson
None
4 comments

In this paper, Simon Knutsson discusses an objection to standard utilitarianism: that it endorses killing many (or all) existing people if that leads to their replacement by beings with higher utility. For example, he lays out the following thought experiment:

Suboptimal Earth: Someone can kill all humans or all sentient beings on Earth and replace us with new sentient beings such as genetically modified biological beings, brains in vats, or sentient machines. The new beings could come into existence on Earth or elsewhere. The future sum of well-being would thereby become (possibly only slightly) greater. Traditional utilitarianism implies that it would be right to kill and replace everyone.

People who identify as utilitarians, do you bite the bullet on such cases? And what is the distribution of opinions amongst academic philosophers who subscribe to utilitarianism?

Answers

answer by Carl_Shulman · 2019-09-06T18:04:59.214Z · score: 22 (12 votes) · EA · GW

I don't identify as a utilitarian, but I am more sympathetic to consequentialism than the vast majority of people, and reject such thought experiments (even in the unlikely event they weren't practically self-defeating: utilitarians should want to modify their ideology and self-bind so that they won't do things that screw over the rest of society/other moral views, so that they can reap the larger rewards of positive-sum trades rather than negative-sum conflict). The contractarian (and commonsense and pluralism, but the theory I would most invoke for theoretical understanding is contractarian) objection to such things greatly outweighs the utilitarian case.

For the tiny population of Earth today (which is astronomically small compared to potential future populations) the idea becomes even more absurd. I would agree with Bostrom in Superintelligence (page 219) that failing to leave one galaxy, let alone one solar system for existing beings out of billions of galaxies would be ludicrously monomaniacal and overconfident (and ex ante something that 100% convinced consequentialists would have very much wanted to commit to abstain from).

comment by Simon_Knutsson · 2019-09-08T11:40:31.380Z · score: 12 (6 votes) · EA · GW

If we are concerned with how vulnerable moral theories such as traditional total act-utilitarianism and various other forms of consequentialism are to replacement arguments, I think much more needs to be said. Here are some examples.

1. Suppose the agent is very powerful, say, the leader of a totalitarian society on Earth that can dominate the other people on Earth. This person has access to technology that could kill and replace either everyone on Earth or perhaps everyone except a cluster of the leader’s close, like-minded allies. Roughly, this person (or the group of like-minded people the leader belongs to) is so powerful that the wishes of others on Earth who disagree can essentially be ignored from a tactical perspective. Would it be optimal for this agent to kill and replace either everyone or, for example, at least everyone in other societies who might otherwise get in the way of the maximization of the sum of well-being ?

2. You talk about modifying one’s ideology, self-bind and commit, but there are questions about whether humans can do that. For example, if some agent in the future would be about to be able to kill and replace everyone, can you guarantee that this agent will be able to change ideology, self-bind and commit to not killing? It would not be sufficient that some or most humans could change ideology, self-bind and commit.

3. Would it be optimal for every agent in every relevant future situation to change ideology, self-bind or commit to not kill and replace everyone or billions of individuals? Again, we can consider a powerful, ruthless dictator or totalitarian leader. Assume this person has so far neither modified nor commited to non-violence. This agent is then in a situation in which the agent could kill and replace everyone. Would it at that time be optimal for the leader to change ideology, self-bind or commit to not killing and replacing everyone?

comment by Carl_Shulman · 2019-09-08T18:55:23.724Z · score: 7 (4 votes) · EA · GW

The first words of my comment were "I don't identify as a utilitarian" (among other reasons because I reject the idea of things like feeding all existing beings to utility monsters for a trivial proportional gains to the latter, even absent all the pragmatic reasons not to; even if I thought such things more plausible it would require extreme certainty or non-pluralism to get such fanatical behavior).

I don't think a 100% utilitarian dictator with local charge of a society on Earth removes pragmatic considerations, e.g. what if they are actually a computer simulation designed to provide data about and respond to other civilizations, or the principle of their action provides evidence [EA · GW] about what other locally dominant dictators on other planets will do including for other ideologies, or if they contact alien life?

But you could elaborate on the scenario to stipulate such things not existing in the hypothetical, and get a situation where your character would commit atrocities, and measures to prevent the situation hadn't been taken when the risk was foreseeable.

That's reason for everyone else to prevent and deter such a person or ideology from gaining the power to commit such atrocities while we can, such as in our current situation. That would go even more strongly for negative utilitarianism, since it doesn't treat any life or part of life as being intrinsically good, regardless of the being in question valuing it, and is therefore even more misaligned with the rest of the world (in valuation of the lives of everyone else, and in the lives of their descendants). And such responses give reason even for utilitarian extremists to take actions that reduce such conflicts.

Insofar as purely psychological self-binding is hard, there are still externally available actions, such as visibly refraining from pursuit of unaccountable power to harm others, and taking actions to make it more difficult to do so, such as transferring power to those with less radical ideologies, or ensuring transparency and accountability to them.

comment by Simon_Knutsson · 2019-09-09T08:26:51.650Z · score: 3 (3 votes) · EA · GW

Carl, you write that you are “more sympathetic to consequentialism than the vast majority of people.” The original post by Richard is about utilitarianism and replacement thought experiments but I guess he is also interested in other forms of consequentialism since the kind of objection he talks about can be made against other forms of consequentialism too.

The following you write seems relevant to both utilitarianism and other forms of consequentialism:

I don't think a 100% utilitarian dictator with local charge of a society on Earth removes pragmatic considerations, e.g. what if they are actually a computer simulation designed to provide data about and respond to other civilizations, or the principle of their action provides evidence [EA · GW] about what other locally dominant dictators on other planets will do including for other ideologies, or if they contact alien life?

Even if these other pragmatic considerations you mention would not be removed by having control of Earth, the question remains whether they (together with other considerations) are sufficient to make it suboptimal to kill and replace everyone. What if the likelihood that they are in a simulation is not high enough? What if new scientific discoveries about the universe or multiverse indicate that taking into account agents far away from Earth is not so important?

You say,

But you could elaborate on the scenario to stipulate such things not existing in the hypothetical, and get a situation where your character would commit atrocities, and measures to prevent the situation hadn't been taken when the risk was foreseeable.

I don’t mean that the only way to object to the form of consequentialism under consideration is to stipulate away such things and assume they do not exist. One can also object that what perhaps make it suboptimal to kill and replace everyone are complicated and speculative considerations about living in a simulation or what beings on other planets will do. Maybe your reasoning about such things is flawed somewhere or maybe new scientific discoveries will speak against such considerations. In which case (as I understand you) it may become optimal for the leader we are talking about to kill and replace everyone.

You bring up negative utilitarianism. As I write in my paper, I don’t think negative utilitarianism is worse off than traditional utilitarianism when it comes to these scenarios that involve killing everyone. The same goes for negative vs. traditional consequentialism or the comparison negative vs. traditional consequentialist-leaning morality. I would be happy to discuss that more, but I guess it would be too off-topic given the original post. Perhaps a new separate thread would be appropriate for that.

You write,

That's reason for everyone else to prevent and deter such a person or ideology from gaining the power to commit such atrocities while we can, such as in our current situation.

In that case the ideology (I would say morality) is not restricted to forms of utilitarianism but also include many forms of consequentialism and views that are consequentialist-leaning. It may also include views that are non-consequentialist but open to that killing is sometimes right if it is done to accomplish a greater goal, and that, for example, place huge importance on the far future so that far future concerns make what happens to the few billion humans on Earth a minor consideration. My point is that I think it’s a mistake to merely talk about utilitarianism or consequentialism here. The range of views about which one can reasonable ask ‘would it be right to kill everyone in this situation, according to this theory?’ is much wider.

comment by Wei_Dai · 2019-09-08T18:16:27.403Z · score: 3 (3 votes) · EA · GW

Another variant along these lines:

Suppose almost all humans adopt utilitarianism as their moral philosophy and fully colonize the universe, and then someone invents the technology to kill humans and replace them with beings of greater well-being. Utilitarianism seems to imply that humans who are utilitarians should commit mass suicide in order to bring the new beings into existence. EDIT: Utilitarianism endorses humans voluntarily replacing themselves with these new beings. (New wording suggested by Richard Ngo [EA · GW].)

comment by MichaelStJules · 2019-09-08T19:33:11.194Z · score: 4 (4 votes) · EA · GW

Well, that is their own belief and they aren't (necessarily) harming others, so I don't think that's nearly as repugnant.

I think if you propose a hypothetical where everyone agrees with what should be done and everyone accepts the consequences, then disagreeing with the outcome seems worse to me than accepting it. In effect, you want your views to trump everyone else's.

comment by Simon_Knutsson · 2019-09-09T09:22:53.364Z · score: 2 (2 votes) · EA · GW

Yea, one can formulate many variants. I can't recall seening yours before. The following is one thing that might seem like nitpicking, but which I think it is quite important: In academia, it seems standard to formulate utilitarianism and other consequentialist theories so that they apply to everyone. For example,

Traditional total act-utilitarianism: An act is right if and only if it results in a sum of well-being (positive well-being minus negative well-being) that is at least as great as that resulting from any other available act.

These theories are not formulated as 'traditional utilitarians ought to ...'. I can't recall ever seeing a version of utilitarianism or consequentialism formluated as 'utilitarians/consequentialists ought to ...'.

So when you write "Utilitarianism seems to imply that humans who are utilitarians should" I would rephrase as 'Utilitarianism seems to imply that humans should' since utilitarianism applies to all agents not only utilitarians. But perhaps you mean 'utilitarianism seems to imply that humans, including those who are utilitarians, should...' which would make sense.

Why does my nitpicking matter? One reason is when thinking about scenarios or thought experiments. For example, I don't think one can reply to world destruction or replacement arguments by saying 'a consequentialist ought to not kill everyone because ...'. We can picture a dictator who has never heard of consequentialism, and who is just about to act out of hatred. And we can ask, 'According to the traditional total act-utilitarian criterion of rightness (i.e. an act is right if and only if ...), would the dictator taking action X (say, killing everyone) be right?

Another reason the nitpicking matters is when thinking about the plausibility of the theories. A theory might sound nicer and more appealing if it merely says 'Those who endorse this theory act in way X' rather than as they are usually roughly formulated 'Everyone act in way X, regardless of whether you endorse this theory or not'.

comment by Wei_Dai · 2019-09-09T10:20:40.168Z · score: 3 (3 votes) · EA · GW

You seem to be assuming moral realism, so that "Utilitarianism seems to imply ..." gets interpreted as "If utilitarianism is the true moral theory for everyone, then everyone should ..." whereas I'm uncertain about that. In particular if moral relativism, subjectivism, or anti-realism is true, then "Utilitarianism seems to imply ..." has to be interpreted as "If utilitarianism is the right moral theory for someone or represents their moral preferences, then that person should ..." So I think given my meta-ethical uncertainty, the way I phrased my statement actually does make sense. (Maybe it's skewed a bit towards anti-realism by implicature, but is at least correct in a literal sense even if realism is true.)

comment by richard_ngo · 2019-09-09T12:55:09.324Z · score: 9 (5 votes) · EA · GW

I think that "utilitarianism seems to imply that humans who are utilitarians should..." is a type error regardless of whether you're a realist or an anti-realist, in the same way as "the ZFC axioms imply that humans who accept those axioms should believe 1+1=2". That's not what the ZFC axioms imply - actually, they just imply that 1+1 = 2, and it's our meta-theory of mathematics which determines how you respond to this fact. Similarly, utilitarianism is a theory which, given some actions (or maybe states of the world, or maybe policies) returns a metric for how "right" or "good" they are. And then how we relate to that theory depends on our meta-ethics.

Given how confusing talking about morality is, I think it's important to be able to separate the object-level moral theories from meta-ethical theories in this way. (For more along these lines, see my post here).

comment by Wei_Dai · 2019-09-09T15:07:21.367Z · score: 2 (2 votes) · EA · GW

Given how confusing talking about morality is, I think it’s important to be able to separate the object-level moral theories from meta-ethical theories in this way. (For more along these lines, see my post here).

I'd have to think more about this, but my initial reaction is that this makes sense. How would you suggest changing my sentence in light of this? The following is the best I can come up with. Do you see anything still wrong with it, or any further improvements that can be made?

"It seems that (at least) the humans who are utilitarians should commit mass suicide in order to bring the new beings into existence, because that's what utilitarianism implies is the right action in that situation."

comment by richard_ngo · 2019-09-11T11:44:19.167Z · score: 1 (1 votes) · EA · GW

My first objection is that you're using a different form of "should" than what is standard. My preferred interpretation of "X should do Y" is that it's equivalent to "I endorse some moral theory T and T endorses X doing Y". (Or "according to utilitarianism, X should do Y" is more simply equivalent to "utilitarianism endorses X doing Y"). In this case, "should" feels like it's saying something morally normative.

Whereas you seem to be using "should" as in "a person who has a preference X should act on X". In this case, should feels like it's saying something epistemically normative. You may think these are the same thing, but I don't, and either way it's confusing to build that assumption into our language. I'd prefer to replace this latter meaning of "should" with "it is rational to". So then we get:

"it is rational for humans who are utilitarians to commit mass suicide in order to bring the new beings into existence, because that's what utilitarianism implies is the right action."

My second objection is that this is only the case if "being a utilitarian" is equivalent to "having only one preference, which is to follow utilitarianism". In practice people have both moral preferences and also personal preferences. I'd still count someone as being a utilitarian if they follow their personal preferences instead of their moral preferences some (or even most) of the time. So then it's not clear whether it's rational for a human who is a utilitarian to commit suicide in this case; it depends on the contents of their personal preferences.

I think we avoid all of this mess just by saying "Utilitarianism endorses replacing existing humans with these new beings." This is, as I mentioned earlier, a similar claim to "ZFC implies that 1 + 1 = 2", and it allows people to have fruitful discussions without agreeing on whether they should endorse utilitarianism. I'd also be happy with Simon's version above: "Utilitarianism seems to imply that humans should...", although I think it's slightly less precise than mine, because it introduces an unnecessary "should" that some people might take to be a meta-level claim rather than merely a claim about the content of the theory of utilitarianism (this is a minor quibble though. Analogously: "ZFC implies that 1 + 1 = 2 is true").

Anyway, we have pretty different meta-ethical views, and I'm not sure how much we're going to converge, but I will say that from my perspective, your conflation of epistemic and moral normativity (as I described earlier) is a key component of why your position seems confusing to me.

comment by richard_ngo · 2019-09-11T11:50:33.241Z · score: 1 (1 votes) · EA · GW

I originally wrote a different response to Wei's comment, but it wasn't direct enough. I'm copying the first part here since it may be helpful in explaining what I mean by "moral preferences" vs "personal preferences":

Each person has a range of preferences, which it's often convenient to break down into "moral preferences" and "personal preferences". This isn't always a clear distinction, but the main differences:

1. Moral preferences are much more universalisable and less person-specific (e.g. "I prefer that people aren't killed" vs "I prefer that I'm not killed").

2. Moral preferences are associated with a meta-preference that everyone has the same moral preferences. This is why we feel so strongly that we need to find a shared moral "truth". Fortunately, most people are in agreement in our societies on the most basic moral questions.

3. Moral preferences are associated with a meta-preference that they are consistent, simple, and actionable. This is why we feel so strongly that we need to find coherent moral theories rather than just following our intuitions.

4. Moral preferences are usually phrased as "X is right/wrong" and "people should do right and not do wrong" rather than "I prefer X". This often misleads people into thinking that their moral preferences are just pointers to some aspect of reality, the "objective moral truth", which is what people "objectively should do".

When we reflect on our moral preferences and try to make them more consistent and actionable, we often end up condensing our initial moral preferences (aka moral intuitions) into moral theories like utilitarianism. Note that we could do this for other preferences as well (e.g. "my theory of food is that I prefer things which have more salt than sugar") but because I don't have strong meta-preferences about my food preferences, I don't bother doing so.

The relationship between moral preferences and personal preferences can be quite complicated. People act on both, but often have a meta-preference to pay more attention to their moral preferences than they currently do. I'd count someone as a utilitarian if they have moral preferences that favour utilitarianism, and these are a non-negligible component of their overall preferences.

comment by Wei_Dai · 2019-09-11T13:35:51.692Z · score: 1 (1 votes) · EA · GW

As I mentioned earlier, I am uncertain about meta-ethics, so I was trying to craft a sentence that would be true under a number of different meta-ethical theories. I wrote "should" instead of "it is rational to" because under moral realism that "should" could be interpreted as a "moral should" while under anti-realism it could be interpreted as an "epistemic should". (I also do think there may be something in common between moral and epistemic normativity but that's not my main motivation.) Your suggestion “Utilitarianism endorses replacing existing humans with these new beings.” would avoid this issue, but the main reason I wrote my original comment was to create a thought experiment where concerns about moral uncertainty and contractarianism clearly do not apply, and “Utilitarianism endorses replacing existing humans with these new beings.” doesn't really convey that since you could say that even in scenarios where moral uncertainty and contractarianism do apply.

comment by richard_ngo · 2019-09-13T09:52:03.451Z · score: 1 (1 votes) · EA · GW

Using those two different types of "should" makes your proposed sentence ("It seems that (at least) the humans who are utilitarians should commit mass suicide in order to bring the new beings into existence, because that's what utilitarianism implies is the right action in that situation.") unnecessarily confusing, for a couple of reasons.

1. Most moral anti-realists don't use "epistemic should" when talking about morality. Instead, I claim, they use my definition of moral should: "X should do Y means that I endorse/prefer some moral theory T and T endorses X doing Y". (We can test this by asking anti-realists who don't subscribe to negative utilitarianism whether a negative utilitarian should destroy the universe - I predict they will either say "no" or argue that the question is ambiguous.) And so introducing "epistemic should" makes moral talk more difficult.

2. Moral realists who are utilitarians and use "moral should" would agree with your proposed sentence, and moral anti-realists who aren't utilitarians and use "epistemic should" would also agree with your sentence, but for two totally different reasons. This makes follow-up discussions much more difficult.

How about "Utilitarianism endorses humans voluntarily replacing themselves with these new beings." That gets rid of (most of) the contractarianism. I don't think there's any clean, elegant phrasing which then rules out the moral uncertainty in a way that's satisfactory to both realists and anti-realists, unfortunately - because realists and anti-realists disagree on whether, if you prefer/endorse a theory, that makes it rational for you to act on that theory. (In other words, I don't know whether moral realists have terminology which distinguishes between people who act on false theories that they currently endorse, versus people who act on false theories they currently don't endorse).

comment by Simon_Knutsson · 2019-09-09T11:58:39.378Z · score: 7 (3 votes) · EA · GW

Very interesting :) I don’t mean to be assuming moral realism, and I don’t think of myself as a realist. Suppose I am an antirealist and I state some consequentialist criterion of rightness: ‘An act is right if and only if…’. When stating that, I do not mean or claim that it is true in a realist sense. I may be expressing my feelings, I may encourage others to act according to the criterion of rightness, or whatever. At least I would not merely be talking about how I prefer to act. I would mean or express roughly ‘everyone, your actions and mine are right if and only if …’. But regardless of whether I would be speaking about myself or everyone, we can still talk about what the criterion of rightness (the theory) implies in the sense that one can check which actions satisfy the criteria. So we can say: according to the theory formulated as ‘an act is right if and only if…’ this act X would be right (simply because it satisfies the criteria). A simpler example is if we understand the principle ‘lying is wrong’ from an antirealist perspective. Assuming we specify what counts as lying, we can still talk about whether an act is a case of lying and hence wrong, according to this principle. And then one can discuss whether the theory or principle is appealing, given which acts it classifies as right and wrong. If repugnant action X is classified as right or if something obviously admirable act is classified as wrong, we may want to reject the theory/criterion, regardless of realism or antirealism.

Maybe all I’m saying is obvious and compatible with what you are saying.

comment by Wei_Dai · 2019-09-09T15:10:48.108Z · score: 4 (3 votes) · EA · GW

I think there is at least one plausible meta-ethical position under which when I say "I think utilitarianism is right." I just mean something like "I think that after I reach reflective equilibrium my preferences will be well-described by utilitarianism." and it is not intended to mean that I think utilitarianism is right for anyone else or applies to anyone else or should apply to anyone else (except insofar as they are sufficiently similar to myself in the relevant ways and therefore are likely to reach a similar reflective equilibrium). Do you agree this is a plausible meta-ethical position? If yes, does my sentence (or the new version [EA · GW] that I gave in the parallel thread) make more sense in light of this? In either case, how would you suggest that I rephrase my sentence to make it better?

comment by Simon_Knutsson · 2019-09-11T06:19:30.067Z · score: 1 (1 votes) · EA · GW

Sure. I’ll use traditional total act-utilitarianism defined as follows as the example here so that it’s clear what we are talking about:

Traditional total act-utilitarianism: An act is right if and only if it results in a sum of well-being (positive well-being minus negative well-being) that is at least as great as that resulting from any other available act.

I gather the metaethical position you describe is something like one of the following three:

(1) When I say ‘I think utilitarianism is right’ I mean ‘I think that after I reach reflective equilibrium I will think that any act I perform is right if and only if it results in a sum of well-being (positive well-being minus negative well-being) that is at least as great as that resulting from any other available act.’

This (1) was about which of your actions will be right. Alternatively, the metaethical position could be as follows:

(2) When I say ‘I think utilitarianism is right’ I mean ‘I think that after I reach reflective equilibrium I will think that any act anyone performs is right if and only if it results in a sum of well-being (positive well-being minus negative well-being) that is at least as great as that resulting from any other available act.’

Or perhaps formulating it in terms of want or preference instead of rightness, like the following, better describes your metaethical position (using utilitarianism as just an example):

(3) When I say ‘I think utilitarianism is right’ I mean ‘I think that after I reach reflective equilibrium I will want or have a preference for that everyone act in a way that results in a sum of well-being (positive well-being minus negative well-being) that is at least as great as that resulting from any other available act.’

My impression is that in the academic literature, metaethical theories/positions are usually, always or almost always formulated as general claims about what, for example, statements such as ‘one ought to be honest’ means; the metaethical theories/position do not have the form ‘when I say “one ought to be honest” I mean …’ But, sure, talking, as you do, about what you mean when you say ‘I think utilitarianism is right’ sounds fine.

The new version of your thought experiment sounds fine, which I gather would go something like the following:

Suppose almost all humans adopt utilitarianism as their moral philosophy and fully colonize the universe, and then someone invents the technology to kill humans and replace humans with beings of greater well-being. (Assume it would be optimal, all things considered, to kill and replace humans.) Utilitarianism seems to imply that at least humans who are utilitarians should commit mass suicide (or accept being killed) in order to bring the new beings into existence, because that's what utilitarianism implies is the optimal and hence morally right action in that situation.

comment by MagnusVinding · 2019-09-08T14:11:35.967Z · score: 7 (8 votes) · EA · GW
The contractarian (and commonsense and pluralism, but the theory I would most invoke for theoretical understanding is contractarian) objection to such things greatly outweighs the utilitarian case.

It is worth noting that this is not, as it stands, a reply available to a pure traditional utilitarian.

failing to leave one galaxy, let alone one solar system for existing beings out of billions of galaxies would be ludicrously monomaniacal and overconfident

But a relevant question here is whether that also holds true given a purely utilitarian view, as opposed to, say, from a perspective that relies on various theories in some notional moral parliament.

It is, of course, perfectly fine to respond to the question "how do most utilitarians feel about X?" by saying "I'm not a utilitarian, but I am sympathetic to it, and here is how someone sympathetic to utilitarianism can reply by relying on other moral frameworks". But then it's worth being clear that the reply is not a defense of pure traditional utilitarianism — quite the contrary.

comment by Carl_Shulman · 2019-09-08T18:13:09.657Z · score: 10 (3 votes) · EA · GW
It is worth noting that this is not, as it stands, a reply available to a pure traditional utilitarian.

That's why the very first words of my comment were "I don't identify as a utilitarian."



comment by Wei_Dai · 2019-09-08T21:04:23.507Z · score: 16 (7 votes) · EA · GW

I think I, and maybe others, are still confused about the point of your top-level comment. Simon Knutsson's argument is against utilitarianism, and I think Richard Ngo wanted to see if there was a good counter-argument against it from a utilitarian perspective, or if a utilitarian just has to "bite the bullet". It seems like the motivation for both people were to try to figure out whether utilitarianism is the right moral philosophy / correct normative ethics.

Your reply doesn't seem to address their motivation, which is why I'm confused. (If utilitarianism is the right moral philosophy then it would give the right action guidance even if one was 100% sure of it and other considerations such as contractarianism didn't apply, so it seems beside the point to talk about contractarianism and overconfidence.) Is the point that utilitarianism probably isn't right, but some other form of consequentialism is? If so, what do you have in mind?

comment by MagnusVinding · 2019-09-10T14:58:30.906Z · score: 3 (2 votes) · EA · GW
That's why the very first words of my comment were "I don't identify as a utilitarian."

I appreciate that, and as I noted, I think this is fine. :-)

I just wanted to flag this because it took me some time to clarify whether you were replying based on 1) moral uncertainty/other frameworks, or 2) instrumental considerations relative to pure utilitarianism. I first assumed you were replying based on 2) (as Brian suggested), and I believe many others reading your answer might draw the same conclusion. But a closer reading made it clear to me you were primarily replying based on 1).

comment by Brian_Tomasik · 2019-09-08T15:25:23.898Z · score: 4 (4 votes) · EA · GW

I think the idea is that even a pure utilitarian should care about contractarian-style thinking for almost any practical scenario, even if there are some thought experiments where that's not the case.

comment by Carl_Shulman · 2019-09-08T19:05:28.039Z · score: 6 (4 votes) · EA · GW

I think that is basically true in practice, but I am also saying that even absent those pragmatic considerations constraining utilitarianism, I still would hold these other non-utilitarian normative views and reject things like not leaving some space for existing beings for a tiny proportional increase in resources for utility monsters.

comment by richard_ngo · 2019-09-09T13:46:29.079Z · score: 3 (3 votes) · EA · GW

Okay, thanks. So I guess the thing I'm curious about now is: what heuristics do you have for deciding when to prioritise contractarian intuitions over consequentialist intuitions, or vice versa? In extreme cases where one side feels very strongly about it (like this one) that's relatively easy, but any thoughts on how to extend those to more nuanced dilemmas?

answer by Halstead · 2019-09-09T22:13:31.967Z · score: 5 (4 votes) · EA · GW

Inevitably, utilitarians would bite the bullet here, since ex hypothesi, there is more utility in the world in which all beings are replaced with beings with higher utility.

I think the question is whether this implication renders utilitarianism implausible. I have several observations.

(1) The assumption here of the thought experiment is that the correct way to assess moral theories is to test them against intuitions about lots of particular cases. And utilitarianism has plenty of counterintuive implications about particular cases: eg the one in the main post, the repugnant conclusion, counting sadistic pleasure, and so on ad infinitum. The problem is that I don't think this is the correct way to assess moral theories.

Many of the moral intuitions people have are best explained by the fact that those intuitions would be useful to have in the ancestral environment, rather than that they apprehend moral reality. eg incest taboos are strong across all cultures, as are beliefs that wrongdoers simply deserve punishment regardless of the benefits of punishment. These would be evolutionarily useful, which makes it hard for us to shake these beliefs. I don't think the belief that subjective wellbeing is intrinsically good is debunkable in the same way, though discussing that is beyond the scope of this post.

Analogy: the current state of moral philosophy is similar to maths if mathematicians judged mathematical proofs and theories on the basis of how intuitive they are. Thus, people's intuition against the Monty Hall Problem was thought to be a good reason to try to build an alternative theory of probability. This form of maths wouldn't get very far. By the same token, moral philosophy doesn't get far in producing agreement because it uses a predictably bad moral epistemology that overwhelmingly focuses on intuitions about particular cases.

(2) Rough outline argument:

a. Subjective experience is all that matters about the world. (Imagine a world without subjective experience - why would it matter? Imagine a world in which people complete their plans but feel nothing - why would it be good?)

b. Personal identity doesn't matter. (See Parfit. or Imagine if you were vaporised in your sleep and then a perfect clone appeared a millisecond afterwards. Why would this be bad?)

From a and b, with some plausible additional premises, you eventually end up with utilitarianism. This means you have to bite the bullet mentioned in the text, and you also find the bullet plausible because you accept a and b and the other premises.

Related to (1), I think a response to utilitarianism that started in the right way would attack these basic premises a and b, along with the other premises. eg It would try and show that something aside from subjective experience matters fundamentally.


answer by Ramiro · 2019-09-06T19:16:52.144Z · score: 3 (3 votes) · EA · GW

It looks like a strawman to me. It conflates (A) a question about evaluation (is Suboptimal Earth axiologically better than current Earth?) with (B) a question about decision/action (would it be right to kill everyone for the sake of Suboptimal Earth), and it omits:

(A) a utilitarian doesn't classify scenarios categorically ("this is good, that is bad"), but through an ordering over possible worlds, such as: (1) current population + everyone alive in Suboptimal Earth is better than (2) Suboptimal Earth scenario minus current population is better than (3) current Earth...

(B) a utilitarian decides according to ex ante expected utility, so it'd have to ask "what's the odds that Suboptimal Earth will occur given my decision?"

Of course, there are huge problems for such reasoning - a more realistic Suboptimal Earth would get close to a Pascal Muggering: imagine that a Super AGI asked you to press this red button, freeing it to turn the whole galaxy into an eternal utopian hedonist simulation, for example.

As someone who has been "fighting" utilitarianism for a long time, I can say that the best objections against it have been produced by utilitarians themselves.

answer by RomeoStevens · 2019-09-06T17:38:33.015Z · score: 3 (3 votes) · EA · GW

Most utilitarian gotchas are either circular or talking about leaky abstractions. 'Assume higher utility from taking option X, but OH NO, you forgot about consideration Y! Science have gone too far!'

See also aether variables.

answer by MichaelStJules · 2019-09-08T21:35:58.115Z · score: 2 (2 votes) · EA · GW

I don't know how many academic philosophers are utilitarians specifically, but 23.6% of respondents to this survey accept or lean towards consequentialism, and I think forms of utilitarianism are the most commonly accepted theories of consequentialism. It would be a guess for me to say how many of those 23.6% or the utilitarians among them accept replacement.

answer by MichaelStJules · 2019-09-08T21:07:27.711Z · score: 2 (2 votes) · EA · GW

I would say that I'm most sympathetic to consequentialism and utilitarianism (if understood to allow aggregation in other ways besides summation). I don't think it's entirely implausible that the order in which harms or benefits occur can matter, and I think this could have consequences for replacement, but I haven't thought much about this, and I'm not sure such an intuition would be permitted in what's normally understood by "utilitarianism".

Maybe it would be helpful to look at intuitions that would justify replacement, rather than a specific theory. If you're a value-monistic consequentialist, treat the order of harms and benefits as irrelevant (the case for most utilitarians, I think), and you

1. accept that separate personal identities don't persist over time and accept empty individualism or open individualism (reject closed individualism),

2. take an experiential account of goods and bads (something can only be good or bad if there's a difference in subjective experiences), and

3. accept either 3.a. or 3.b., according to whether you accept empty or open individualism:

3. a. (under empty individualism) accept that it's better to bring a better off individual into existence than a worse off one (the nonidentity problem), or or,

3. b. (under open individualism) accept that it's better to have better experiences,

then it's better to replace a worse off being A with better off one B than to leave A, because the being A, if not replaced, wouldn't be the same A if left anyway. In the terms of empty individualism, there's A1 who will soon cease to exist regardless of our choice, and we're deciding between A2 and B.

A1 need not experience the harm of death (e.g. if they're killed in their sleep), and the fact that they might have wanted A2 to exist wouldn't matter (in their sleep), since that preference could never have been counted anyway since A1 never experiences the satisfaction or frustration of this preference.

For open individualism, rather than A and B, or A1, A2 and B, there's only one individual and we're just considering different experiences for that individual.

I don't think there's a very good basis for closed individualism (the persistence of separate identities over time), and it seems difficult to defend a nonexperiental account of wellbeing, especially if closed individualism is false, since I think we would have to also apply this to individuals who have long been dead, and their interests could, in principle, outweigh the interests of the living. I don't have a general proof for this last claim, and I haven't spent a great deal of time thinking about it, though, so it could be wrong.

Also, this is "all else equal", of course, which is not the case in practice; you can't expect attempting to replace people to go well.

Ways out for utilitarians

Even if you're a utilitarian, but reject 1 above, i.e. believe that separate personal identities do persist over time and take a timeless view of individual existence (an individual is still counted toward the aggregate even after they're dead), then you can avoid replacement by aggregating wellbeing over each individual's lifetime before aggregating across individuals in certain ways (e.g. average utilitarianism or critical-level utilitarianism, which of course have other problems), see "Normative population theory: A comment" by Charles Blackorby and David Donaldson.

Under closed individualism, you can also believe that killing is bad if it prevents individual lifetime utility from increasing, but also believe there's no good in adding people with good lives (or that this good is always dominated by increasing an individual's lifetime utility, all else equal), so that the killing which prevents individuals from increasing their lifetime utilities would not be compensated for by adding new people, since they add no value. However, if you accept the independence of irrelevant alternatives and that adding bad lives is bad (with the claim that adding good lives isn't good, this is the procreation asymmetry), then I think you're basically committed to the principle of antinatalism (but not necessarily the practice). Negative preference utilitarianism is an example of such a theory. "Person-affecting views and saturating counterpart relations" by Christopher Meacham describes a utilitarian theory which avoids antinatalism by rejecting the independence of irrelevant alternatives.

answer by Simon_Knutsson · 2019-09-07T14:10:50.546Z · score: 2 (10 votes) · EA · GW

Hi Richard. You ask, “People who identify as utilitarians, do you bite the bullet on such cases? And what is the distribution of opinions amongst academic philosophers who subscribe to utilitarianism?”

Those are good questions, and I hope utilitarians or similar consequentialists reply.

It may be difficult to find out what utilitarians and consequentialists really think of such cases. Such theories could be understood as sometimes prescribing ‘say whatever is optimal to say; that is, say whatever will bring about the best results.’ It might be optimal to pretend to not bite the bullet even though the person actually does.

Regarding the opinions among academic philosophers who subscribe to traditional utilitarianism. I don’t know of many such people who are alive, but a few are Torbjörn Tännsjö, Peter Singer, Yew-Kwang Ng (is my impression), and Katarzyna de Lazari-Radek (is also my impression). And Toby Ord has written, “I am very sympathetic towards Utilitarianism, carefully construed.” Tännsjö (2000) says, “Few people today seem to believe that utilitarianism is a plausible doctrine at all.” Perhaps others could list additional currently living academic philosophers who are traditional utilitarians, but otherwise it’s a very small population when talking about a distribution. Here is a list https://en.wikipedia.org/wiki/List_of_utilitarians#Living, but it includes people who are not academic philosophers, like Krauss, Layard, Lindström, Matheny and Reese, and it lists negative utilitarian David Pearce, and I doubt it is correct regarding the academic philosophers included in the list.

I can’t think of any traditional utilitarian who has discussed the replacement argument (i.e., the one that involves killing and replacing everyone). Tännsjö has bitten a bullet on another issue that involves killing. As I write here https://www.simonknutsson.com/the-world-destruction-argument/#Appendix_Reliability_of_intuitions, Tännsjö thinks that a doctor ought to kill one healthy patient to give her organs to five other patients who need them to survive (if there are no bad side effects). He argues that if this is counterintuitive, that intuition is unreliable partly because it is triggered by something that is not directly morally relevant. The intuition stems from an emotional reluctance to kill in an immediate manner using physical force, which is a heuristic device selected for us by evolution, and we should realize that it is morally irrelevant whether the killing is done using physical force (Tännsjö 2015b, 67–68, 205–6, 278–79). And as I also write in my paper, he has written, among other things, “Let us rejoice with all those who one day hopefully … will take our place in the universe.” I like his way of writing. It is illuminating, he feels straightforward, and he often writes as if he teaches (in a good way). But I could only speculate about what he thinks about the replacement argument against his form of utilitarianism.

comment by richard_ngo · 2019-09-07T14:37:48.467Z · score: 6 (5 votes) · EA · GW

Thanks for the informative reply! And also for writing the paper in the first place :)

"Such theories could be understood as sometimes prescribing ‘say whatever is optimal to say; that is, say whatever will bring about the best results.’ It might be optimal to pretend to not bite the bullet even though the person actually does."

I think we need to have high epistemic standards in this community, and would be dismayed if a significant number of people with strong moral views were hiding them in order to make a better impression on others. (See also https://forum.effectivealtruism.org/posts/CfcvPBY9hdsenMHCr/integrity-for-consequentialists [EA · GW])

comment by JulianAbernethy · 2019-09-08T11:25:35.674Z · score: 3 (3 votes) · EA · GW

I have to admit I only skimmed the paper, can you explain to me what "bullet" I ought to bite? It seems like a neutral proposition to say, what if we replace "us" with beings of greater well being.

My best guess is that our intuition for self-preservation and our intuition that killing sentient beings is bad most of the time should make me feel like that it would be a horrible idea? I'd rather throw the intuitions out than the framework, so to me this seems like an easy question. But maybe you have other arguments that make me not want to replace us.

comment by Simon_Knutsson · 2019-09-08T15:21:46.727Z · score: 8 (5 votes) · EA · GW

To bite the bullet here would be to accept that it would be morally right to kill and replace everyone with other beings who, collectively, have a (possibly only slightly) greater sum of well-being. If someone could do that.

The following are two similar scenarios:

Traditional Utilitarian Elimination: The sum of positive and negative well-being in the future will be negative if humans or sentient life continues to exist. Traditional utilitarianism implies that it would be right to kill all humans or all sentient beings on Earth painlessly.

Suboptimal Paradise: The world has become a paradise with no suffering. Someone can kill everyone in this paradise and replace them with beings with (possibly only slightly) more well-being in total. Traditional utilitarianism implies that it would be right to do so.

To bite the bullet regarding those two scenarios would be to accept that killing everyone would be morally right in those scenarios.

comment by Halstead · 2019-09-13T09:58:15.576Z · score: 2 (1 votes) · EA · GW

If you actually think that the only thing that matters is wellbeing, then personhood doesn't matter, so it makes sense that you would endorse these conclusions in this thought experiment.

38 comments

Comments sorted by top scores.

comment by MagnusVinding · 2019-09-07T09:37:20.632Z · score: 8 (8 votes) · EA · GW

Thanks for posting this, Richard. :-)

I think it is worth explaining what Knutsson's argument in fact is.

His argument is not that the replacement objection against traditional/classical utilitarianism (TU) is plausible. Rather, the argument is that the replacement objection against TU (as well as other consequentialist views it can be applied to, such as certain prioritarian views) is roughly as plausible as the world destruction argument is against negative utilitarianism (NU). And therefore, if one rejects NU and favors TU, or a similarly "replacement vulnerable" view, because of the world destruction argument, one must explain why the replacement argument is significantly less problematic for these other views.

That is, if one rejects such thought experiments in the case of TU and similar views because 1) endorsing or even entertaining such an idea would be sub-optimal in the bigger picture for cooperation reasons, 2) because it would be overconfident to act on it even if one finds the underlying theory to be the most plausible one, 3) because it leaves out "consideration Y", 4) because it seems like a strawman on closer examination, Knutsson's point is that one can make similar points in the case of NU and world destruction with roughly equal plausibility.

As Knutsson writes in the abstract:

>The world destruction argument is not a reason to reject negative utilitarianism in favour of these other forms of consequentialism, because there are similar arguments against such theories that are at least as persuasive as the world destruction argument is against negative utilitarianism.



comment by richard_ngo · 2019-09-08T18:26:53.929Z · score: 5 (5 votes) · EA · GW

I agree and didn't mean to imply that Knutsson endorses the argument in absolute terms; thanks for the clarification.

comment by Tobias_Baumann · 2019-09-07T20:31:33.923Z · score: 6 (5 votes) · EA · GW

I don't understand why this question has been downvoted by some people? It is a perfectly reasonable and interesting question. (The same holds for comments by Simon Knutsson and Magnus Vinding, which to me seem informative and helpful but have been downvoted.)

comment by Brian_Tomasik · 2019-09-08T15:43:46.603Z · score: 1 (1 votes) · EA · GW

I see here [EA · GW] it says: "Aside from mass voting, you can vote using any other criteria you choose." Presumably some people use votes to express dislike rather than to rate the quality of the comment.

(I expect most of us are guilty of this to some extent. I don't downvote comments merely because I disagree, but I upvote more often on comments with which I do agree...)