Less Wrong is a community blog devoted to refining the art of human rationality. Please visit our About page for more information.

Metacontrarian Metaethics

2 Post author: Will_Newsome 20 May 2011 05:36AM

Designed to gauge responses to some parts of the planned “Noticing confusion about meta-ethics” sequence, which should intertwine with or be absorbed by Lukeprog’s meta-ethics sequence at some point.

Disclaimer: I am going to leave out many relevant details. If you want, you can bring them up in the comments, but in general meta-ethics is still very confusing and thus we could list relevant details all day and still be confused. There are a lot of subtle themes and distinctions that have thus far been completely ignored by everyone, as far as I can tell.

Problem 1: Torture versus specks

Imagine you’re at a Less Wrong meetup when out of nowhere Eliezer Yudkowsky proposes his torture versus dust specks problem. Years of bullet-biting make this a trivial dilemma for any good philosopher, but suddenly you have a seizure during which you vividly recall all of those history lessons where you learned about the horrible things people do when they feel justified in being blatantly evil because of some abstract moral theory that is at best an approximation of sane morality and at worst an obviously anti-epistemic spiral of moral rationalization. Temporarily humbled, you decide to think about the problem a little longer:

"Considering I am deciding the fate of 3^^^3+1 people, I should perhaps not immediately assert my speculative and controversial meta-ethics. Instead, perhaps I should use the averaged meta-ethics of the 3^^^3+1 people I am deciding for, since it is probable that they have preferences that implicitly cover edge cases such as this, and disregarding the meta-ethical preferences of 3^^^3+1 people is certainly one of the most blatantly immoral things one can do. After all, even if they never learn anything about this decision taking place, people are allowed to have preferences about it. But... that the majority of people believe something doesn’t make it right, and that the majority of people prefer something doesn’t make it right either. If I expect that these 3^^^3+1 people are mostly wrong about morality and would not reflectively endorse their implicit preferences being used in this decision instead of my explicitly reasoned and reflected upon preferences, then I should just go with mine, even if I am knowingly arrogantly blatantly disregarding the current preferences of 3^^^3 currently-alive-and-and-not-just-hypothetical people in doing so and thus causing negative utility many, many, many times more severe than the 3^^^3 units of negative utility I was trying to avert. I may be willing to accept this sacrifice, but I should at least admit that what I am doing largely ignores their current preferences, and there is some chance it is wrong upon reflection regardless, for though I am wiser than those 3^^^3+1 people, I notice that I too am confused."

You hesitantly give your answer and continue to ponder the analogies to Eliezer’s document “CEV”, and this whole business about “extrapolation”...

(Thinking of people as having coherent non-contradictory preferences is very misleadingly wrong, not taking into account preferences at gradient levels of organization is probably wrong, not thinking of typical human preferences as implicitly preferring to update in various ways is maybe wrong (i.e. failing to see preferences as processes embedded in time is probably wrong), et cetera, but I have to start somewhere and this is already glossing over way too much.)

Bonus problem 1: Taking trolleys seriously

"...Wait, considering how unlikely this scenario is, if I ever actually did end up in it then that would probably mean I was in some perverse simulation set up by empirical meta-ethicists with powerful computers, in which case they might use my decision as part of a propaganda campaign meant to somehow discredit consequentialist reasoning or maybe deontological reasoning, or maybe they'd use it for some other reason entirely, but at any rate that sure complicates the problem...” (HT: Steve Rayhawk)

Comments (75)

Comment author: DanArmak 22 May 2011 12:35:59PM 1 point [-]

The original dust speck vs. torture problem isn't probabilistic. You're asked to choose whether to dust-speck 3^^^3 people, or to torture someone; not whether to do it to yourself.

If we reformulate it to apply to the person making the decision, we should be careful - the result may not be the same as for the original problem. For instance, it's not clear that making a single decision to dust-speck 3^^^3 people is the same as making 3^^^3 independent decisions to dust-speck each of them. (What do you do when the people disagree?)

Comment author: RichardChappell 22 May 2011 12:28:49AM 6 points [-]

It's confusing that you use the word 'meta-ethics' when talking about plain first-order ethics.

Comment author: Will_Newsome 22 May 2011 02:28:28AM 0 points [-]

...You're right, that was pretty sloppy. I felt vaguely justified in doing so since I often think about meta-ethics implied by or represented in first-order ethics (not that the levels are easily distinguishable in the first place in practice) and thus sort of made a point of not distinguishing them carefully. In hindsight that was dumb and especially dumb to to fail to acknowledge.

Comment author: Eugine_Nier 21 May 2011 03:36:19PM 3 points [-]

The problem with basing decisions on events with a probability of 1-in-3^^^^^3, is that you're neglecting to take into account all kinds of possibilities with much higher (though still tiny probabilities).

For example, your chances of finding that the Earth has turned into your favorite fantasy novel, i.e., the particles making up the earth spontaneously rearranged themselves into a world closely resembling the world of the novel due to quantum tunneling, and then the whole thing turning into a giant bowl of tapioca pudding a week later, is much much higher then 1-in-3^^^^^3.

Comment author: Amanojack 22 May 2011 02:04:24PM *  0 points [-]

The problem with basing decisions on events with a probability of 1-in-3^^^^^3, is that you're neglecting to take into account all kinds of possibilities with much higher (though still tiny probabilities).

Especially the probability that the means by which you learned of these probabilities is unreliable, which is probably not even very tiny. (How tiny is the probability that you, the reader of this comment, are actually dreaming right now?)

Comment author: jimrandomh 22 May 2011 05:02:14PM 0 points [-]

Actually, considering the possibility that you've misjudged the probability doesn't help with Pascal's Mugging scenarios, because

P(X|judged that X has probability p) >= p\*P(judgment was correct)

And while P(judgment was correct) may be small, it won't be astronomically small under ordinary circumstances, which is what it would take to resolve the mugging.

(My preferred resolution is to restrict the class of admissable utility function-predictor pairs to those where probability shrinks faster that utility grows for any parameterizable statement, which is slightly less restrictive than requiring bounded utility functions.)

Comment author: Eugine_Nier 22 May 2011 05:41:53PM 1 point [-]

BTW, you realize we're talking about torture vs. dust spec and not Pascal's mugging here?

Comment author: Amanojack 22 May 2011 05:49:20PM 0 points [-]

I think he's just pointing out that all you have to do is change the scenario slightly and then my objection doesn't work.

Still, I'm a little curious about how someone's ability to state a large number succinctly makes a difference. I mean, suppose the biggest number the mugger knew how to say was 12, and they didn't know about multiplication, exponents, up arrow notation, etc. They just chose 12 because it was the biggest number they could think of or knew how to express (whether they were bluffing totally or were actually going to torture 3^^^3 people). Should I take a mugger more seriously just because they know how to communicate big numbers to me?

Comment author: Eugine_Nier 22 May 2011 06:43:46PM 1 point [-]

The point of stating the large number succinctly is that it overwhelms the small likelihood of the muggers story being true, at least if you have something resembling a Solomonoff prior. Note also that the mugger isn't really necessary for the scenario, he's merely there to supply a hypothesis that you could have come up with on your own.

Comment author: Amanojack 22 May 2011 08:42:02PM 0 points [-]

Good point. I guess the only way to counter these odd scenarios is to point out that everyone's utility function is different, and then the question is simply whether the responder wants to self-modify (or would be happier in the long run doing so) even after hearing some rationalist arguments to clarify their intuitions. The question of self-modification is a little hard to grasp, but at least it avoids all these far-fetched situations.

Comment author: Eugine_Nier 22 May 2011 09:28:54PM 1 point [-]

For the Pascal's mugging problem, I don't think that will help.

Comment author: Amanojack 22 May 2011 09:48:09PM -2 points [-]

Isn't Pascal's mugging just this?

"Give me five dollars, or I'll use my magic powers from outside the Matrix to run a Turing machine that simulates and kills 3^^^^3 people."

I'd just walk away. Why should I care? If I thought about it for so long that I had some lingering qualms, and I got mugged like that a lot, I'd self-modify just to enjoy the rest of the my life more.

As an aside, I don't think people really care that much about other people dying unless they have some way to connect to it. Someone probably was murdered while you were reading this comment. Is it going to keep you up? On the other hand, people can cry all night about a video game character dying. It's all subjective.

Comment author: endoself 22 May 2011 10:11:30PM 1 point [-]

As an aside, I don't think people really care that much about other people dying unless they have some way to connect to it. Someone probably was murdered while you were reading this comment. Is it going to keep you up? On the other hand, people can cry all night about a video game character dying. It's all subjective.

There's a difference between mental distress and action-motivating desire. If I were asked to pay $5 to prevent someone from being murdered with near-certainty, I would. On the other hand, I would not pay $5 more for a video game where a character does not die, though I can't be sure of this self-simulation because I play video games rather infrequently. If I only had $5, I would definitely spend it on the former option.

I do not allow my mental distress to respond to the same things that motivate my actions; intuitively grasping the magnitude of existential risks is impossible and even thinking about a fraction of that tragedy could prevent action, such as by causing depression. However, existential risks still motivate my decisions.

Comment author: Will_Newsome 23 May 2011 09:51:31AM 0 points [-]

(My preferred resolution is to restrict the class of admissable utility function-predictor pairs to those where probability shrinks faster that utility grows for any parameterizable statement, which is slightly less restrictive than requiring bounded utility functions.)

It's still way too restrictive though, no? And are there ways you can Dutch book it with deals where probability grows faster (instead of the intuitively-very-common scenario where they always grow at the same rate)?

Comment author: Prismattic 20 May 2011 11:45:15PM *  8 points [-]

I think torture v. dust specks and similar problems can be illuminated by flipping them around and examining them from the perspective of the potential victims. Given a choice between getting a dust speck in the eye with probability 1 or a 1-in-3^^^^^3 chance of being tortured, I suspect the vast majority of individuals will actually opt for the dust speck, and I don't think this is just insensitivity to the scope of 3^^^^^3. Dust specks are such a trivial inconvenience that people generally don't choose to do any of the easy things they could do to minimize the chances of getting one (e.g. regularly dusting their environment, wearing goggles, etc.) On the other hand, most people would do anything to stop being tortured, up to and including suicide if the torture has no apparent end point. The difference here is arguably not expressible as a finite number.

Pardon me, I have to go flush my cornea.

Comment author: Kevin 25 May 2011 12:43:36PM 1 point [-]

1 in 3^^^^^3 is a really, really, really, really, really, really accurate specification of 0.

Comment author: NihilCredo 21 May 2011 10:21:27PM *  6 points [-]

FWIW, I would take the 1 in 3^^^^^3 chance of torture over a single dust speck*, but I would give 3^^^^^3 people a dust speck each than subject one person to torture, because the world with the dust specks looks nicer to me than the world with the torture. (I find fairness to be aesthetically pleasant.)

* Though since I feel like being pedantic, the - irrational - momentary anxiety of rolling the torture die would be a worse feeling than a dust speck, which in practice would make the dust speck the better choice.

Comment author: Kevin 26 May 2011 08:03:29AM *  0 points [-]

That's not a world with dust specks, that's millions upon millions of universes cycling endlessly of dust specks.

Comment author: FAWS 21 May 2011 08:46:34PM *  3 points [-]

and I don't think this is just insensitivity to the scope of 3^^^^^3.

Nope, unless they have an arbitrary discontinuity in their valuation of harm it's really just scope insensitivy, 3^^^^^3 is that big. Quibbling about a few orders of magnitude of their caring about dust specks ("dust specks" just means the smallest harm they care about at all) and torture is a waste of time, they could consider torture a billion times worse than they do and it wouldn't change anything.

It's another matter to volunteer to be dust-specked to save someone from torture, the personal satisfaction from having done so might very well outweigh the inconvenience. But if you choose for them and they never learn about it they don't get that satisfaction.

Comment author: Eugine_Nier 21 May 2011 02:39:16PM 1 point [-]

And yet we don't give Pascal's mugger the 10 bucks.

Comment author: Rain 21 May 2011 03:16:37AM *  4 points [-]

We have eyelashes, blink a lot, have reflex actions to shield our eyes, and will quickly stop all activity around us because "I got something in my eye."

And 1-in-3^^^^^3 odds are so trivial as to never happen within the lifetime of this universe or many such universes, so I think it is indeed scope insensitivity.

Comment author: Kevin 21 May 2011 04:52:23AM 3 points [-]

Yup. To get on the right scope, I would look at it as a choice between getting a dust speck in my eye or not getting a dust speck in my eye.

Comment author: Prismattic 21 May 2011 03:41:58PM 0 points [-]

I am just rephrasing the original torture v. dust specks problem. We have already established that exactly 1 individual in some universe somewhere will being tortured. The odds of that individual being you are greater-than-astronomical but that is not the same as "will never happen."

Comment author: Rain 22 May 2011 12:55:38PM *  3 points [-]

The odds of that individual being you are greater-than-astronomical but that is not the same as "will never happen."

Why would people only accept "zero probability", considering zero is not a probability?

Comment author: FAWS 22 May 2011 02:01:28PM 1 point [-]

The odds of that individual being you are greater-than-astronomical but that is not the same as "will never happen."

Yes, it's vastly less probable than things that merely "will never happen." It's far more likely that you spontaneously turn into chocolate raspberry ice cream and are eaten by a pink Yeti who happens to be Miss New South Wales.

Would you like to be dust specked in exchange for a reduction in the base rate of that specific thing happening to you in the next second by one part in a trillion? It's a much better deal than the torture offer.

Comment author: Prismattic 23 May 2011 04:26:25AM *  0 points [-]

Trying again here, as I'm not sure I expressed myself well in the previous attempt. The original problem is that you either torture one person or give 3^^^^^3 people dust specks. For purposes of this thought experiment, therefore, we must assume that there exist at least 3^^^^^3 people in some number of universes. So if you asked each and every one of them the formulation I offered in the grandparent, and they all chose the chance of torture based on your reasoning, exactly one of those people would be in for a very unpleasant surprise.

Also, the probability of the torture is greater than the probability of spontaneously converting into a frozen treat for an improbably attractive sasquatch, since we have no reason whatsoever to expect that there is anything that could cause the latter, but we have already taken as a given that there is some individual in the multiverse with the capacity to choose to inflict torture or dust specks.

[Edited] to correct the numbers to be of identical powers.

Comment author: wedrifid 23 May 2011 10:34:40AM *  2 points [-]

The original problem is that you either torture one person or give 3^^^^^3 people dust specks. For purposes of this thought experiment, therefore, we must assume that there exist at least 3^^^^^^3 people in some number of universes.

3^^^^^^3? Do you know the difference between 3^^^^^^3 and 3^^^^^3? Compare the size of an electron and the size of the entire universe. Then forget that because that doesn't come close to demonstrating it. This means that the fact that you said the original problem is about 3^^^^^3 when it is actually about a mere 3^^^3 is a relatively minor error.

That you are so willing to jump from 3^^^3 to 3^^^^^^3 in the same conversation suggest you don't really grasp the point here. The amount of people experiencing discomfort really does make a difference. There is an unimaginable amount of pain at stake.

Comment author: Prismattic 23 May 2011 10:39:50PM *  0 points [-]

The sixth carrot in 3^^^^^^3 was a typo, which I am correcting.

Comment author: Peterdjones 23 May 2011 10:43:03PM 0 points [-]

And "carrot"?

Comment author: Prismattic 23 May 2011 10:44:57PM 0 points [-]

The name of the grammatical symbol "^" is a carrot, as far as I know.

Comment author: Peterdjones 23 May 2011 10:54:39PM 0 points [-]
Comment author: FAWS 23 May 2011 09:36:30AM 1 point [-]

You do not seem to understand how large 3^^^^^3 is.

Comment author: ArisKatsaris 21 May 2011 05:25:21AM 0 points [-]

Given a choice between getting a dust speck in the eye with probability 1 or a 1-in-3^^^^^3 chance of being tortured, I suspect the vast majority of individuals will actually opt for the dust speck, and I don't think this is just insensitivity to the scope of 3^^^^^3.

I think it is. Imagine that the same trade is offered you a trillion times. Or imagine that it's automatically offered or rejected (unconsciously by default, but you have the ability to change the default) every second of your life.

After spending a week getting a dust speck in the eye every single second, I think you'll do the math and opt to be choosing the 1-in-3^^^^^3 chance of torture instead.

Comment author: wedrifid 21 May 2011 02:39:49PM *  8 points [-]

After spending a week getting a dust speck in the eye every single second, I think you'll do the math and opt to be choosing the 1-in-3^^^^^3 chance of torture instead.

That is an entirely different scenario than what Prismattic is describing. In fact, a dust speck in the eye every single second would be an extremely effective form of torture.

Comment author: NihilCredo 21 May 2011 04:23:00PM *  3 points [-]

Indeed. More abstractly: pleasure and suffering aren't so nice as to neatly add and multiply like pretty little scalars.

Even if you wish to talk about "utils/utilons" - it is by no means obvious that ten dust specks are worth exactly ten times as many (negative) utils as one dust speck.

Comment author: Prismattic 21 May 2011 03:38:43PM 1 point [-]

Getting a dust speck for a moment is a minor nuisance. Getting an uninterrupted series of dust specks forever is torture. It's not a particularly invasive form, but it is debilitating.

Comment author: Rain 22 May 2011 12:57:21PM *  2 points [-]

He was trying to show the difference between something which actually happens and has an effect versus something which will only happen with a 1-in-3^^^^^3 chance: one exists in this universe, the other does not.

If we changed it to be one dust speck per second and a 1-in-3^^^^^3 chance every second that you are quantum teleported to a random planet in the universe, you'd swallow that planet in the ever-expanding black hole you'd become long, long before you're teleported there.

Comment author: ArisKatsaris 22 May 2011 10:48:34AM 0 points [-]

Okay, what about a dust speck per hour or a dust speck per ten minutes. Still a minor nuisance, but has it reached to the point you'd prefer to have a 1-in-3^^^^^3 chance of being tortured?

Comment author: Nisan 20 May 2011 07:52:46PM 3 points [-]

What is a "meta-ethical preference"? Do you just mean a moral judgment that is informed by one's metaethics? Or do you mean something like a second-order moral judgment based on others' first-order moral judgments?

Comment author: jsalvatier 20 May 2011 08:15:37PM 2 points [-]

I didn't understand the phrase "preferences at gradient levels of organization". Can you clarify?

Comment author: Rain 20 May 2011 12:15:32PM *  8 points [-]

My favorite realist injection into the trolley problem is that there will be far more uncertainty: you won't know that the fat man will stop the trolley. I keep picturing someone tipping the poor guy over, watching him fall and break a few legs, moaning in agony, and then get mowed down by the trolley, which continues on its merry way and kills the children tied to the tracks regardless.

Comment author: NihilCredo 21 May 2011 10:13:40PM 4 points [-]

Have you come up with a better scenario for the trolley problem?

The one I currently like the best is:

  • Trolley: You're a surgeon, you have a dying patient in your care, he needs five full litres of healthy blood to survive the operation; fortunately, you have exactly five liters available. You've just opened him up when five more emergency patients arrive, each of whom could survive with just one litre.

  • Fat man: There is no dying surgery patient, but the same five new emergencies have just arrived and you have no blood reserves at all. What you do have is a healthy but unconscious patient, with five litres of good blood in his veins.

It's still not perfect though, because the role of doctors has deep cultural roots (Hyppocratic oath and so on), so the idea of a doctor doing harm to a patient feels repugnant and blasphemous, and since a patient feels like he is "entrusting" himself to a surgeon there's also an overtone of betrayal. Modern hospitals, huge and anonymous, have only partially deleted such feelings (and they try their best not to).

Comment author: Rain 22 May 2011 01:06:04AM *  1 point [-]

Having played a healer in many online games, I've discovered that triage (what you've described above) quickly becomes second nature, to the point where if someone performs the wrong response to a threat, I will literally say aloud, "go die then" because I have more important people to take care of. I consider it a signature of the best healers that they will abort a spell on someone who needs it to instead cast it on someone who is more important.

A true trolley problem would have to be contrived by a murderous, insane villain ala Saw; the uncertainty remains in any real world scenario that I've come across

Though perhaps organ transplants can serve as a stand-in. Tons of otherwise healthy people just need one organ to live out a good life, and tons of people who are somewhat negative utility have nice, juicy organs. We'll know we're on our way to fixing the trolley problem when organ donation is mandatory, and can be a punishment handed down similar to the death penalty for those who harm society.

Comment author: MixedNuts 20 May 2011 08:02:33PM 1 point [-]

That's what makes heroism so poignant in real life. If you have to shoot an innocent but firmly believe it'll save the world, you'll probably brood a little (especially if there's also a cost to you), but mostly you'll get massive fuzzies. (I've never shot an innocent, but defending a cause you think is just is... more pleasant than it should be.) If you have to shoot an innocent and you expect it won't save the world but on average it's worth it anyway, it gnaws at you.

Comment author: Prismattic 21 May 2011 01:37:34AM 3 points [-]

Heroism is throwing yourself on the tracks to save the greater number of people. Pushing somebody else may be an example of decisiveness, or courage (in the sense of grace under pressure) but there is nothing heroic about it.

Comment author: orthonormal 21 May 2011 08:16:27PM 3 points [-]

This is a definitional dispute and an attempt at applause lights rather than a helpful comment.

Comment author: wedrifid 21 May 2011 09:20:20PM *  4 points [-]

This is a definitional dispute and an attempt at applause lights rather than a helpful comment.

I don't agree with your judgement. The applause light reference in particular doesn't seem fair.

Comment author: MixedNuts 21 May 2011 06:07:56AM 1 point [-]

If it's really purely a cost to others, okay. But usually there's also a cost to yourself - you push someone else and get a death sentence, or a life sentence, you push a loved one, or you're just eaten alive by guilt for the rest of your days.

Comment author: Will_Newsome 20 May 2011 12:35:05PM 0 points [-]

Ha. I like Steve's version better though, since you can assume away all the environmental uncertainty leaving just indexical uncertainty (the distinction doesn't really exist but bear with me), and yet people still can't just go "LCPW!", because this version is both way less convenient and more probable. I bet it'd still annoy the hell out of a real philosopher, though.

Comment author: lessdazed 20 May 2011 09:27:33AM 4 points [-]

suddenly you have a seizure during which you vividly recall all of those history lessons where you learned about the horrible things people do when they feel justified in being blatantly evil because of some abstract moral theory

People often mistakenly think they are above average at tasks and skills such as driving. This has implications for people who are members of the set of people who believe themselves above average, without changing how well members of the set of people who are actually above average at driving can drive.

Humans often mistakenly think they face trolley problems when they really don't. This has implications for people who believe they face a trolley problem, without directly changing what constitutes a good response by someone who actually faces a trolley problem.

If I expect that these 3^^^3+1 people are mostly wrong about morality and would not reflectively endorse their implicit preferences being used in this decision instead of my explicitly reasoned and reflected upon preferences, then I should just go with mine

If your decision depends on referencing people's hypothetical reflectively endorsed morality, then you are not simply going with your preferences about morality, divorced from the moral systems of the many people in question. Your original thought process was about the morality of the act independent of those people's preferences,and it determined one choice was right. Having checked others' reflective morality, it's in an important sense a coincidence that you conclude that the same act is the right one. You are performing a new calculation (that it is largely comprised of the old one is irrelevant), and so should not say you are "just" going with "[your]" preferences.

That you are ignoring people's stated preferences in both calculations (which remember, have the same conclusion) is similarly irrelevant. In the second but not the first you weigh people's reflective morality, so despite other (conspicuous) similarities between the calculations, there was no going back to the original calculation in reaching the second conclusion.

I am knowingly arrogantly blatantly disregarding the current preferences of 3^^^3 currently-alive-and-and-not-just-hypothetical people in doing so and thus causing negative utility many, many, many times more severe than the 3^^^3 units of negative utility I was trying to avert.

If in your hypothetical they are informed and this hurts them, they're getting more than a speck's worth, eh?

I may be willing to accept this sacrifice

You're willing to accept the sacrifice of others having negative utility?

blatantly evil with respect to their current preferences

It's OK to admit an element of an action was bad - it's not really an admission of a flaw. We can celebrate the event of the death of Bin Laden without saying it's good that a helicopter crashed in the operation to get him. We can celebrate the event of his death without saying that one great element of his death was that he felt some pain. We can still love life and be happy for all Americans and especially the SEAL who shot him, that he got to experience what really has to feel FANTASTIC, without saying it's, all else equal, good OBL is dead rather than alive. All else is not equal, as we are overall much better off without him.

The right choice will have some negative consequences, but to say it is partly evil is misleadingly calling attention to an irrelevancy, if it isn't an outright misuse of "evil".

Bonus problem 1: Taking trolleys seriously

Or to test the subject to see if he or she is trustworthy...or for reasons I can't think of.

Comment author: Will_Newsome 20 May 2011 10:29:01AM *  2 points [-]

Humans often mistakenly think they face trolley problems when they really don't. This has implications for people who believe they face a trolley problem, without directly changing what constitutes a good response by someone who actually faces a trolley problem.

I'm having trouble inferring your point here... The contrast between 'those who are dreaming think they are awake, but those who are awake know they are awake' and "I refuse to extend this reply to myself, because the epistemic state you ask me to imagine, can only exist among other kinds of people than human beings" is always on the edges of every moral calculation, and especially every one that actually matters. (I guess it might sound like I'm suggesting reveling in doubt, but noticing confusion is always so that we can eventually become confused on a higher level and about more important things. Once you notice a confusion, you get to use curiosity!)

If your decision depends on referencing people's hypothetical reflectively endorsed morality, then you are not simply going with your preferences about morality, divorced from the moral systems of the many people in question.

Yeah, so this gets a little tricky because the decision forks depending on whether or not you think most people would themselves care about their future smarter selves' values, or whether you think they don't care but they're wrong for not caring. (The meta levels are really blending here, which is a theme I didn't want to avoid but unfortunately I don't think I came up with an elegant way to acknowledge their importance while keeping the spirit of the post, which is more about noticing confusion and pointing out lots of potential threads of inquiry than it is an analysis, since a real analysis would take a ton of analytic philosophy, I think.)

That you are ignoring people's stated preferences in both calculations (which remember, have the same conclusion) is similarly irrelevant. In the second but not the first you weigh people's reflective morality, so despite other (conspicuous) similarities between the calculations, there was no going back to the original calculation in reaching the second conclusion.

Ah, I was trying to hint at 3 main branches of calculation; perhaps I will add an extra sentence to delineate the second one more. The first is the original "go with whatever my moral intuitions say", the second is "go with whatever everyone's moral intuitions say, magically averaged", and the third is "go with what I think everyone would upon reflection think is right, taking into account their current intuitions as evidence but not as themselves the source of justifiedness". The third and the first are meant to look conspicuously like each other but I didn't mean to mislead folk into thinking the third explicitly used the first calculation. The conspicuous similarity stems from the fact that the actual process you would go through to reach the first and the third positions are probably the same.

The right choice will have some negative consequences, but to say it is partly evil is misleadingly calling attention to an irrelevancy, if it isn't an outright misuse of "evil"

I used some rhetoric, like using the word 'evil' and not rounding 3^^^3+1 to just 3^^^3, to highlight how the people whose fate you're choosing might perceive both the problem and how you're thinking about the problem. It's just... I have a similar reaction when thinking about a human self-righteously proclaiming 'Kill them all, God will know His own.', but I feel like it's useful that a part of me always kicks in and says I'm probably doing the same damn thing in ways that are just less obvious. But maybe it is not useful.

Comment author: lessdazed 20 May 2011 12:15:31PM 4 points [-]

The contrast between 'those who are dreaming think they awake, but those who are awake know they are awake

I have an uncommon relationship with the dream world, as I remember many dreams every night. I often dream within a dream, I might do this more often than most because dreams occupy a larger portion of my thoughts than they do in others, or I might just be remembering those dreams more than most do. When I wake up within a dream, I often think I am awake. On the other hand, sometimes in the middle of dreams I know I am dreaming. Usually it's not something I think about while asleep or awake.

I also have hypnopompic sleep paralysis, and sometimes wake up thinking I am dead. This is like the inverse of sleep walking - the mind wakes up some time before the body can move. I'm not exactly sure if one breathes during this period or not, but it's certainly impossible to consciously breathe and one immediately knows that one cannot, so if one does not think oneself already dead (which for me is rare) one thinks one will suffocate soon. Confabulating something physical blocking the mouth or constricting the trunk can occur. It's actually not as bad to think one is dead, because then one is (sometimes) pleasantly surprised by the presence of an afterlife (even if movement is at least temporarily impossible) and one does not panic about dying - at least I don't.

So all in all I'd say I have less respect for intuitions like that than most do.

Comment author: lessdazed 20 May 2011 11:39:24AM 1 point [-]

I'm having trouble inferring your point here...

One point is that I feel very unconfused. That is, not only do I not feel confused now, I once felt confused and experienced what I thought was confusion lifting and being replaced by understanding.

I feel like it's useful that a part of me always kicks in

Which one, if just one, criteria for usefulness are you using here? It is useful for the human to have pain receptors, but there is negative utility in being vulnerable to torture (and not just from one's personal perspective).

Surely you don't expect that even the most useful intuition is always right? This is similar to the Bin Laden point above, that the most justified and net-good action will almost certainly have negative consequences.

I'm willing to call your intuition useful if it often saves you from being misled, and its score on any particular case is not too important in its overall value.

However, its score on any particular case is indicative of how it would do in similar cases. If it has a short track record and it fails this test, we have excellent reason to believe it is a poorly tuned intuition because we know little other than how it did on this hypothetical, though its poor performance on this hypothetical should never be considered a significant factor in what makes it generally out of step with moral dilemmas regardless. This is analogous to getting cable ratings from only a few tracked boxes: we think many millions watched a show because many of the thousands tracked did, but do not think those thousand constitute a substantial portion of the audience.

Comment author: Will_Newsome 20 May 2011 12:18:12PM *  2 points [-]

Which one, if just one, criteria for usefulness are you using here? It is useful for the human to have pain receptors, but there is negative utility in being vulnerable to torture (and not just from one's personal perspective).

That's the one I'm referencing. My fear of having been terribly immoral (which could also be even less virtuously characterized as being or at least being motivated by an unreasonable fear of negative social feedback) is useful because it increases the extent to which I'm reflective on my decisions and practical moral positions, especially in situations that pattern match to ones that I've already implicitly labeled as 'situations where it would be easy to deceive myself into thinking I had a good justification when I didn't', or 'situations where it would be easy to throw up my hands because it's not like anyone could actually expect me to be perfect'. Vegetarianism is a concrete example. The alarm itself (though perhaps not the state of mind that summons it) has been practically useful in the past, even just from a hedonic perspective.

Comment author: lessdazed 20 May 2011 12:34:13PM 1 point [-]

OK, sometimes you will end up making the same decision after reflection and having wasted time, other times you may even change from a good decision (by all relevant criteria) to a bad one simply because your self-reflection was poorly executed. That doesn't necessarily mean there's something wrong with you for having a fear or with your fear (though it seems too strong in my opinion).

This should be obvious - it wasn't to me until after reading your comment the second time - but "increases the extent to which I'm reflective" really ought to sound extraordinarily uncompelling to us. Think about it: a bias increases the extent to which you do something. It should be obvious that that thing is not always good to increase, and the only reason it seems otherwise to us is that we automatically assume there are biases in the opposite direction that won't be exceeded however much we try to bias ourselves. Even so, to combat bias with bias - it's not ideal.

Comment author: Multiheaded 15 April 2012 07:38:04AM 0 points [-]

I used some rhetoric, like using the word 'evil' and not rounding 3^^^3+1 to just 3^^^3, to highlight how the people whose fate you're choosing might perceive both the problem and how you're thinking about the problem. It's just... I have a similar reaction when thinking about a human self-righteously proclaiming 'Kill them all, God will know His own.', but I feel like it's useful that a part of me always kicks in and says I'm probably doing the same damn thing in ways that are just less obvious. But maybe it is not useful.

Umm, didn't you (non-trollishly) advocate indiscriminately murdering anyone and everyone accused of heresy as long as it's the Catholic Church doing it?

Comment author: Kevin 20 May 2011 10:22:13AM 1 point [-]

Yes, but I don't like the word evil.

Comment author: Will_Newsome 20 May 2011 10:40:41AM *  2 points [-]

2 out of 2 comments thus far complaining about this means I have to change it... /sigh. I like the word evil because good/evil and good/bad (depending on context) is how most of the world thinks (at one level of abstraction at least), and in meta-ethics it's not entirely obvious you're allowed to just throw away such value judgments as the results of misguided human adaptations.

Comment author: Kevin 21 May 2011 06:17:42AM 0 points [-]

I feel like the word "evil" is only coherent under absolute morality (which I'm pretty sure is wrong) and falls apart trivially under relative morality. Just because it is regular used as a metaphor by the population doesn't seem worth the linguistic precision lost by a concept so widely misunderstood.

This post is slightly less interesting with the word evil removed because negative utility is harder to feel for most people, but it makes more sense to me.

I would be interested in hearing you more clearly define the nature of evil as negative utility and having a comment thread descend into various

Comment author: Peterdjones 21 May 2011 02:08:04PM *  1 point [-]

I feel like the word "evil" is only coherent under absolute morality (which I'm pretty sure is wrong) and falls apart trivially under relative morality. Just because it is regular used as a metaphor by the population doesn't seem worth the linguistic precision lost by a concept so widely misunderstood.

Do you mean absolute or objective morality? The two are not the same (see the WP articles).

I would be interested in hearing you more clearly define the nature of evil as negative utility

The commitments you have already made, if true, make that impossible. Utility is clearly relative. You say Evil falls apart under relative interpretations. Therefore, there can be no interpretation of Evil in terms of utility where it doesn't fall apart.

Comment author: army1987 24 December 2012 10:10:48AM 0 points [-]

But... that the majority of people believe something doesn’t make it right, and that the majority of people prefer something doesn’t make it right either.

Gotta love how this sentence is perfectly clear even though the word right means two different things (‘true’ and ‘good’ respectively) in two seemingly parallel places.

Comment author: wedrifid 20 May 2011 11:02:45AM 0 points [-]

"Considering I am deciding the fate of 3^^^3+1 people, I should perhaps not immediately assert my speculative and controversial meta-ethics. Instead, perhaps I should use the averaged meta-ethics of the 3^^^3+1 people I am deciding for

I am careful to avoid putting people in a position of such literal moral hazard. That is, where people I care about will end up having their current preferences better satisfied by having different preferences than their current preferences. I don't average.

Comment author: lessdazed 20 May 2011 11:50:59AM 1 point [-]

I'm confused. All people are always in position to have current preferences better satisfied by having somewhat different preferences, no?

I have no doubt the case you're thinking of meets the criteria for "That is...". One mistake I have made recently is to think my description fit reality because an important scenario fit according to thedescription, as I diligently checked...but so did every other outcome. Perhaps I am oversensitive right now to seeing this mistake around me, perhaps this is a false positive, or a true positive I wouldn't have otherwise spotted, or even a true positive I would have without having experience with that mistake - seeing the flaws in others' arguments is so very much easier than seeing them in one's own. This is particularly true of gaps, which one naturally fills in.

If you could share some examples of when you were put in the position of putting people in the position of moral hazard, that would be great.

Comment author: wedrifid 20 May 2011 12:51:09PM 3 points [-]

All people are always in position to have current preferences better satisfied by having somewhat different preferences, no?

Preferences A will be more satisfied if the agent actually had preferences B than they will be if they actually have preferences A. So the way you get what you would have wanted is by wanting something different. For example, if I have a preference for '1' but I know that someone is going to average my preferences with someone who prefers 0 then I know I will make '1' happen by modifying myself to prefer '2' instead of '1'. So averaging sucks.

Comment author: Will_Newsome 20 May 2011 01:34:18PM *  1 point [-]

Yeah, evolutionary (in the Universal Darwinian sense that includes Hebbian learning) incentives for a belief, attention signal, meme, or person to game differential comparisons made by overseer/peer algorithms (who are themselves just rent-seeking half the time) whenever possible is a big source of dukkha (suffering, imperfection, off-kilteredness). An example at the memetic-societal level: http://lesswrong.com/lw/59i/offense_versus_harm_minimization/3y0k .

In the torture/specks case it's a little tricky. If no one knows that you're going to be averaging their preferences and won't ever find out, and all of their preferences are already the result of billions of years of self-interested system-gaming, then at least averaging doesn't throw more fuel on the fire. Unless preferences have evolved to exaggerate themselves to game systems-in-general due to incentives caused by the general strategy of averaging preferences, in which case you might want to have precommited to avoid averaging. Of course, it's not like you can avoid having to take the average somewhere, at some level of organization...

Comment author: Perplexed 20 May 2011 02:20:12PM 0 points [-]

So averaging sucks.

Averaging by taking the mean sucks. Averaging by taking the median sucks less. It is a procedure relatively immune to gaming by would-be utility monsters.

The median is usually the 'right' utilitarian algorithm in any case. It minimizes total collective distance from the 'average'. The mean minimizes total collective distance^2 from the 'average'. There is no justification for squaring.

Comment author: magfrump 20 May 2011 11:42:11PM 2 points [-]

Is there a justification for not-squaring?

What's the appropriate metric on the space of preferences? This seems like something people would have different opinions about; i.e. "People who are smart should have more say!" "People who have spent more time self-reflecting should have more say!" "People who make lifestyle choices like this should be weighted more heavily!" "People who agree with me should have more say!"

Depending on the distribution, squaring could be better, because more (might be) lost as you get further away. And of course you can only take the median if your preferences are one dimensional.

Comment author: Perplexed 21 May 2011 12:11:20AM 0 points [-]

Is there a justification for not-squaring?

Personally, I am unconvinced that there is any fundamental justification for considering anyone's utility but one's own. But, if you have reason to respect the principles of democracy, the median stands out as the unique point acceptable to a majority. That is, if you specify any other point, a majority would vote to replace that point by the median.

What's the appropriate metric on the space of preferences?

That depends on what kinds of preferences you are comparing. If you are looking at the preferences of a single person, the standard construction of that person's utility function sets the "metric". But if you attempt to combine the preferences of two people, you either need to use the Nash Bargaining solution or Harsanyi's procedure for interpersonal comparison. The first gives a result that is vaguely median-like. The second gives an answer that is suitable for use with the mean.