utilitymonster

Karma: 339

utilitymonster Jul 3, 2010, 2:13 PM
4 points
on: Open Thread: July 2010
Is there a principled reason to worry about being in a simulation but not worry about being a Boltzmann brain?

Here are very similar arguments:
- If posthumans run ancestor simulations, most of the people in the actual world with your subjective experiences will be sims.
- If two beings exist in one world and have the same subjective experiences, your probability that you are one should equal your probability that you are the other.
- Therefore, if posthumans run ancestor simulations, you are probably a sim.
vs.
- If our current model of cosmology is correct, most of the beings in the history of the universe with your subjective experiences will be Boltzmann brains.
- If two beings exist in one world and have the same subjective experiences, your probability that you are one should equal your probability that you are the other.
- Therefore, if our current model of cosmology is correct, you are probably a Boltzmann brain.
Expanding your evidence from your present experiences to all the experiences you’ve had doesn’t help. There will still be lots more Boltzmann brains that last for as long as you’ve had experiences, having experiences just like yours. Most plausible ways of expanding your evidence have similar effects.

I suppose you could try arguing that the Boltzmann brain scenario, but not simulation scenario, is self-defeating. In the Boltzmann scenario, your reasons for accepting the theory (results of various experiments, etc) are no good, since none of it really happened. In the simulation scenario, you really did see those results, all the results were just realized in a funny sort of way that you didn’t expect. It would be nice if the relevance of this argument were better spelled out and cashed out in a plausible Bayesian principle.

edited for format

utilitymonster Jun 1, 2010, 5:31 PM
0 points
in reply to: jimrandomh’s comment on: Diseased thinking: dissolving questions about disease

Consequentialist value systems are a huge class; of course not all consequentialist value systems are praiseworthy! But there are terrible agent-neutral value systems, too, including conventional value systems with an extra minus sign, Clippy values, and plenty of others.

Yeah, the objection wasn’t supposed to be that because there was an implausible consequentialist view on that definition of “consequentialism”, it was a bad definition. The objection was that pretty much any maximizing view could count as consequentialist, so the distinction isn’t really worth making.

utilitymonster Jun 1, 2010, 4:36 PM
0 points
in reply to: jimrandomh’s comment on: Diseased thinking: dissolving questions about disease
This is where things get murky. The traditional definition is this:

Consequentialism: an act is right if no other option has better consequences

You can say that it is consistent with consequentialism (in this definition) to favor yourself, as such, only if you think that situations in which you are better off are better than situations in which a relevantly similar other is better off. Unless you think you’re really special, you end up thinking that the relevant sense of “better” is relative to an agent. So some people defend a view like this:

Agent-relative consequentialism: For each agent S, there is a value function Vs such that it is right for S to A iff A-ing maximizes value relative to Vs.

When a view like this is on the table, consequentialism starts to look pretty empty. (Just take the value function that ranks outcomes solely based on how many lies you personally tell.) So some folks think, myself included, that we’d do better to stick with this definition:

Agent-neutral consequentialism: There is an agent-neutral value function v such that an act is right iff it maximizes value relative to v.

I don’t think there is a lot more to say about this, other than that paradigm historical consequentialists rejected all versions of agent-relative consequentialism that allowed the value function to vary from person to person. Given the confusion, it would probably be best to stick to the latter definition or always disambiguate.

utilitymonster Jun 1, 2010, 3:26 PM
1 point
in reply to: Vladimir_Nesov’s comment on: Diseased thinking: dissolving questions about disease
Our question is this: is there a consequentialist view according to which it is right for someone to care more about his own welfare, as such? I said there is no such view, because consequentialist theories are agent-neutral (i.e., a consequentialist value function is indifferent between outcomes that are permutations of each other with respect to individuals and nothing else; switching Todd and Steve can’t make an outcome better, if Steve ends up with all of the same properties as Todd and vice versa).

I agree that a preference utilitarian could believe that in a version of the example I described, it could better to help yourself. But that is not the case I described, and doesn’t show that consequentialists can care extra about themselves, as such. My “consequentialist” said:

“I should get the morphine. It would be better if I got it, and the only reason it would be better is because I was me, rather than him, who received it.”

Yours identifies a different reason. He says, “I should get the morphine. This is because there would be more total preference satisfaction if I did this.” This is a purely agent-neutral view.

My “consequentialist” is different from your consequentialist. Mine doesn’t think he should do what maximizes preference satisfaction. He maximizes weighted preference satisfaction, where his own preference satisfaction is weighted by a real number greater than 1. He also doesn’t think his preferences are more important in some agent-neutral sense. He thinks that other agents should use a similar procedure, weighing their own preferences more than the preferences of others.

You can bring out the difference between by considering a case where all that matters to the agents is having a minimally painful death. My “consequentialist” holds that even in this case, he should save himself (and likewise for the other guy). I take it that on the view you’re describing, saving yourself and saving the other person are equally good options in this new case. Therefore, as I understand it, the view you described is not a consequentialist view according to which agents should always care more about themselves, as such. Perhaps we are engaged in a terminological dispute about what counts as caring about your own welfare more than the welfare of others, just because it is yours?

utilitymonster Jun 1, 2010, 2:16 PM
1 point
in reply to: Vladimir_Nesov’s comment on: Diseased thinking: dissolving questions about disease

Consequentialism doesn’t work by reversing the hedonic/deontological error of focusing on the agent, by refusing to consider the agent at all. A consequentialist cares about what happens with the whole world, the agent included.

I hope I didn’t say anything to make you think I disagree with this.

I’d say it’s the only correct understanding for human consequentialists to care especially about themselves, though of course not exclusively.

I noted that there might be instrumental reasons to care about yourself extra if you’re a consequentialist. But how could one outcome be better than another, just because you, rather than someone else, received a greater benefit?

Example: you and another person are going to die very soon and in the same kind of way. There is only enough morphine for one of you. Apart from making one of your deaths less painful, there will be nothing relevant hangs on who gets the morphine.

I take it that it isn’t open to the consequentialist to say, “I should get the morphine. It would be better if I got it, and the only reason it would be better is because I was me, rather than him, who received it.”

utilitymonster Jun 1, 2010, 10:57 AM
1 point
in reply to: Nick_Tarleton’s comment on: Diseased thinking: dissolving questions about disease
Well, you clearly could be extra concerned about your own welfare because it is instrumentally more valuable than the welfare of others (if you’re happy you do more good than your neighbor, perhaps). Or, you could be a really great guy and hold the view that it’s good for great guys to be extra happy. But I was thinking that if you thought that your welfare was extra important just because it was yours you wouldn’t count as a consequentialist.

As I was mentioning in the last post, there’s some controversy about exactly how to spell out the consequentialist/non-consequentialist distinction. But probably the most popular way is to say that consequentialists favor promoting agent-neutral value. And thinking your welfare is special, as such, doesn’t fit that mold.

Still, there are some folks who say that anyone who thinks you should maximize the promotion of some value or other counts as a consequentialist. I think this doesn’t correspond as well to the way the term is used and what people naturally associate with it, but this is a terminological point, and not all that interesting.

utilitymonster Jun 1, 2010, 2:02 AM
6 points
in reply to: Scott Alexander’s comment on: Diseased thinking: dissolving questions about disease

you’d need a pretty convoluted consequentialist system to promote blame (and if you were willing to go that far, you could call a deontologist someone who wants to promote states of the world in which rules are followed and bad people are punished, and therefore a consequentialist at heart). Likewise, you could imagine a preference utilitarian who wants people to be punished just because e or a sufficient number of other people prefer it.

I’m not sure how complicated it would have to be. You might have some standard of benevolence (how disposed you are to do things that make people happy) and hold that other things being equal, it is better for benevolent people to be happy. True, you’d have to specify a number of parameters here, but it isn’t clear that you’d need enough to make it egregiously complex. (Or, on a variant, you could say how malevolent various past actions are and hold that outcomes are better when malevolent actions are punished to a certain extent.)

Also, I don’t think you can do a great job representing deontological views as trying to minimize the extent to which rules are broken by people in general. The reason has to do with the fact that deontological duties are usually thought to be agent-relative (and time-relative, probably). Deontologists think that I have a special duty to see to it that I don’t break promises in a way that I don’t have a duty to see to it that you don’t break promises. They wouldn’t be happy, for instance, if I broke a promise to see to it that you kept two promises of roughly equal importance. Now, if you think of the deontologists as trying to satisfy some agent-relative and time-relative goal, you might be able to think of them as just trying to maximize the satisfaction of that goal. (I think this is right.) If you find this issue interesting (I don’t think it is all that interesting personally), googling “Consequentializing Moral Theories” should get you in touch with some of the relevant philosophy.

utilitymonster May 31, 2010, 10:43 AM
2 points
in reply to: Nick_Tarleton’s comment on: Diseased thinking: dissolving questions about disease
Agreed. (Though I agree with the general structure of your post.)

A better name for your position might be “basic desert skepticism”. On this view, no one is intrinsically deserving of blame. One reason is that I don’t think the determinism/indeterminism business really settles whether it is OK to blame people for certain things. As I’m sure you’ve heard, and I’d imagine people have pointed out on this blog, the prospects of certain people intrinsically deserving blame, independently of benefits to anyone, are not much more cheering if everything they do is a function of the outcome of indeterministic dynamical laws.

Another reason is that you can have very similar opinions if you’re not a consequentialist. Someone might believe that it is quite appropriate, in itself, to be extra concerned about his own welfare, yet agree with you about when it is a good idea to blame folks.

utilitymonster May 27, 2010, 11:19 AM
4 points
on: Abnormal Cryonics
Thanks for this post. I tend to lurk, and I had some similar questions about the LW enthusiasm for cryo.

Here’s something that puzzles me. Many people here, it seems to me, have the following preference order:

pay for my cryo > donation: x-risk reduction (through SIAI, FHI, or SENS) > paying for cryo for others

Of course, for the utilitarians among us, the question arises: why pay for my cryo over risk reduction? (If you just care about others way less than you care about yourself, fine.) Some answer by arguing that paying for your own cryo maximizes x-risk reduction better than the other alternatives because of its indirect effects. This reeks of wishful thinking and doesn’t fit well with the preference order above. There are plenty of LWers, I assume, who haven’t signed up for cryo, but would if someone else would pay the life insurance policy. If you really think that paying for your own cryo maximizes x-risk reduction, shouldn’t you also think that getting others signed up for cryo does as well? (There are some differences, sure. Maybe the indirect effects aren’t as substantial if others don’t pay their own way in full. But I doubt this justifies the preference.) If so, it would seem that rather than funding x-risk reduction through donating to these organizations, you should fund the cryo preservation of LWers and other willing people.

So which is it utilitarians: you shouldn’t pay for your own cryo or you should be working on paying for the cryo of others as well?

If you think paying for cryo is better, want to pay for mine first?

utilitymonster May 11, 2010, 12:37 AM
0 points
in reply to: thomblake’s comment on: Rationality quotes: May 2010
I’m not sure I understand what you mean when you say

This seems at odds with our notion of subjective probability, where we assume that significant lingering doubt after confidently assigning a 99%+ probability is evidence that your calibration is poor, and your estimate should have been lower.

The most confusing part about this is the part about poor calibration.

As for the rest, I don’t deny that the fact that the man is unwilling to undertake the voyage is evidence that he doesn’t think it is worthwhile, at least in ordinary contexts. But I think there is little to recommend the view that acting against your best reflective judgment is impossible or even extremely rare.

utilitymonster May 10, 2010, 12:47 PM
4 points
in reply to: pricetheoryeconomist’s comment on: Rationality quotes: May 2010
Fair enough. Maybe it is typically reasonable to charge people with hypocrisy when they neglect to follow their professed ethical codes.

I still like the quote, even if it is hyperbolic. It is useful to be reminded that there are important cases where failure to live up to one’s professed code does not warrant this kind of criticism. Being overly concerned with hypocrisy can make you be unconcerned with living up to a meaningful ethical code. This is especially important in the context of consequentialist morality. This is just a hunch, but I think there are a fair number of intelligent people who shy away from a demanding code for fear of being charged with hypocrisy. But there need be no genuine hypocrisy, at least in any deeply regrettable sense, in professing a demanding ethical code and failing to live up to it. Better to try to live up to a demanding code and fail than meet the demands of an uninspiring and mundane one. (In this kind of case, of course, you aren’t just professing the code to curry political favor.)

utilitymonster May 8, 2010, 4:18 PM
20 points
0
on: Rationality quotes: May 2010

Nothing is more unjust, however common, than to charge with hypocrisy him that expresses zeal for those virtues which he neglects to practice; since he may be sincerely convinced of the advantages of conquering his passions, without having yet obtained the victory, as a man may be confident of the advantages of a voyage, or a journey..., without having courage or industry to undertake it, and may honestly recommend to others, those attempts which he neglects himself.

--Samuel Johnson

utilitymonster Apr 20, 2010, 11:12 PM
5 points
on: Attention Lurkers: Please say hi
Hi.

utilitymonster Apr 19, 2010, 12:31 PM
6 points
on: Welcome to Less Wrong!
I’m a philosophy PhD student. I studied math and philosophy as an undergrad. I work on ethics and a smattering of Bayesian topics. I care about maximizing the sum of desirable experiences that happen in the future. In less noble moments, I care more immediately about advancing my career as a philosopher and my personal life.

I ran into OB a couple years ago when Robin Hanson came and gave a talk on disagreement at a seminar I was attending. I started reading OB, and then I drifted to LW territory a few months ago.

At first, much of the discussion here sounded crazy to me. It often still does. But I thought I’d give it a detailed look, since everyone here seems to have the same philosophical prejudices as me (Bayesian utilitarian atheist physicalists).

I like discussion of Bayesian topics and applied ethics best.

utilitymonster Apr 2, 2010, 1:45 AM
−1 points
in reply to: Cyan’s comment on: The I-Less Eye
I think this guy disagrees: Weatherson, Brian. Should We Respond to Evil with Indifference? Philosophy and Phenomenological Research 70 (2005): 613-35. Link: http://brian.weatherson.org/papers.shtml

utilitymonster Mar 26, 2010, 3:59 PM
10 points
in reply to: Drahflow’s comment on: The “show, don’t tell” nature of argument
At least in philosophy (my discipline), people often say ‘it can be argued that p’ when arguments for p have been produced by others, and these arguments are thought to have considerable force, but it would take the discussion too far astray to rehearse those arguments. It is sometimes worthwhile to say this, rather than producing an argument for p.

This isn’t just for people who are already familiar with those arguments for p. Knowing these arguments exist is often importance evidence related to p, and can guide one’s future thinking.

utilitymonster Mar 18, 2010, 1:33 PM
0 points
on: Let There Be Light
“Where do you get your priors, when you start modeling yourself seriously instead of doing it by halfhearted intuition?”

Is the thought that here that you should try to find out what your priors ought to be by figuring out how you work psychologically? That seems odd. If you think some priors are special, why think they’d be the ones that you’d have in particular? If you think no priors are special, why bother trying to use your own?

Apart from the intrinsic interest of self-knowledge, I would have thought the point of making your psychological life more luminous would be to avoid known biases and misfiring heuristics that have bad consequences (epistemic and otherwise). Of particular interest here would be information about your current credences (which probably don’t form a probability function, though they may approximate one in good cases) and your updating method (which probably isn’t a form of conditionalization, though it may approximate one in good cases). Knowing your credences and updating method could help you know when you’re subject to biases and misfiring heuristics.

But note that this other goal does seem to rely on “intuition” (scare quotes because no one knows that intuitions are) at some level (though I don’t know if its half-hearted). You’ll have to come to the table with some assumptions about what would be irrational so that you can compare your judgments to that standard. It seems doubtful to me that such assumptions could be vindicated purely empirically, since the questions seem thoroughly normative. It seems doubtful, for instance, that you could justify these assumptions about what would make some pieces of reasoning rational by appeal to psychological facts about yourself or anyone else.

utilitymonster Feb 17, 2010, 3:13 AM
5 points
on: Open Thread: February 2010, part 2
I’m new to Less Wrong. I have some questions I was hoping you might help me with. You could direct me to posts on these topics if you have them. (1) To which specific organizations should Bayesian utilitarians give their money? (2) How should Bayesian utilitarians invest their money while they’re making up their minds about where to give their money? (2a) If your answer is “in an index fund”, which and why?