G0W51

Karma: 218

G0W51 Jun 12, 2015, 5:27 AM
0 points
on: Open Thread, Jun. 8 - Jun. 14, 2015
I’ve been reading the discussion between Holden et al on the utility of charities aimed at directly decreasing existential risk, but the discussion seems to have ended prematurely. It (basically) started with this post, then went to this post. Holden made a comment addressing the post, but I think it didn’t fully address the post and I don’t think Holden’s comment was fully addressed either. Is there any place that continues the discussion?

G0W51 Jun 12, 2015, 5:11 AM
1 point
in reply to: HoldenKarnofsky’s comment on: Bayesian Adjustment Does Not Defeat Existential Risk Charity

That said, I don’t accept any of the arguments given here for why it’s unacceptable to assign a very low probability to a proposition. I think there is a general confusion here between “low subjective probability that a proposition is correct” and “high confidence that a proposition isn’t correct”; I don’t think those two things are equivalent.

I don’t think you’ve really explained why you don’t accept the arguments in the post. Could you please explain why and how the difference between assigning low probability to something and having high confidence it’s incorrect is relevant? I have several points to discuss, but I need to fully understand your argument before doing so.

And yes, I know I am practicing the dark art of post necromancy. But the discussion has largely been of great quality and I don’t think your comment has been appropriately addressed.

G0W51 Jun 12, 2015, 4:18 AM
0 points
in reply to: steven0461’s comment on: Bayesian Adjustment Does Not Defeat Existential Risk Charity

Karnofsky has, as far as I know, not endorsed measures of charitable effectiveness that discount the utility of potential people.

Actually, according to this transcript on page four, Holden finds that the claim that the value of creating a life is “some reasonable” ratio of the value of saving a current life is very questionable. More exactly, the transcript sad:

Holden: So there is this hypothesis that the far future is worth n lives and this causing this far future to exist is as good as saving n lives. That I meant to state as an accurate characterization of someone else’s view.

Eliezer: So I was about to say that it’s not my view that causing a life to exist is on equal value of saving the life.

Holden: But it’s some reasonable multiplier.

Eliezer: But it’s some reasonable multiplier, yes. It’s not an order of magnitude worse.

Holden: Right. I’m happy to modify it that way, and still say that I think this is a very questionable hypothesis, but that I’m willing to accept it for the sake of argument for a little bit. So yeah, then my rejoinder, as like a parenthetical, which is not meant to pass any Ideological Turing Test, it’s just me saying what I think, is that this is very speculative, that it’s guessing at the number of lives we’re going to have, and it’s also very debatable that you should even be using the framework of applying a multiplier to lives allowed versus lives saved. So I don’t know that that’s the most productive discussion, it’s a philosophy discussion, often philosophy discussions are not the most productive discussions in my view.

G0W51 Jun 10, 2015, 8:00 AM
0 points
in reply to: Philosophist’s comment on: Reason Poetry: f(me.0)
I don’t think you were being obtuse. Your post wasn’t bad per se, it was just off-topic as, unless I am failing to interpret it, it doesn’t really add anything new to rationality or applied rationality.

Also, in case you had trouble locating the open thread, just click “Discussion” at the top of the page, then click the link under “Latest Open Thread” on the right of the page.

G0W51 Jun 8, 2015, 5:46 AM
1 point
on: Reason Poetry: f(me.0)
I don’t think Less Wrong discussion is the best place for poetry, though someone please correct me if I am mistaken.

G0W51 Jun 7, 2015, 11:55 PM
0 points
in reply to: Eliezer Yudkowsky’s comment on: Pascal’s Muggle: Infinitesimal Priors and Strong Evidence
You’re absolutely right. I’m not sure how I missed or forgot about reading that.

G0W51 Jun 7, 2015, 8:39 PM
0 points
in reply to: Eliezer Yudkowsky’s comment on: Pascal’s Muggle: Infinitesimal Priors and Strong Evidence
The article said the leverage penalty “[penalizes] hypotheses that let you affect a large number of people, in proportion to the number of people affected.” If this is all the leverage penalty does, then it doesn’t matter if it takes 3^^^3 atoms or units of computation, because atoms and computations aren’t people.

That said, the article doesn’t precisely define what the leverage penalty is, so there could be something I’m missing. So, what exactly is the leverage penalty? Does it penalize how many units of computation, rather than people, you can affect? This sounds much less arbitrary than the vague definition of “person” and sounds much easier to define: simply divide the prior of a hypothesis by the number of bits flipped by your actions in it and then normalize.

G0W51 Jun 7, 2015, 2:23 AM
2 points
on: Pascal’s Muggle: Infinitesimal Priors and Strong Evidence
What if the mugger says he will give you a single moment of pleasure that is 3^^^3 times more intense than a standard good experience? Wouldn’t the leverage penalty not apply and thus make the probability of the mugger telling the truth much higher?

I think the real reason the mugger shouldn’t be given money is that people are more likely to be able to attain 3^^^3 utils by donating the five dollars to an existential risk-reducing charity. Even though the current universe presumably couldn’t support 3^^^3 utils, there is a chance of being able to create or travel to vast numbers of other universes, and I think this chance is greater than the chance of the mugger being honest.

Am I missing something? These points seem too obvious to miss, so I’m assigning a fairly large probability to me either being confused or that these were already mentioned.

G0W51 Jun 4, 2015, 5:31 AM
1 point
on: Logical and Indexical Uncertainty
The second problem can easily be explained by having your utility function not be linear in the number of non-destroyed universes.

G0W51 Jun 1, 2015, 10:36 PM
1 point
on: Open Thread, Jun. 1 - Jun. 7, 2015
Is Solomonoff induction a theorem for making optimal probability distributions or a definition of them? That is to say, has anyone proved that Solomonoff induction produces probability distributions that are “optimal,” or was Solomonoff induction created to formalize what it means for a prediction to be optimal. In the former case, how could they define optimality?

(And another question: I posted this a couple days ago on the last open thread, but it was closed before I got a response. Is it okay to repost it?)

G0W51 May 31, 2015, 2:27 AM
1 point
on: Open Thread, May 25 - May 31, 2015
Is Solomonoff induction a theorem for making optimal probability distributions or a definition of them? That is to say, has anyone proved that Solomonoff induction produces probability distributions that are “optimal,” or was Solomonoff induction created to formalize what it means for a prediction to be optimal. In the former case, how could they define optimality?

G0W51 May 30, 2015, 10:48 PM
0 points
in reply to: Lumifer’s comment on: How can I reduce existential risk from AI?
Noted. I’ll invest my efforts on x-risk reduction into something other than voting.

G0W51 May 25, 2015, 2:29 AM
0 points
in reply to: ChristianKl’s comment on: How can I reduce existential risk from AI?
There are already are well-argued articles, I’m not sure how useful more articles would be. Perhaps a more accessible version of Existential Risk as a Global Priority would be useful, though.

G0W51 May 25, 2015, 2:22 AM
0 points
in reply to: ChristianKl’s comment on: How can I reduce existential risk from AI?
Yeah, I suppose you’re right. Still, once something that could pose a large existential risk comes into existence or looks like it will soon come into existence, wouldn’t politicians then consider existential risk reduction? For example, once a group is on the verge of developing AGI, wouldn’t the government think about what to do about it? Or would they still ignore it? Would the responses of different parties vary?

You could definitely be correct, though; I’m not knowledgeable about politics.

G0W51 May 24, 2015, 1:45 AM
−1 points
in reply to: Lumifer’s comment on: How can I reduce existential risk from AI?
Why not? I imagine that different political parties have different views on what the government should do about existential risk and voting for the ones that are potentially more willing to decrease it would be beneficial. Currently, it seems like most parties don’t concern themselves at all with existential risk, but perhaps this will change once strong AI becomes less far off.

G0W51 May 24, 2015, 1:44 AM
0 points
in reply to: ChristianKl’s comment on: How can I reduce existential risk from AI?
Remember that there may still be some value in voting for candidates who aren’t mainstream.

G0W51 May 19, 2015, 10:39 PM
0 points
on: How can I reduce existential risk from AI?
I know discussing politics on LW is discouraged, but is voting in elections a viable method of decreasing existential risk by making it more likely that those who are elected will take more action to decrease it? If so, what parties should be voted for? If this isn’t something that should be discussed on LW, just say so and I can make a reddit post on it.

G0W51 May 19, 2015, 9:36 PM
1 point
in reply to: Capla’s comment on: Should we go all in on existential risk? - Considering Effective Altruism
Given that you are only one individual, the increase in the chance of a p-sing for each unit of money you give is roughly linear, so diminishing marginal returns shouldn’t be an issue.

G0W51 13 Apr 2015 23:10 UTC
7 points
on: Open Thread, Apr. 13 - Apr. 19, 2015
Perhaps multiple choice tests in schools make people extra susceptible to privileging hypotheses. As a simplified example, if a student’s probability distribution to the answer of a question on a multiple choice test before seeing the choices is uniformly distributed amongst all integers from 1 to n, simply seeing an arbitrary integer as one of the, say, four options in a multiple choice test justifiably increases its probability of being correct to 0.25, a tremendous increase when n is large. Thus, on multiple choice tests, privileging a possible answer being “suggested” by being one of the answers the student can pick is a useful strategy on multiple choice tests, so we need to remember that privileging the hypothesis is not a viable strategy in other contexts. Maybe multiple choice tests aren’t so good after all...

G0W51 10 Apr 2015 3:24 UTC
0 points
in reply to: NancyLebovitz’s comment on: Open Thread, Apr. 06 - Apr. 12, 2015

I’ve seen not-especially-theoretical discussion about this problem for human organizations, though mostly from the point of view of lower status people complaining that they’re being given impossible, incomprehensible, and/or destructive commands.

Remember that making humans in organizations cooperate is a rather different from making the many parts of an AI cooperate with the other parts, because people in organizations can’t (feasibly) be reprogrammed to have their values be aligned, but AI HLAs can.

What I’m concluding from this is that if Friendliness is to work, it has to pervade the hierarchy of agents.

True. The real issue is that if you give a lower-level action the same goals and utility functions as the higher ones, you’ve lost all benefit of having HLAs!.