Jiro

Karma: 5,312

Jiro 21 Mar 2026 18:51 UTC
2 points
−1
in reply to: Bowl of Cereal’s comment on: It’s a Good Thing to Respond to Internet Trolls

Instead of focusing on saying “this is wrong,” try to add new information that is related to the discussion and that could stimulate readers’ minds

That allows the trolls to control the direction of the discussion by picking a topic and having other people add new information to it. Some topics are inherently inflammatory or inherently are likely to attract people who behave badly, so even letting the troll pick the topic can be a disaster. You do not want to have a troll arguing that Jews descend from Cthulu, provide “new information”, and attract people sincerely arguing that Jews are not descendants of Cthulhu but are as evil as if they were, or even just Holocaust deniers.

Jiro 21 Mar 2026 15:20 UTC
2 points
0
in reply to: AlphaAndOmega’s comment on: My Willing Complicity In “Human Rights Abuse”
You get to specify that the trolley doesn’t have brakes because you can say “I will only apply the conclusion I am getting from the trolley problem in real life situations that are similar to the trolley problem in that there are no (metaphorical) brakes.”

The equivalent for your slavery contract is “I will permit slavery only in situations where I will starve if a slavery contract is not allowed”. You can’t do this, because you can’t have a slavery law that applies only in a specific, unlikely, scenario. A law permitting slavery will apply in the vastly more common scenario where the slaver is still incentivised to get some money from you and sell you the food.

Jiro 21 Mar 2026 13:26 UTC
4 points
1
in reply to: AlphaAndOmega’s comment on: My Willing Complicity In “Human Rights Abuse”
It’s “wrong” because you describe a scenario where if slavery is prohibited, you will starve, but in that scenario if slavery is prohibited you won’t starve—because if slavery is banned, the slaver is still incentivized to sell you the food at a price less than a slave contract.

You can’t just say “it’s my scenario so that doesn’t happen”, because to do that you have to have a scenario where incentives don’t work, and if you do the scenario has no relevance to the real world.

Jiro 21 Mar 2026 4:13 UTC
4 points
1
in reply to: AlphaAndOmega’s comment on: My Willing Complicity In “Human Rights Abuse”
But one argument you gave is:

I would be mad if someone else stepped in and prevented the transaction on abolitionist grounds. It beats starving to death.

This argument is flat out wrong. If slavery is forbidden, the would be slaveowner is incentivized to sell you the food for a price that is large but less than a slave contract. Thus, you would not starve to death, and in fact you would be strictly better off than if the slave contract was allowed.

Jiro 19 Mar 2026 22:55 UTC
4 points
−2
in reply to: AlphaAndOmega’s comment on: My Willing Complicity In “Human Rights Abuse”
Incentives matter. If it was illegal for someone to make a starving person pay for food with an indentured servant contract, he could just refuse to sell you the food, but his incentive is to sell it to you at the maximum legal price. The result in the real abolitionist world would be that the starving person is better off, and the person selling the food only gets a high fee but misses out on an indentured servant contract.

Libertarians usually can’t handle fixing situations with bad incentives or improving incentives because people respond to the bad incentives voluntarily, so there is no way for a libertarian to decide that prohibiting the bad incentive can improve things

I do not intend to fully generalize this argument, even if I understand it can be interpreted that way.

If you don’t intend to generalize it, on what grounds do you decide that it doesn’t need to be generalized? After all it looks as though the reasoning still applies. That seems like it would foreclose refusing to generalize it.

Jiro 17 Mar 2026 17:03 UTC
3 points
0
on: New LessWrong Editor! (Also, an update to our LLM policy.)
If I think someone else’s text (either from here or from elsewhere on the Internet) is likely unlabelled LLM text, and I wish to quote it here, what should I do under this policy? Neither labelling nor not labelling it as LLM really seems appropriate.

Jiro 14 Mar 2026 3:46 UTC
2 points
−1
in reply to: Chris_Leong’s comment on: The Refined Counterfactual Prisoner’s Dilemma: An Attempt to Explode Decision-Theoretic Consequentialism
Giving you a limited amount to decide, where that limited amount of time means that you can’t use some lines of reasoning, means that the question is now “what should an imperfect reasoner do when faced with Omega”. The answer to this question 1) depends on the details of the imperfect reasoning and 2) leads to rather less insight on the main question.

Jiro 14 Mar 2026 3:43 UTC
2 points
−2
in reply to: Algon’s comment on: Fake Updates
The basic problem is epistemic learned helplessness. You know that your own reasoning process isn’t perfect and that there are arguments that are wrong, but which you can’t detect as wrong. In other words, “once you fully understand a point” is a state that you, an imperfect reasoner, can’t know you are in. Advice on what to do in that state is therefore useless. You need advice on what to do when you seem to be in that state, which may be different than advice on what to do if you know for sure that you’re in that state.

Jiro 12 Mar 2026 8:36 UTC
2 points
0
on: The Refined Counterfactual Prisoner’s Dilemma: An Attempt to Explode Decision-Theoretic Consequentialism
Is Omega meant to be immune to the Halting Problem here? (Your own strategy could conceivably depend on trying to predict what Omega would do under some circumstances.) Or are you supposed to be an imperfect reasoner (which may be true in real life, but isn’t normally assumed to be true in these hypothetical scenarios)?

Jiro 11 Mar 2026 23:37 UTC
2 points
0
on: Veganism is Necessary
Consider the example that I often use to refute this sort of thing: Video game characters. We have no idea exactly where future AIs will draw the line. It is entirely possible that future AIs will think “humans kill a lot of video game characters, so it’s okay for us to kill lots of humans”.

Of course, this sounds ludicrous because nobody thinks that killing video game characters is wrong, but some people think killing animals is wrong. But if you’re postulating future intelligences whose moral system will be broad enough to save us only if our own system is broad enough, we really don’t know exactly how broad it has to be for this to work. Neither do we know that it will match something like veganism that humans actually believe in.

In fact, we can generalize this. It’s the same as one of the problems with Pascal’s Wager: the wager applies to all gods and even hypothetical gods that don’t have religions, and you don’t know which one to follow. Likewise, the “AI wager” applies to all ideologies, not just to veganism, and including ludicrous ones that I just made up.

Jiro 10 Mar 2026 18:29 UTC
−1 points
−2
on: The case for satiating cheaply-satisfied AI preferences
I say no. See https://www.lesswrong.com/posts/bWdtxtyTiEsxSZtK3/a-world-without-violet-peculiar-consequences-of-granting .

I would also add that people are often poorly calibrated on what “cheaply satisfied” means, and dismiss personal inconvenience or displeasure as not truly being costs. If the AI prefers that nobody in the world use the color violet, that’s not a cheaply satisfied preference, even if eliminating violet just means more personal inconvenience.

Jiro 10 Mar 2026 16:54 UTC
2 points
2
on: Fake Updates
I’m pretty sure there was a post about why effortless updates are a bad idea. I couldn’t find the exact post, but I suggest remembering epistemic learned helplessness. The correct thing to do when faced with an argument that seems to disprove one of your beliefs is usually to nod politely and ignore it. After you have 1) thought about it for a while and 2) observed experts changing their mind on it, you may then actually consider adopting it.

Also, remember Yudkowsky’s AI box experiment where the experimenter playing an AI in a box can convince someone to let it out. The “pretty good point” may be the equivalent of Eliezer convincing you to let the AI out because he is convincing, not because he actually has a pretty good point. (This goes double for actual AIs.)

Jiro 9 Mar 2026 19:42 UTC
2 points
0
in reply to: Andrew_Critch’s comment on: Schelling Goodness, and Shared Morality as a Goal
Because that’s a variable value principle and it’s also been proposed as a solution to the Repugnant Conclusion, and it doesn’t work.

Jiro 8 Mar 2026 21:52 UTC
2 points
−2
in reply to: Andrew_Critch’s comment on: Schelling Goodness, and Shared Morality as a Goal
The problem with “grass has almost no moral value” is not that you then need to argue the magnitude of that moral value. It’s that no matter what that magnitude is, there will be some point where enough grass becomes the most important thing morally. If you want to believe that a human always outweighs grass morally, and you believe that moral comparisons are even possible, you must believe that grass has zero morality, or your beliefs contradict themselves.

Jiro 28 Feb 2026 2:54 UTC
2 points
0
on: What is a species?

Most alarming of all (from a scientific perspective), some taxonomists use political considerations to define a species, calling a population a new species just to bring attention to it or give it legal protection, even when everyone in the room privately believes it’s not scientifically accurate. There is no consistent rule being applied.

If they are aware that it isn’t scientifically accurate, and they call it a species anyway, they are lying. It’s just that it’s hard to prove that they are lying without being able to read their mind, and they could probably say something that makes it superficially sound like a species that would be hard to rebut in a couple of seconds.

The fact that someone can undetectably lie about whether something is a species doesn’t affect the definition of species at all.

Jiro 23 Feb 2026 0:29 UTC
5 points
3
on: Defending Classical Utilitarianism

Ultimately, I think that people conclude that happiness is bad not because it is bad, but instead because they have strong social pressures to believe so. I believe that, given enough time and resources to engage in meaningful moral reflection, they would come to the belief that happiness is morally good.

You are just dismissing disagreement. Blissful ignorance results in happiness. I believe that it is bad anyway. Obviously I can’t prove that this is not due to social pressures, but that claim is unfalsifiable.

You also don’t seem to be addressing common criticisms. For instance, your post makes no mention of utility monsters, or of the repugnant conclusion.

Jiro 17 Feb 2026 0:44 UTC
9 points
2
on: Contra Caplan on higher education

A key problem with Caplan’s trinity is that most of it is easily replaceable. Getting good grades at college does signal intelligence and conscientiousness, but these could be signaled far more easily and cheaply. It’s very easy to signal intelligence via test scores: IQ is surprisingly predictive of many other desirable cognitive traits. This need not require literal IQ tests—standardized tests like the SAT or GRE are highly correlated with intelligence.

Good luck not getting a disparate impact suit if you try this. You have to prove that the test that you’re using is necessary for the job. College is basically always considered necessary. The coding interviews you mention are obviously necessary. Trying to use an IQ-like test or even the SAT (unless you’re a college) will get you in court.

Jiro 2 Feb 2026 17:22 UTC
4 points
0
on: The Smoking Lesion Doesn’t Really Distinguish EDT from CDT
My response to the smoking lesion question is:
1. If the lesion doesn’t affect your ability to reason, then deciding to smoke as the result of a reasoning process isn’t correlated with cancer, even though deciding to smoke in general is, so you should smoke.
2. If the lesion does affect your ability to reason, the question is ill defined because it may not be possible for you to execute some strategies, or only possible to execute them with some probability.
Of course “as the result of a reasoning process” needs to be phrased more precisely (every decision can be said to be the result of some reasoning process). Maybe “when you make some particular deduction, compared to a counterfactual world where you did not make that deduction”.

Jiro 2 Feb 2026 17:08 UTC
2 points
0
in reply to: James Diacoumis’s comment on: Would you kill a vulcan to save a shrimp?
You’re thinking of a p-Vulcan as a slight variation on a human. But the context includes:

Remember, it’s highly likely that shrimps have some form of phenomenal consciousness and experience some form of suffering. Shrimp suffering is bad. Even though we lack the capability to accurately predict its intensity, the shrimp certainly suffers more than the Vulcan would. Vulcans totally lack the capacity to suffer.

A p-Vulcan doesn’t have to be very humanlike. It only needs to be shrimplike, but without the ability to suffer. I do think that many existing video game characters would qualify, by that standard, as similar to a p-Vulcan. Is it wrong to kill those video game characters?

Jiro 2 Feb 2026 2:27 UTC
2 points
0
in reply to: James Diacoumis’s comment on: Would you kill a vulcan to save a shrimp?
Repeat the robots question, except ask the question about video game characters instead. Now, there are some game characters that have very simple patterns of behavior, but there are some that are a lot more complex, even if still describable by a set of algorithms. I’m sure there are characters that can and will beg you not to kill them. Is it wrong to play the video game and sacrifice the video game characters?