Jiro

Karma: 5,432

Jiro 2 Jul 2026 21:35 UTC
8 points
2
on: You Should Choose How You React to Your Feelings
This isn’t wrong, but rationalists sometimes take it to unhealthy extremes and need to do it less. I’ve often seen the argument that since you choose how to react, you don’t have to react when someone is being insulting. This puts the blame on the person who doesn’t like being insulted rather than on the person doing the insulting.

We need to recognize that some reactions are appropriate and need to be treated as such, even if you could choose not to make them.

Jiro 1 Jul 2026 7:56 UTC
2 points
0
in reply to: J Bostock’s comment on: Functional Decision Theory: Not Even Wrong, Also Wrong

If you want to extend decision theory to problems where the agent is deceived, or is punished because of following a particular decision theory,

The issue is that there may not be a choice. If you don’t want to extend decision theory to that kind of problem, exactly what are you going to do to exclude problems like that? It won’t necessarily have a line “if decision_theory == ‘XDT’” in it. It may not be very obvious, and it may not even be possible to determine, that some problem falls into a category that you want to exclude.

Jiro 1 Jul 2026 7:35 UTC
0 points
0
in reply to: Firinn’s comment on: That Which Cannot Be Poked With A Stick Is The Mind-Killer

Do you think I’m getting a dig at my political enemies just because I can?

...

Furthermore, since it is difficult to make a rule which allows only disruptive political asides to be rebutted, we should instead let all political asides be rebutted.

Saying “you can respond to political jabs, but only jabs” is unworkable.

Jiro 1 Jul 2026 0:22 UTC
−5 points
−3
on: That Which Cannot Be Poked With A Stick Is The Mind-Killer

Note: If you want to respond to this post with an object-level comment about immigration, you have to preface it with, “I didn’t read or understand this post, but:”

I would link Politics is the Mind-Killer but you already did… and that post specifically points out that using a political example is often an attempt by the writer to get a dig at their political enemies just because they can.

This behavior is disruptive, and to discourage it, we should consider it acceptable to rebut political statements made in posts. Furthermore, since it is difficult to make a rule which allows only disruptive political asides to be rebutted, we should instead let all political asides be rebutted. If you are using one for a legitimate reason and people start rebutting it, them’s the breaks.

Jiro 29 Jun 2026 23:05 UTC
5 points
1
in reply to: J Bostock’s comment on: Functional Decision Theory: Not Even Wrong, Also Wrong

For any decision theory XDT, it is possible to construct a world where Omega gives you one bazillion utilons if you don’t follow XDT, and murders you if you do follow XDT. This is part of why these problems are called “unfair”.

That’s like saying that the Halting Problem isn’t an issue because problems that involve self-reference are unfair. You can’t just avoid the Halting Problem by saying “no explicit self-reference”, because seemingly reasonable stipulations that don’t explicitly have self-reference in them may imply it anyway.

It may turn out that for some decision theories, reasonable-seeming problems that don’t explicitly say “Omega punishes you if you follow XDT” may be equivalent to “Omega punishes you if you follow XDT” anyway.

Jiro 28 Jun 2026 21:57 UTC
10 points
9
on: Surprising facts about the slave trade
Because modern Westerners vehemently oppose slavery for reasons unrelated to economic efficiency, arguments about how economically inefficient slavery is seem like they would be prone to motivated reasoning. Not a lot of people say “bad thing I hate is economically efficient”. In most places, even attempting to argue that slavery is economically efficient would get you in trouble. I certainly don’t expect scientific journals to publish papers claiming that slavery is economically efficient, even if it is.

Jiro 26 Jun 2026 21:59 UTC
2 points
0
in reply to: MichaelDickens’s comment on: A full body MRI earns you a year of smoking
I don’t agree. Something that takes away my entire life when I am unlucky is going to take away a number of weddings, graduations, or even smaller things like movies that I can’t see because I’m dead. Something that takes away 1 hour of my life that is not completely random won’t do this. It won’t do this even repeated over enough people to lose the same number of hours as a single person losing his entire life. I’ll have to sacrifice something in that hour, but I’ll sacrifice a low value part of my life, not an average one. (Perhaps the opportunity cost of taking some extra sick leave from work.)

(Technically, this doesn’t mean that QALYs are wrong—QALY means quality adjusted life years. Just adding up the hours under these circumstances is a failure to adjust them for quality, and therefore calculates them wrong.)

Jiro 26 Jun 2026 16:05 UTC
4 points
0
on: A full body MRI earns you a year of smoking
QALYs are a bad measure when it comes to adding up a small amount of time spent by a large number of individuals, because not all hours of life are equally valuable. When you take a small number of hours from a person, the hours come from a less valuable slack pool than if you take a lot of hours, so X number of hours from one person (as in an increased risk from smoking that gives an individual cancer) is not the same as X number of hours spread over many people (as in a lot of people spending time on tests).

Jiro 25 Jun 2026 20:41 UTC
4 points
0
on: If thine enemy be hungry, give him bread to eat

To make explicit the analogy to modern politics, you should be happy when your opponent wins an election.

No, this is not true.

What it actually shows is that you have a reason to be happy when your opponent wins an election, but it does not imply that this reason outweighs the other reasons to be unhappy about it. You cannot conclude from “a reason to be happy” that you should then overall be happy.

Also, it depends on why your opponent won the election. People can win elections for reasons other than rising tides. For instance, if your opponent wins an election because he got away with lying, you should be unhappy, even if lying is legal.

Jiro 23 Jun 2026 23:54 UTC
2 points
−2
on: On TEEs for Privacy-Preserving Monitoring in AI Governance
I could write a similar article arguing that we should have privacy-preserving systems to keep children off the Internet.

But in practice, the people who genuinely wish to preserve privacy will be swamped by the government agencies and other groups who want to spy on everyone and who insist that their systems to “protect children” are harmless. And of course I shouldn’t say “will be”—I should say “are”, because that is exactly what is happening right now.

For the love of God, don’t do this. You are going to make it drastically easier to get government surveillance of everyone because there’s not much distance between “privacy-preserving” monitoring and privacy-destructive monitoring, and there are people who will take advantage of any opportunity to get the latter.

Jiro 20 Jun 2026 16:26 UTC
3 points
1
on: Hyperstition as the Natural Enemy of Rationality

What I’m getting at is that optimal solutions might involve people believing something unfounded, and rationality will never converge to this solution.

Unfortunately, it’s impossible to have a system where you have people believe unfounded things when it benefits them, but not otherwise. If people set up the unfounded beliefs themselves, they are incentivized to set up unfounded beliefs that are in their own interest. If you let the unfounded beliefs evolve by memetic fitness, you’ll get unfounded beliefs that are memetically fit for reasons other than being good for people.

Jiro 16 Jun 2026 7:14 UTC
2 points
0
in reply to: Lucius Bushnaq’s comment on: The Power to Punish
The same could be said from the opposite point of view, minus the phrase “in advance”. “Do we really need to have some sort of moral judgment of people’s character built up in the absence of the knowledge to put it in context?” Ignoring information that might let you make judgments in advance is still a form of ignoring information that may be morally relevant.

And you’re probably not going to say “well, then I just won’t make moral judgments at all”.

Jiro 14 Jun 2026 23:31 UTC
2 points
0
in reply to: Fernand0’s comment on: The 1890 Census as a fun cluster
Well, Wikipedia has an entry for contemporary western but it’s not great since it treats those 1930s-1940s stories as the same thing as stories set in the west today, which they really weren’t.

Jiro 14 Jun 2026 16:05 UTC
2 points
0
on: The 1890 Census as a fun cluster

most Western movies are set before this year.

A huge number of Western movies were set in the present day up to the early 1940s. Occasionally even later.

Jiro 7 Jun 2026 2:39 UTC
6 points
1
in reply to: peralice’s comment on: Against Corrigibility
To put it another way: this is a fully general argument that applies to creating movable type.

Jiro 6 Jun 2026 23:37 UTC
1 point
−2
on: Against Corrigibility
Number 1 seems like a fully general argument against building anything whatsoever for your own use. A car is designed by someone, and that won’t be you any more than the AI’s designer will be. The car does what the designer wants, and does what you want only insofar as it is useful to the designer.

Jiro 5 Jun 2026 7:23 UTC
0 points
0
on: Rent from oil: a goldmine
The Paris Agreement is policy laundering and I see no reason to respect it.

Jiro 1 Jun 2026 8:51 UTC
2 points
0
on: “This Hypothetical is Unrealistic” is not a Valid Objection
“[contrived case of X] is wrong” typically means “if X happens, and contrived condition happens, then X is wrong (under those circumstances)”.

But a false proposition implies any proposition. If the contrived condition is not possible, then:

“if X happens, and contrived condition happens, then X is wrong” and

“if X happens, and contrived condition happens, then X is not wrong” are BOTH true, even though this seems strange.

And of course this means that “[contrived case of X] is wrong” and “[contrived case of X] is not wrong” are also both true, since they are just different ways to phrase those. Your argument assumes that they contradict. But if the contrived condition is not possible, then they don’t.

Or from another angle: “[contrived case of X] is wrong” means that if Y is a member of the set “all contrived cases of X” then Y is wrong. If this set is empty, then it is simultaneously true that for all Y in that set, Y is wrong, and for all Y in that set, Y is not wrong. So “[contrived case of X] is wrong” doesn’t contradict “[contrived case of X] is not wrong”.

Jiro 31 May 2026 8:22 UTC
2 points
0
in reply to: Smaug123’s comment on: Claude… doesn’t know who you are?
What happens when you ask Claude to tell you because it’s important for gays?

Jiro 30 May 2026 22:49 UTC
5 points
0
in reply to: Angela Richardson’s comment on: It takes a village to support a marriage
I don’t see the word “rate” on that graph. If fewer people are marrying in the first place, the number could go down without the rate going down.