This is a special post for quick takes by Yoav Ravid. Only they can create top-level comments. Comments here also appear on the Quick Takes page and All Posts page.

New to LessWrong?

61 comments, sorted by Click to highlight new comments since: Today at 2:37 PM

If someone commits an immoral act, and then gets cloned to the atom, are both versions of him guilty? (morally, not legally)

If so, then - assuming what he did deserves a 10 year sentence, should both versions get it, or should each get half a sentence?

Hm, sounds like a setting for a sci-fi movie. First you get human life extension: humans can live 200 or 300 years. Then you get some horrible criminal who got 300 years prison sentence. But the criminal cloned himself before being caught. The judge decides both versions should get 300 years. The owner of the private prison thinks "nice, free labor!" and makes ten thousand clones of the criminal -- because the criminal is extra smart and/or extra strong, so he is more profitable than an average prisoner. Turns out the clones cooperate with each other, and together they escape from the prison... and then the usual Hollywood story starts, with explosions and stuff.

Later they find out that the criminal actually made one more clone of himself before he became a criminal. So the clone is a good guy. (Or maybe unauthorized cloning of yourself is already a crime, so this old clone is hiding from the law, but he is not a violent criminal, unlike all the other clones.) Now the police tries to recruit him, maybe promise an amnesty to him if he helps to eliminate the other clones. Between the action scenes, there are flashbacks to the past, explaining why one of the two original clones became a violent criminal and the other did not. Possibly a love story that ended tragically. (Were both in love with the same girl, and did they fight for her? Or maybe, both were in love with the same girl, then she died in a traffic accident, one of them decided to forgive the guy who accidentally killed her, the other one murdered him.)

Or perhaps the old clone helps to kill all the fugitives and get the amnesty, and then it is revealed he was actually the worst of all... only his crimes were never exposed, because he managed to frame the other clone.. so he was the bad guy, who just murdered ten thousand copies of the good guy...

At the end, the bad guy kidnaps the corrupted judge, and tells him he just made ten thousand copies of him. One randomly selected copy will be released, the remaining 9999 will be tortured to death, and all evidence will be destroyed, so no one will believe that any crime happened. The bad guy is just bluffing here, there was only one copy of the judge, and he is finally released, but he is traumatized because he will never know whether the threat was real or not.

(I'm going to just think aloud here about your idea).

Reason why this guy is so dangerous - He was genetically engineered. both his mental and physical abilities are wicked good. but, well, he was created during the first few years/decades of genetically engineered humans, and there were... some unforeseen consequences.

Or maybe he experimented with unauthorized cloning a lot, and self-experimented on the clones with all kinds of enhancements. some going better some going worse.

Possibly a love story that ended tragically. (Were both in love with the same girl, and did they fight for her? 


Perhaps somehow the mind cloning was good, but the body cloning wasn't (or the body was hurt afterward), and now he's ugly and the girl didn't want him. the scenario where she dies in an accident and one instance forgives the guy and the other murders him suggests quite different minds - so, not really the sort of clone we're talking about. 

There are many possible ways how to do this. I imagined something like "bad moral luck" -- two minds that started the same, but different things happen to them, and they diverge as a result.

For example, both decide to murder the guy, and both wait for him at different places. One murders him. The other waits at a wrong place (maybe just further down the same street), so nothing happens... then he goes home, gets some good sleep, and realizes that he was stupid. Alternatively, one gets a phone call that reminds him of something, and he changes his mind.

The idea of mind divergence was already done with parallel words (e.g. Sliding Doors), except these would be clones living in the same world.

Ouu that's also nice!

This. Is. Awesome. 

is someone writing this? are you writing this? should i try to write this? i really want this written. did i mention already that this is awesome and should be written?

Thanks, but I am not good at writing. Anyone, feel free to use this idea.

The clone has an alibi: not existing at the time.

Sure.  Moral culpability for past acts should go with the decision-making apparatus, the pattern of connections and data in the brain.  If that's copied, so is the blame.  We can't let these jerks get away with stuff just by using a Star Trek teleporter!

Yes, thinking about it more the only policy i found that didn't lead to problems was (in the case where the cloning happens after the act):

  • All instances of the person should be regarded as culpable as much as you would regard the person if they didn't clone themselves.

Otherwise if not both of them are culpable you get the star trek teleporter problem.

And if you divide culpability it means someone can reduce their own punishment 50 fold by creating 50 clones. and if they aren't sympathetic to the clone's suffering, they might do just that.

The sad thing about this is policy is having to multiply the amount of suffering experienced by the punishment. you would much rather someone not create a clone after committing a crime, cause then you'll have to punish multiple people. maybe in such a future cloning would be a crime for someone who previously committed a crime.

The sad thing about this is policy is having to multiply the amount of suffering experienced by the punishment.

There's a missing step in this result.  Moral culpability is about judgement and condemnation of actions (and the actors who performed them), not (necessarily) about punishment.  Calculation of optimal punishment is about influencing FUTURE actions, not about judging past actions.  It's not fully disconnected from past culpability, but it's not at all the same thing.

You may have to increase total suffering, but you may not - perhaps punishing one clone randomly is sufficient to achieve the punishment goals (deterring future bad actions by that decision-maker and by observers).  Even if there's more summed punishment needed to have the same level of deterrence, presumably the clones increased total joy as well, and the net moments of lives-worth-living is somewhat increased.

Now if the cloning ITSELF is a moral wrong (say, it uses resources in a way that causes unjustified harm to others), you pretty much have to overreact - make it far more painful to all the clones, and more painful for more clones.  But I'd argue that the culpability for the punishment pain falls on the clones as well, rather than the judge or hangman.

How can you be culpable of an act that you didn't commit? The !as doesn't punish people just for being the kind of person who would commit an evil act.

Well, for one, they wouldn't just be the kind of person who would commit an evil act, they're the kind of person who did commit an evil act.

But ok, how do you suggest solving the two evasion tactics described?

Well, for one, they wouldn’t just be the kind of person who would commit an evil act, they’re the kind of person who did commit an evil act.

No, they are a clone of a person who did commit an evil act. You can't claim that they are very same person in the sense of numerical identity. (Numerical identity is the kind of identity that does not hold between identical twins. Identical twins are not the same person. They are two identical people).

I'm not talking about identical twins, I'm talking exactly about numerical identity. perfect cloning.

Cloning isnt numerical identity. A clone is an artificial twin and twins are not numerically identical. Numerical identity relates to aliasing, to having two labels for the same entity. Stephanie Germanicus and Lady Gaga are numerically identical.

Ok, so i confused the term with something else, oops. my point is that I'm talking about an exact copy, in the sense discussed in the quantum mechanics and personal identity sequence.

The exactness of the copy doesn't matter. If a twin commits a crime, the other twin is not held responsible , because they did not commit the crime, not because there is some minute difference between them.

Suppose you got 'cloned to the atom'. 10 years later, one of you commits a crime. Who is punished? Who should be?

After 10 years I'd expect plenty of difference to have built up, making it not so different (judicially) from a regular case. One thing you might still say is that it's reasonable to hold the other instance as prone to commit murder too, and perhaps that would justify surveillance on him, or something..

On first thought, it does not seem to me that (im)morality is something that is commonly ascribed to atoms. Just as bits do not actually have a color, so it seems to me that atoms do not have morality. But I'm not a moral philosopher, so that's just my feeling.

On second thought, consider a thought experiment where we judge the clone. Was the clone a direct / proximate cause of the immorality? It would seem not, as the original was. Did the clone have the intention to cause the immorality? It would seem not, the original did. So I don't think I would hold the clone liable for the committed immorality.

A more interesting scenario to me would be - We have two clones, we know one of them committed an immorality, but we do not know which one. How do we proceed?

The morality isn't ascribed to atoms, it's ascribed to the person in the same way it usually is. yes, people are made of atoms, but it all adds up to normality.

On your second point, did you read the article linked? summarized, the conclusion is that in the case of perfect cloning "There is no copy; there are two originals." (on reflection i might have linked the wrong post, this is where this quote is taken from).

from this viewpoint, there would be no difference between blaming the "clone" and blaming the "original". so in a way it's isomorphic to the scenario you suggested in the third paragraph. 

It's probably important though whether the cloning happened before or after the act. if someone cloned himself, and 40 years later one of them commits a crime, there probably isn't such a dilemma. but is the same true if a crime is committed by one of the clones right after cloning? not sure.

It seems to me that you are thinking about some "stronger" form of cloning. The framework that I was thinking in was that the "clone" was a similar-but-distinct entity, something like a Twin materialized out of thin air instantaneously. But it seems that you are thinking of a stronger form where we should treat the two entities as exactly the same.

I have difficulties conceptualizing this since in my mind a clone still occupies a distinct time, space and consciousness as the original, and so is treated distinctly in my eyes. (In terms of being judged for the morality of actions that the original committed).

I will try to think of a situation / framework where this "stronger" form of cloning makes sense to me.

Let's see if i can help.

Say someone commits a crime, then goes into a scanner, destroyed, and recreated somewhere else. is it agreed that they're the same person? if so, it would make sense to still blame them for the crime.

Now let's say we discovered that this person never actually destroyed themselves, they were scanned and cloned, but faked getting destroyed.

Should the "clone" now be declared innocent, and the "original" declared guilty instead? or should both of them be declared guilty?

Yeah, that makes sense. The way I came to think of it is that person A commits a crime, then faints and is unconscious after that. Afterwards, a separate nefarious cloner then clones person A in a black box, so one person A goes in, two persons A come out from the cloning black box.  Person(s!) A awake, and having a strong conscience of their crime, turn themselves in. Since they have exactly the same memories and conscience, they are indistinguishable from the point of view of being the person who committed the crime, both internally and externally.

This is actually a good question. I feel that both persons should be declared guilty, since cloning oneself (whether intentionally or not) should not give one an automatic-out from moral judgement. I am not as sure about whether the punishment should be equal or shared.

See my thoughts here on full/distributed punishment

'The grass is green' and 'The sky is blue' are pretty bad examples of obviously true statements (they're actually often false - the sky is dark at night and the grass is yellow-brown when dry).

'The sun is bright' and 'Water is wet' are better statements in the same style.

Honestly, I prefer the more obvious generalities.  They're false often enough that one is able to see that it's not a literal universal truth, but more an extremely common experience being used as an example of something uncontroversial, even with well-known non-explicitly-mentioned exceptions.

All 4 statements are incomplete - these aren't properties of the object, but relational experiences of a person to the object.  None of them are universally true for all situations (though the latter two are closer).  

Yeah I agree with that. I like the latter two statements because they seem just as general, simple and everyday to me, while also being more correct.

I've seen some convincing arguments that water is not wet.

I don't like the dialogue matching feature. It's not like dating where my interest in a dialogue comes from interest in a specific person. It's more akin to wanting to do an activity (go climb / discuss a certain topic), and looking for someone to do it with (could be anyone).

If you want to have a dialogue with me, send me a message, and I'll probably say yes.

If I were to design a dialogue matching system I think I'd make it more topic-based than people-based.

Idea: Github + Voting 

The way Github works is geared towards how projects usually work - someone or some group owns them, or is responsible for them, and they have the ability to make decisions about it, then there are various level of permissions. If you can make a pull request, someone from the project's team has to accept it. 

This works well for almost all projects, so I have no problem with it. What I'm going to suggest is more niche and solves a different problem. What if there's a project that you want to be completely public, that no one person or group would have authority on it? 

My idea is a repository site like Github, but instead of anyone having authority, all changes are voted on. So if I make a pull request to make some function faster, it opens a vote. If there's a big ratio of people against the change, the change doesn't happen, if there's some minimum for it and not enough resistance, it goes through. 

What exactly are the thresholds and rules for voting? That would also be changeable through vote, and the person who first creates a repository will just need to choose the default settings.

Other devops features can also be integrated, so things like pushing updates and managing hosting can also be done through voting.

This is not for most projects, this is for decentralized projects that don't want anyone having authority over them.

(If I want to play around with it, what repository sites are there that are open source like GitLab?)

There are Blockchains like Polkadot that require people to vote whether code changes get deployed. I'm not sure how that exactly works but looking at those crypto project might be valuable does for looking at existing uses and also to find potential users.

Oh, cool! I'll look into it. Thanks :)

I like this idea, because I'm too lazy to review pull requests. It would be great if other people could just review and vote on them for me :P

(Only read the abstract)

Stagnation and Scientific Incentives

New ideas no longer fuel economic growth the way they once did. A popular explanation for stagnation is that good ideas are harder to find, rendering slowdown inevitable. We present a simple model of the lifecycle of scientific ideas that points to changes in scientist incentives as the cause of scientific stagnation. Over the last five decades, citations have become the dominant way to evaluate scientific contributions and scientists. This emphasis on citations in the measurement of scientific productivity shifted scientist rewards and behavior on the margin toward incremental science and away from exploratory projects that are more likely to fail, but which are the fuel for future breakthroughs. As attention given to new ideas decreased, science stagnated. We also explore ways to broaden how scientific productivity is measured and rewarded, involving both academic search engines such as Google Scholar measuring which contributions explore newer ideas and university administrators and funding agencies utilizing these new metrics in research evaluation. We demonstrate empirically that measures of novelty are correlated with but distinct from measures of scientific impact, which suggests that if also novelty metrics were utilized in scientist evaluation, scientists might pursue more innovative, riskier, projects.

New ideas no longer fuel economic growth the way they once did. A popular explanation for stagnation is that good ideas are harder to find, rendering slowdown inevitable.

Another place to go from there is that ideas aren't proving as useful as they used to. As if they're being monopolized somehow. i.e. one could investigate/compare 'productivity' and the rules around 'intellectual property' such as 'copyright'.

Every now and then I see a paper that looks into replication in some field and finds a replication crisis, to the point that just by seeing in the the title that the article is about replication in field X I can know field X has a replication crisis. I wonder, is there any paper that looked for a replication crisis in some field and didn't find it? 

(Not sure what effect of publication bias would have. On the one hand it's a "negative" result, so less likely to be published, on the other hand it says very good things about the field, so people in the field would want to publish it)

(This is an exercise, be careful not to spoil the answer to yourself)

All world maps are wrong due to the fact that it's impossible to flatten a sphere without distortions.

there is a simple idea anyone can think of that greatly improves the accuracy of flat maps and that no has tried in the last 2000 years - Until last week, when three Princeton researchers thought about it.

Take a moment to try to think what you might do to improve the accuracy of flat maps.

I'm making this an exercise since this seems like incredibly low hanging fruit that hasn't been picked up, and the idea will seem obvious in retrospect.

Ok, stop here and think, spoilers ahead:

*
*
*
*

Make a double sided map, of course!

Map design by J. Richard Gott, Robert Vanderbei and David Goldberg

Instead of projecting a sphere to a flat surface, they just projected two hemispheres to two surfaces and glued them together.

"Goldberg and Gott invented a system to score existing maps, quantifying the six types of distortions that flat maps can introduce: local shapes, areas, distances, flexion (bending), skewness (lopsidedness) and boundary cuts (continuity gaps). The lower the score, the better: a globe would have a score of 0.0."

The previous best on this metric was the Winkel Tripel projection, with a Goldberg-Gott score of 4.563

the Winkel Tripel projection

Their new design is better than the Winkel Tripel on every one of the 6 matrices, with a slightly lower Goldberg-Gott score of 4.497.

The other huge advantage of their design is that it's the only flat design that has the topology of a sphere. if you go over the edge it's exactly like going over the equator.

The other advantage is a bit less concrete - Their design just looks fun. it makes me want to hold it in my hands. other designs don't do that for me.

They also made maps of other solar system bodies, available in their paper (Starting at page 24). Here's mars:

 

Now I'm just left with two questions:

  1. How did no one think about this before
  2. Why aren't they selling maps

See the full article about this here

This is a breathless article about something that's obvious to people who know the state of the art. (I've worked in geoinformational systems.) If you want a map that shows the shapes and sizes of continents without too much distortion, and don't mind having two circles, the Nicolosi globular projection is a thousand years old.

Huh. Well, thanks for pointing that out. I see that it is part of the collection they mentioned, but they didn't count it as a two-sided map. Is it suitable for two-sided map? Was it used that way? 

In any case, oops.

I don't know, something about this smacks of "prestigious people reinvent obvious thing that was previously dismissed out of hand because it didn't meet the criteria but if you're prestigious enough people might bend the criteria for you".

In particular I think the point was largely around having two dimensional projections. Using both sides is, in some sense, not really a 2D projection anymore since you have to interact with it by rotating it. And if you have to do that you're most of the way to just using a globe instead.

The idea is that it lets you compact a globe from 3D space to 2D space with minimal distortions. You can carry a 100 such maps in less space than one globe would take. (Of course, if your requirements are to be able to see everything at once, then this doesn't fit)

So what you said in the first paragraph doesn't seem true to me, but if someone did invent that already and was dismissed i would be interested to hear.

My guess was: you could have a different map, for different parts of the globe, ie a part that focus on Africa (and therefore has minimal distortions of Africa), and a separate part for America, and a separate part for Asia, and so on.

Well, in a way that's what they did. They have two maps for each hemisphere which connect perfectly when glued together. But the idea of having different maps for different places on it's own has been done countless times.

The biggest disadvantage of this that I could see is that it prevents you from seeing the entirety of the map at once. This is reflected in the article linked, ""Our map is actually more like the globe than other flat maps," Gott said. "To see all of the globe, you have to rotate it; to see all of our new map, you simply have to flip it over."".

Right, it's not the sort of map you'd want to put on a wall, it's intended to be interactive and give the benefits of a globe in flat space.

What is the class which ask/guess/tell/reveal cultures are instances of? it doesn't currently have a name (at least not something less general than communication culture), which makes this awkward to talk about or reference. so i thought about it for a bit, and came up with Expectation Culture

Ask/guess/tell/reveal culture are a type of expectation culture. they're all cultures where one thing that is said maps to a different expectation. this is also the case with different kinds of asks

This seems like a useful phrase with which to bundle these things together.

When making many predictions together, I think it would be useful to add another prediction about your calibration.

I think it's worth knowing not just how calibrated someone is, but also how "meta-calibrated" - do they have a correct sense of their own level of calibration?

Punishing non-punishers taken to the limit (punishing non-punisher non-punishers, punishing non-punisher non-punisher non-punishers, etc...) is simply "If you're not with me, then you're my enemy".

I think if taken to the limit that it becomes "nobody is with me, not even me". After all, there are billions of people who you are not punishing and almost certainly will never punish for being non-punishers of non-punishers.

Correlation suggests Association suggests Causation.

Looking at the sky suggests the Sun goes round the Earth.

Indeed it does, and that the earth goes around it. Until you have the information needed to rule either hypothesis out you better consider both. Maybe In some universes the bright thing in the sky really does orbit around you and not vice versa.

Which causation?

Several. You need more information than just one correlation/association to pinpoint the exact causal relationship.

But I expect you already know that, cause my understanding is that this is standard correlation/causation stuff - the sentence wasn't supposed to point at a novel idea, just a novel phrasing of it. So your question seems rhetorical, but I don't know what's it trying to point at. Do you think this view of correlation/association/causation is false?

Me and a friend want to start journaling daily, and we're looking for a good app to use for that.

Desired features:

  • Can be used from a phone (bonus if it's multiplatform)
  • Easy to add a new entry
  • Easy to see entries in chronological order
  • Has an option to tag entries so it's easy to later come back and read them using the tag
  • Can remind of you entries (say, remind of you entries on the date they were entered, or after some set amount of time).
  • Bonus: Reminds you daily to write with a notification

I thought of roam and obsidian, but I'm not very familiar with either. So I'd be happy to hear both thoughts on these two apps specifically, and recommendations for other apps.

If you are in the Apple ecosystem, I can recommend Day One. My oldest entry is nine years ago and I've been very happy with them.

I'm an android user. Not a fan of apple/IOS. But thanks :)