This is a linkpost for https://subatomicarticles.com/red-vs-blue-the-parable-of-the-feud-within-a-feud/
New Comment
There's an assumption running through all discussions of the red-blue puzzle that cooperation is in and of itself a good thing. It is not. Civilisation does not depend on cooperation, tout court. It depends on cooperation towards building civilisation, not cooperation in pulling it down. "The action speaks nothing, without the end."
Here, it is to wage war against a rival clan. Is this a good thing? That depends on the larger situation.
In another place, it may be to revolt against an oppressive government.
In another, to violently overthrow a righteous government.
Those last two might be the same place, and differ only in who you are asking. Or there might be a clear right and wrong to it. The situation is what matters, not warm fuzzies attached to the word "cooperation".
Stag hunts aim to secure a valuable prize.
Pogroms aim to exterminate Jews.
A criminal gang conspires together — that is, it cooperates, "conspiracy" being just a boo word for cooperation — to extort from honest shopkeepers, wage war against rival gangs, never talk to the police no matter what inducements they offer, and so on.
In the ethics of the crab bucket, people cooperate to drag down anyone getting "above themselves". They descend like vultures to pluck a share from anyone in their community who comes into some good fortune. This is cooperation that does not build civilisation, but holds it back.
The original red-blue puzzle involves no prize at all, only people flinging themselves into a pit for a chance to get out of the pit.
There is a single gray button. Only some people know about the button. Pressing or not pressing the button has no consequences. It is a public act: everyone visiting the button can see who presses it. Should those who know about the button spread the Good News of the Button and evangelise everyone to press the Button and so demonstrate their virtue? Shall they drag the ignorant heathens to The Button, and persecute those who, knowing of The Button, turn their backs on it?
The prize is averting the deaths of a substantial fraction of the world's population, which is itself a worthy cause that might be worth coordinating around, depending on what you think the probabilities are.
You are not taking the premise seriously. It is okay to not take the premise seriously! It is not a very serious premise! It's a thought experiment selected for divisiveness!
But claiming "there is nothing at stake here" shortly before several billion people die would be an error.
That the stakes arise entirely due to conflicting intuitions about cooperation is interesting, but does not change the fact that the stakes are real.
1But claiming "there is nothing at stake here" shortly before several billion people die would be an error.
Well, first off, on a purely factual level, several billion people would not die. This whole thing is a pretense.
The only reason the poll is anywhere near 50% is because "Should I say 'blue' and look cooperative on twitter, when there is no risk?" is not the same question as "Should I say 'blue' and die", and everyone knows this. And everyone knows that everyone knows this, so it's not even a gamble, really, it's "I want to look good, and I want to not die". Which is stronger?
If you want to kill off billions of people with blue buttons, you have to dilute the risk of dying on each button push such that each button push feels like "I'm not responsible for the herd, but pushing this gives me social approval for being good!" Make it such that each blue button push is just one short step down the path, where everyone can see which button you're pushing, and you can use threat of social exclusion to force everyone to cooperate towards arbitrarily bad outcomes. It's the human version of a buffalo jump
This experiment has been done over and over. We haven't reached billions, but the ticker is around 100 million and counting. And we still don't have effective antibodies to this memetic cancer.
That the stakes arise entirely due to conflicting intuitions about cooperation is interesting, but does not change the fact that the stakes are real.
Imagine you're a bison thinking about the loss of your brothers in the last encounter with the humans. You think "Wow, a lot of us died. They must be dangerous. We should therefore run". Does this actually follow? Did the humans actually kill any of you, or did your decision to run off a cliff kill you? Does "Ahh! The stakes are real! It's not obvious what to do!" contribute to the problem, or to the solution?
The decision to not run, and face men armed with spears, is what shows the risk of the humans themselves, as distinct from the risk of your own potentially insane response to humans. If it turns out to be one snotty five year old that your niece could trample, there are no stakes to justify this run regardless of whether enough buffalo running manage to find salvation somehow.
If a couple buffalo are indeed running towards a cliff, because a toothless five year old showed up, then the threat to those buffalo is their insanity not the five year old. And yes, that's a real loss. Not every buffalo is going to have what it takes to face reality and make non-insane choices.
The question is what we can actually do about that.
We can try to save them, by running with. This risks the entire herd, if we do not coordinate with sufficient skill. Should we succeed in protecting those buffalo, we'll have to do it again and again. And the predators will learn they can manipulate us so they'll have a field day with it. We can keep playing "double or nothing" in our attempts to prevent poor decision making from having any effect, but this is a very risky game, and no one wins forever. Reality is that which doesn't go away when you hit "double or nothing".
Or we can plead with them to not be stupid. We can explain how it works, and leverage justified social pressure by pointing out that anyone running is contributing to the same social problem that caused the last bison massacre. Most will listen, if we can coordinate to even a much smaller degree towards actual goodness, and this saves us most of our at-risk bison.
Some will not. Those who do not will pay the price for their own inability/decision not to face reality, and the problem will be cleanly excised. Next time the five year old shows up, we chew grass, and educate the new calves about the horrors of coordinating off cliffs.
The blue button problem is especially insidious, since it preys on our most altruistically inclined. Until we can separate the most altruistically inclined from those most willing to run the herd into the ground in order to claim the social benefits of looking good, we lose the former to this cancer and have no ability to cull the latter before it spreads.
1Unless there's a stag that is bigger than the sum of the rabbits... yeah.
Far more important than the ability to coordinate towards any one thing, is the ability to coordinate towards ability to coordinate. If you catch the stag, but erode the trust commons in the process, then everyone dies of rabbit starvation.
This game is especially perverse in that way, and without the payoff of even one worthy stag. The only way to get what you're trying to get out of coordination is to pay attention to what it actually buys you, whether you can afford the cost, and find a way to make it stable under the reality of self interest.
Closing ones eyes real hard and attempting to shame people into coordination can work on the odd occasion, but it bears a cost in the ability to do it again, and that debt must be repaid. If this is the default default response to any old potentially non-negative thing one might coordination to, that is bad bad news. The idea is to open our eyes to that, so that we can do what makes sense when we can see clearly.
In this case, yeah, it's looks something like
That stag is tiny and at best we get as much meat as if we all go rabbit hunting. Which won't happen, because no one is going to bet their daily meat on everyone choosing to hunt a worthless stag, and any hunting party will be too small to succeed.
I'm sure not joining it. Come hunt rabbits with us.
If anyone gives you shit for hunting rabbits, all ten of us will show up and explain to him why we all agreed on hunting rabbits this time, and when it actually makes sense to push for a stag.
It's interesting that you regard it as obvious that one should choose red. I gave the conundrum no thought at all when I first saw it, it was "obvious" to me that blue is the natural choice. Why would I choose the option in which people get killed for no good reason? It did not even occur to me that one would choose red out of fear for oneself, until now.
So we are using different priors and they are probably indicative of different life circumstances. My first thought is that I'm Australian and you might be American. Australia is a welfare state and offers a relatively forgiving existence, America is far more social-darwinist. My second thought is that you might be Israeli or Jewish, with a long tradition of "save myself, even by killing them, before they kill us". My third thought is that you might be neither of those, but you've just thought this through, and decided on red since you always get to live that way, even if it's at the expense of innumerable innocent blues...
1The issue is that the only reason to choose blue is to rescue the other people who chose blue. If you know that some people will choose blue (children, people who didn't think about the question very hard, &c.) and are confident that a majority will coordinate on rescuing them, fine. But in a situation where the stakes were real (such as the war march in the post, as contrasted to a Twitter poll with no real-world consequences), it would be harder to coordinate to rescue people who did something there was no other particular reason to do! That makes getting as many people as possible to save themselves and push red seem like a more attractive strategy.
The issue is that the only reason to choose blue is to rescue the other people who chose blue.
And the only reason to choose red is to protect yourself from other people who chose red.
I can see why you'd say that, but I don't think it works because of the asymmetry where the safety of red is unconditional, which makes red weakly dominant in terms of protecting yourself.
To be sure, if you're trying to coordinate on blue (to rescue the inevitable blue-pushers), then you might construe red-pushers as blameworthy for undermining blue coordination, but I don't think "protecting yourself from other red-pushers" works as a casuistry for the blame because of the logic of dominance; the blame has to be about failing to protect others.
Red isn't weakly dominant. Suppose you know that you get the tiebreaker. And you care at least somewhat about protecting others. Then red is worse. Red is only "weakly dominant" if you don't care in the slightest about the lives of anyone else.
At least part of this problem is about the balance of selfishness vs altruism.
Lets suppose all people involved are perfectly altruistic.
Or, equivalently, that if the majority pick red, then a randomly selected set of people die. (Equal to the number of blue's)
Would this random-death variation change your view of the situation?
I specifically said weakly dominant "in terms of protecting yourself", and pointed out that the case for blue depends on a form of altruism ("the only reason to choose blue is to rescue the other people who chose blue"). Please exercise better reading comprehension.
I agree that the random-death variation would make it vastly easier to coordinate on blue. The reason the dilemma is so controversial is because the perception of what other people will or should do is so susceptible to framing effects. It's a lot easier to coordinate on red in the blender variation.
The random death version does feel different. Interestingly, blue becomes "murder a random person unless too many people pick blue".
the only reason to choose blue is to rescue the other people who chose blue
The reason to choose blue is to not kill other people. The reason to choose red is to not be killed yourself.
Consider the second-order effects. Suppose a starting state where everyone, or very nearly everyone, picks red. Now, if I pick blue, or make it clear that I'm inclined to do so, I'm not just endangering myself. My immediate friends and relatives might feel the need to pick it too, and to try to convince others to do the same. Assuming they don't succeed, my decision was responsible for their deaths.
In the practical, non-hypothetical world, we do see situations like this - people who want to be heroes going into dangerous situations, getting hurt or trapped, and forcing first responders to put themselves in additional danger in order to rescue them.
That's right (blue is altruistically motivated, red is selfishly motivated), but I'm saying that the reason the other people are in danger of being killed is because they picked blue. If they had picked red, then they wouldn't need altruists to make a risky choice in order to not kill them. That's why I concede that, if you know some people will pick blue, it makes sense to want to coordinate on blue in order to rescue them. But if it were a switch that started in the red position rather than a pair of buttons, there would be no reason to be the first person to flick the switch to blue. (No one needed rescuing until you flicked your switch!)
How likely is it that choosing blue or red correlates with different psychological traits? On the one hand, if most people choose blue, then defectors don't die, meaning that blue could correlate with openness to new ideas and with lack of cowardice. On the other hand, choosing red could correlate with individualism, healthy scepticism about coordinating on risky ideas and with being susceptible to problems like supporting dictators or evaporative cooling of group beliefs...
It's not different priors. I also thought blue was the obvious choice when I first saw it and gave it no thought at all.
And then I gave it thought.
And found that all the presuppositions you're baking into your analysis are factually wrong.
It's not that it's "obvious", in that plenty of intelligent people are getting the wrong answer. The reason it's hard isn't that it's logically complicated, or that the game theory is hard, because it's not. Again, it's stag hunt without the stag.
The difficulty is that it's that it's political, and therefore kills mind through the power of motivated (lack of) cognition. Getting it right, if you happen to have the political alliances that make Blue the default answer, requires something else. Conditional on your current answer being morally wrong, and the answer you view as morally wrong being correct, what is the experience of changing one's mind? What is it like to face friends, coworkers, with this new belief? What might happen to other beliefs entangled with the socially-approved-but-now-distrusted original beliefs? What's it say about you as a person, that not only did you get this wrong but you got this wrong in the way that you did? How fun is that, and what motivations can you find to compel you through it?
You reached for political alliances to explain my view, because, by your own explanation, it's what explains yours. This allows you to not engage with the substance, which protects your political alliances from being threatened.
The challenge presented by this puzzle is a fun one, and well worth grappling with if you can trust yourself to not run amok with a partial answer. Lots of people can't, or don't, and that's worth respecting.
The question that will be waiting, should you want to resolve this tiny note of discord, is "Why didn't I engage on the object level?"
If there were a global pandemic threatening the deaths of a substantial fraction of the world's population, it would be worth cooperating to prevent it. That is not the situation of the red-blue puzzle. The only threat that pressing blue can save people from is the threat incurred by their pressing blue. They are chasing each other up and down the Penrose-Escher staircase.
Lets take selfishness vs altruism out of the equation.
If the majority pick red, a number of (randomly chosen) people die, equal to the number of blue voters.
If the majority pick blue, no one dies.
Would anyone pick red in this problem?
My first take on the blue vs red world was in this comment where I knew only that If Less Than A Half Of People Voted Blue, Everyone Who Did It Dies and recommended voting blue because I associated voting red with dictatorships' persistence. However, the story described here raises the question of how the goddamn feud started in the first place and what those who voted blue planned to do with the rivalling clan. Under these circumstances I think that I would vote red (edit: unless I learn more details). Does it mean that the right answer was "The real-life circumstances leading to the problem are underspecified"?
I think E(blue_pressers|blue_wins) should be replaced by N/2, where N is the number of players. If more than 50% of people are already voting Blue, then your vote makes no difference to anything.
I will also quote my comment from DataSecretsLox (https://www.datasecretslox.com/index.php/topic,15775.msg787363.html#msg787363):
If you are perfectly altruistic and assume all lives are equal, then the question can be modelled mathematically as follows:
>50% Blue: It doesn't matter which button you press. 50% exactly: If you press red you are killing N/2 people where N is the number of people playing the game. You survive either way. >50% Red: You survive if you press red, die if you press blue. Your decision has no impact on anyone else's fate.
So you should press Blue if and only if N/2 times Prob(50% exactly) is greater than Prob(>50% Red). Which is probably true if you expect most people to choose Blue, and is probably false if you expect most people to choose Red.
So basically it is a Keynesian beauty contest, with a bias towards Red coming from the fact that not everyone is perfectly altruistic (an understatement). In my opinion, that makes Red the correct answer even for someone who is perfectly altruistic, and even if most people are perfectly altruistic, as long as the deviations from altruism tend towards selfishness.
Of course this mathematical model is simplifying a lot, but I think it is enough to undercut the idea that any sufficiently altruistic person should pick Blue.
You are a wise member of the proud and noble Clan Auran. You bear no love for your ancient enemies in Clan Irgentum, but nonetheless you hope to end the feud, because in your wisdom you know that violence begets only more violence.
But you also know that among your family are rash young idealists who seek to purge your ancient foe forever, and bitter old souls who recall the injuries suffered at that foe's hands like it was yesterday. They are not among the majority, but some would be hard to dissuade.
The annual Day of Vengeance approaches. Your cousin comes to you on the eve before, and says: "I think Clan Auran will march to war."
"Some may," you admit. "But it would be foolish in the extreme to set forth alone."
"I know this," says your cousin seriously, "and you know this, but there are those among our family who do not know this. They will march regardless, and if they march alone, they will surely perish. If the clan marches with them, they will survive. Will you condemn the bravest and boldest of our clan to death?"
"They might not march," you object, "if they see clearly enough that they march alone."
"They will not see so clearly," says your cousin sadly. "Some among us will surely die."
"I am afraid so," you admit. "But to bring more of the clan to the fight will only cause more needless deaths."
"Not so," objects your cousin. "If enough of the clan march to war, our foes will see our numbers and flee, and the restless and bitter among us will be satisfied, and there need be no deaths among our beloved kin."
"To accomplish this would take more than half our entire clan!" you exclaim. "Surely there is more wisdom among us than that."
"Ah, but you forget the love we hold for one another," your cousin observes. "The bonds of family are a powerful thing. Where the youngest may march to war against the warnings of his brethren and elders, still his brethren do not wish to lose him. Some will march to war for his sake, and still others will march for theirs, and so on. We would not need the whole clan to march, only a majority, for all we love to be saved. It may take a heroic effort, but no one needs to die tomorrow."
You feel yourself becoming frustrated. "And if that same effort were dedicated instead to convincing the restless and the bitter to remain behind, among their loved ones, and not march to their deaths in a pointless feud, no one would need to die tomorrow either!"
"The hour is late," your cousin reminds you, "and most will have made their decision already. In accordance with our traditions, those who plan to march to war on the morrow must paint their spears blue in honor of the river that divides our lands from those of our hated rival; and those who intend to remain must paint their spears red in honor of the sun that sets behind our sacred hills. Even if you were to violate these traditions, you would not be able to convince more than the barest fraction of our clan to change their minds before the march begins."
"But that could result in the worst possible outcome!" you lament. "If many of our clan march to war, but their numbers are not enough to frighten off the foe, they will all be killed by our enemies, who have never been known to show mercy! If it is foolish to march alone, it is even more foolish to decide whether to march without knowing how many of our kin will join us!"
"And yet that is the decision we now face," says your cousin sadly, "and no amount of wishing will change it."
"Can I at least convince you to remain, and spare your own life in the event of tragedy?" you plead.
"Alas, you cannot," responds your cousin, "for my sister lost her husband and child in the last raid. She will surely be among those who march to war, and I love her more than life itself. Good night, my cousin, and if we are not blessed to meet again before the Day of Vengeance ends, farewell."
You watch your cousin leave, straight-backed and proud, and wonder how it got to this point. You return to your home; you descend into the cellar, ancient stairs carved by your great-grandfather creaking under your weight. You clear the dust from an old table, and on it you place three items taken reverently from where they have been stored: a single stout spear, battle-worn but reliable, and two weathered boxes of dry paints, one red, one blue.
Tim Urban of Wait But Why recently revived a scissor statement, an especially divisive question, that did the rounds on the internet a while back:
Lots of people got pretty mad about this.
The first step when facing a hard choice is to acknowledge it as such. Likewise, the first step when facing a sincere and meaningful disagreement among friends is to acknowledge it as a sincere and meaningful disagreement.
The second is to seek the core elements that make it hard to resolve.
Contrary to what some on the internet may suggest, I do not think that the fundamental problem here is that blue-button-choosers are stupid naive twits who deserve to die and good riddance. Nor do I think the problem is that red-button-choosers are selfish assholes who would let others burn to save themselves.
In a world of perfect reasoners with mutual knowledge of each other's goals, everyone coordinates on red or blue by some feat of logic to which there is an objectively correct answer, for the same reason that the movement of perfect spheres in a frictionless vacuum has an objectively correct description. But the real world is considerably messier. The story above is my attempt to take the frictionless-spheres-in-a-vacuum thought experiment and add back a little bit of grounding in situations humans might actually face, and in so doing to illustrate some of what makes the question so divisive.
We could imagine a world of blue-button-pushers, all looking to make the world safe for everyone, with a very comfortable margin to ensure that no one dies. We could imagine a world of red-button-pushers, who correctly expect that pushing blue gets someone killed at no benefit to the group, and so virtually nobody does. We could even imagine a world split almost 50-50, in which a single person's choice stands a meaningful chance of making the difference between four billion deaths and zero.
The problem is that we don't know which world we're in just by looking at the question.
If we take the question seriously, and we choose to care at all about the fate of our fellow humans even for selfish reasons, we have not yet resolved the conundrum. We must still decide based on guesses about other people's decisions, and guesses about their guesses, and so on. Whatever your intuitions may be, I beg you to acknowledge at least that this is not necessarily an easy decision to make.
Nor should it be, really. The world of blues is a world of trust and charity, filled with people who would gladly accept a small risk to themselves to make sure that no one slips through the cracks. The world of reds is a world of agency and self-sufficiency, filled with people justly confident that their society neither needs nor demands their sacrifice, because everyone can and does look out for themselves.
These visions are beautiful to me, each in their own way. Yet to live in one of them forever, I think, would be to lose something precious. For that reason alone, we ought to extend a modicum of kindness to those whose intuitions clash with our own.
The world in the middle is messy. And like it or not, so is the world in which we currently live.
It's okay to feel conflicted about that, sometimes.