[The ideas here followed from some great discussion with Jerome Warren in Brussels 2022]

Karl Popper’s Paradox of Tolerance argues that an absolutely tolerant society is impossible, since it would be vulnerable to intolerant people coming in and breaking it (or just killing everyone). To me this sounds a bit like an unstable equilibrium from dynamical systems theory – trying to balance a pencil on its tip. The paradox is thus that to have a stable tolerant society, it must be intolerant of intolerance – and that immediately creates the gray-zone of what exactly we should be intolerant of and how violent we should be about it.

I’ve been annoyed with this argument ever since I learned about it. My idealist self really wants an absolutely tolerant utopia to be possible, at least in principle, without compromises. As such, I’ve been trying to get a deeper understanding of the argument and its assumptions so as to find some way to dislodge it. Part of my discomfort with it comes from thinking of the stories of non-violent protests, such as exemplified by Jesus, Gandhi, MLK, and others. Now before you point out all the ways in which those people were actually quite intolerant of intolerance, I just want to suspend disbelief and entertain the possibility that the dominant force in their reformations was actually the tolerance itself. To balance a pencil on its tip, we need some forces to hold it up – and very small forces may actually be enough (in fact, we have found many fun ways to balance pencils on their tips - like by using oscillating drives or active control) So if we can understand a way in which tolerance itself can be a force for change – an active influence, rather than just passive tranquility - then we may have our solution.

One implicit assumption that the Paradox of Tolerance seems to rest on is that to turn an intolerant person, we must actively and intentionally do something to them. This may involve using force and violence, propaganda, manipulation, seduction, or incentives – but one way or another, we would try to change them, and thus be intolerant towards them.

But let’s now for a second suppose that tolerance is contagious, even just a little bit. By contagious here I mean that merely being tolerant would have some tendency to make others more tolerant – with no effort or even intention to the matter. We could think of this in the context of memetics – where not people, but ideas ("memes") are seen as the individual beings that spread and multiply using the medium of our brains. Like in epidemiology, the most contagious and resilient ideas will then dominate the society in the long-run. Note that this view is in sharp contrast to one that sees society as a collection of individual humans with fixed beliefs. This latter view, which is often implicitly assumed in much of western thinking, would indeed lead to the Paradox of Tolerance – where the only way to get rid of a particular belief or behavior (such as intolerance) is to kill or coerce the people carrying it.

On the other hand, from the memetic perspective, the possibility of an absolutely tolerant society is just a matter of how contagious tolerant behavior is. One curious thing that seems to be true historically is that martyrdom makes it more contagious! If so, then killing tolerant non-resisting people actually fuels the “tolerance epidemic” – thus increasing the push towards tolerance. This could explain how non-violent protests were able to become such strong social forces. It also seems to lead to a conclusion that is just the reverse of the Paradox of Tolerance: that in an absolutely tolerant society, rise of intolerance would create martyrs, making tolerant behavior more contagious, and thus more resilient.

What makes this perspective non-obvious is that the effect of such “contagion” will only be prevalent at scale, and may not be obvious in the immediate local and short-term consequences. A psychopath (i.e., one “immune to the tolerance virus”) with a machine gun in an absolutely tolerant society can certainly do much damage. Similarly, millions of martyred Jews in WW2 did not seem to stop the intolerance towards them – only force did. Nonetheless, now, nearly a century later, the lesson from those WW2 martyrs still impacts legislation and decisions of governments around the globe in very concrete tangible ways (e.g., link).

This brings us to another interesting point: epidemics require not only a high contagion rate, but also tight social networks to spread over. This way, Gandhi’s non-violent protests may never have succeeded if it weren’t for the rising globalization, and namely the global media reporting on India’s protests back in England. If so, then for an absolutely tolerant society to be stable, it may require a certain minimal degree of interconnectivity of social bonds (and especially empathic ones). I.e., a loosely connected social network, such as sparse early-human tribes, may indeed be incapable of total tolerance – unlike a globalized economy and culture. This line of reasoning further makes me think that there may be other subtle pre-requisites for absolute tolerance that we are currently ignoring, and that will only become clear with deeper research and modeling of sociology. Perhaps it is just such missing requirements that make an absolutely tolerant society seem far-fetched to our intuition now.

To me, while these ideas certainly do not prove the stability or possibility of an absolutely tolerant society in practice, they do open up a slight gap in Popper’s impossibility proof, making it less water-tight. From this point on, I think there is a wide research opportunity to study the stability of absolute tolerance in the context of various social network connectivities, including external threats, neurodivergences that could give an “immunity to tolerance,” temporal dynamics, including phase transitions from intolerance to tolerance and vice versa, etc. And of course, all of this is hard because we don’t quite know how to do quantitative sociology yet – but fun to think about!

So, could tolerance be contagious, without any intentional action to make it so (violent or otherwise)? If so, could that make the existence of an absolutely tolerant society conceivable? Share in comments! 

(Cross-posted from my blog at pchvykov.com/blog

New to LessWrong?

New Comment
8 comments, sorted by Click to highlight new comments since: Today at 9:52 AM

I once saw someone argue (haven't found the citation) that, in context, Popper used the word "tolerate" to mean something like "refrain from using violence to suppress".  If that is so, then the "intolerant" are those who advocate using violence to suppress their opponents.  Whereas today's common usage of "intolerant" probably includes anyone who expresses negative opinions about groups of people.  It does seem to me that the definition of "tolerance" is pretty important, when interpreting the quote, and I feel much better about it when I think that "intolerant" implies "violent" [bold added]:

Less well known [than other paradoxes] is the paradox of tolerance: Unlimited tolerance must lead to the disappearance of tolerance. If we extend unlimited tolerance even to those who are intolerant, if we are not prepared to defend a tolerant society against the onslaught of the intolerant, then the tolerant will be destroyed, and tolerance with them.—In this formulation, I do not imply, for instance, that we should always suppress the utterance of intolerant philosophies; as long as we can counter them by rational argument and keep them in check by public opinion, suppression would certainly be most unwise. But we should claim the right to suppress them if necessary even by force; for it may easily turn out that they are not prepared to meet us on the level of rational argument, but begin by denouncing all argument; they may forbid their followers to listen to rational argument, because it is deceptive, and teach them to answer arguments by the use of their fists or pistols. We should therefore claim, in the name of tolerance, the right not to tolerate the intolerant. We should claim that any movement preaching intolerance places itself outside the law and we should consider incitement to intolerance and persecution as criminal, in the same way as we should consider incitement to murder, or to kidnapping, or to the revival of the slave trade, as criminal.

I think the relevant limits on free speech in America are generally taken to be "making specific, credible threats of violence, or advocating imminent lawless action".  So, under the above interpretation, Popper is saying that, for those who advocate an ideology that says its followers should commit certain crimes, we should prefer to oppose them peacefully, but we would be justified in opposing them by force and should do so when necessary; this would amount to somewhat broadening the above encroachments on free speech (e.g. by saying "advocating lawless action, even if it's not imminent, still counts if certain additional conditions apply").  I'm certainly sympathetic to his position, and might agree with it—it would be nice to see a proper statement of what the new speech limit should be.

On the notion of "paradox", here's a related thing: Libertarians often subscribe to the non-aggression principle, which is usually interpreted to mean that violence is aggressive and therefore bad, unless it is used to either punish a prior act of aggressive violence (usually with some notion of proportionality), or stop an ongoing or possibly an imminent act of aggressive violence.  One could look at this and say "So you want to minimize violence, but you're on board with committing violence; that's paradoxical."  But I think most people usually don't consider it paradoxical to draw distinctions between kidnapping an innocent and throwing a convicted criminal in jail, or between murder and killing in self-defense.

While these are relevant elaborations on the paradox of tolerance, I'd also be curious to hear your opinion on the proposal I'm making here - could tolerance be contagious, without any intentional action to make it so (violent or otherwise)? If so, could that make the existence of an absolutely tolerant society conceivable? 

Tolerance (of whatever form) certainly can be contagious.  All behaviors are at least a little contagious—I think we have a module in our brains that observes what other people do and at least considers what it would be like if we did it too.  Tolerance as "go ahead and disagree with me and insult me, and I will defend to the death your right to do so" I find inspiring; tolerance as "absolute pacifism" I do not, but I can imagine respecting someone who truly stuck to it—e.g. refusing to fight back even to the point of their own death—and probably some others do find it inspiring.  Martyrs in general can be inspiring and tend to make those who killed them look bad.

Also, tolerance in probably any form decreases the incentive for others to be intolerant: one of the major use cases for violence, insults, or other "behavior meant to hurt someone" is as revenge/punishment to disincentivize others from hurting you, and if others' tolerance means they're unlikely to hurt you in the first place, then there's less need for aggressive behavior to be top-of-mind for you.

That said, some of the above arguments also imply that intolerance is contagious.  Intolerant behavior is at least slightly contagious due to monkey-brain imitation, and creates an incentive for others to become violent (for self-defense if nothing else).

Is an absolutely tolerant society conceivable?  Here I think it does depend crucially on what "tolerant" means.

One point I'd make is, a small percentage of the population is truly bad apples.  Sociopaths, who don't care about morality or others' pain; sadists, who enjoy causing it.  People who respond to "turning the other cheek" with "Great, I'll hit that one too!"; people who would accept charity and then take the opportunity to steal more from their benefactors.  Probably some of them suffer, and feel wronged (by someone or by society), and therefore feel entitled to grab what they want, or want to make others suffer as a kind of revenge.  Others think "morality" is a stupid delusion, and anyone bound by it is weak and contemptuous, and that the weak are annoying and should be crushed; others may be animated by an ideology that says "these people are oppressors and deserve everything we might do to them", and perhaps even "those people are enablers and deserve no better".  Also, when one adds various kinds of insanity into the mix... Well, I don't claim to understand evil particularly well, and am not eager to understand it better.

The point is: (a) evil exists, (b) in many forms, (c) some percentage of which is completely incorrigible and will just do evil things.  The question is then, is (c) going to destabilize your 100% tolerant society?  If "tolerance" means letting criminals torch the city unchecked, I'd say yes.  If "tolerance" means letting the Nazis and the communists hold their meetings and distribute their pamphlets, without the FBI doing anything until the activists get sufficiently specific in their plans for violent revolution... I'd say it might work, and might not work; that whether those movements would grow to the point that they're a real threat depends a lot on the population, the culture, and so on.  Karl Popper (in my interpretation) would say he hopes that would be fine, but that if they become a big enough threat we should send in the FBI preemptively and feel justified doing so.

So yes, I agree that intolerance can also be contagious - and it's sort of a quantitative question of which one outweighs the other. I don't personally believe in "evil" (as you sort of hint there, I believe that if we are sufficiently eager to understand, we can always find common humanity with anyone) - but all kinds of neurodivergences, such as biological lack of empathy, do exist, and while we need not stigmatize them, they may be socially disruptive (like torching a city). Again, the question of whether our absolutely tolerant society can be stable in face of psychopaths torching cities once in a while I think is a quantitative one. 

But what I'm excited about here is that in the case that those quantities are sufficient (tolerance is sufficiently contagious, psychopaths are sufficiently rare, etc), then we could have an absolutely tolerant society - even in that pacifist way you don't quite like. And that possibility in itself I find exciting. And that possibility is something that I think Popper did not see. 

I think that successful behavior is contagious, unsuccessful behavior much less so. People will try to copy what you do if they see that you gain something from that behavior. (The rewards may be social, and they are not necessarily worth it. For example, martyrs are admired by fellow believers, and in some people this triggers the reaction "I want to be admired, too". Even if, from selfish perspective, dying is too much of a cost, and you will not live to experience the benefits of the admiration. Yet the emotion "I want to be admired, too" can be strong.)

If this is correct, then tolerance will be copied only when it is generally recognized as successful. For example, in today's society, most people see neo-Nazis as losers. Like, yeah, they may be dangerous if you meet them alone, but that probably happens rarely. More importantly, they will never be invited to someone's party, unless that person is also a fellow neo-Nazi. They are a despised minority -- and tolerance wins because few people volunteer to join a despised minority, and some members of the despised minority gradually figure out that it's not worth it and they leave.

Now imagine a Third-Reich scenario, where Nazis are the people in power, they are rich and successful, they collectively own the government, business, academia, and press. Not approving of Nazis violence gets you cancelled or worse. How many people would copy tolerance in this scenario?

Just because good behavior is contagious doesn't mean it will necessarily spread. There are people and/or organizations that purposefully want to troll or be bad actors either for monetary gain (false short report to manipulate a stock price), political gain (foreign interference in an election) or out of their enjoyment or possibly some mental disorder or to show off or enjoyment of breaking norms. A rational discussion or showing good behavior might not be effective in stopping the intolerant speech or behavior. So just displaying good behavior in response to bad behavior doesn't always solve the problem and sometimes the bad or evil actions need to be fought and/or not tolerated. In terms of hate speech, I've noticed that often people that conduct hate speech don't want to engage in rational discussions and a lot of the hate speech on social media is generated by bots. That intolerant hate speech then might become contagious in a negative way!

I'd say the paradox of tolerance is probably correct, in the sense that an Adversarial example can always be constructed as a counterexample to a perfectly tolerant society, and I suspect it's related to the reason you can't optimize all goals at once, or why no learning algorithm performs better than any other on all possible problems.

I think your perspective also relies on an implicit assumption which may be flawed. Not quite sure what it is exactly - but something around assuming that agents are primarily goal-directed entities. This is the game-theoretic context - and in that case, you may be quite right. 

But here I'm trying to point out precisely that people have qualities beyond the assumptions of a game-theoretic setup. Most of the times we don't actually know what our goals are or where those goals came from. So I guess here I'm thinking of people more as dynamical systems.