I mean, I don't know that we need to coin a new term "diversity libertarianism"... to me this just seems to be, well, classical liberalism. (Or if I'm wrong about that, because I might be, there's also already the existing therm "thick libertarianism".) I mean, go read "On Liberty", right? OK, "On Liberty" isn't so much about the value of diversity in the marketplace, but still, a lot of the other ideas mentioned here are there.
This to my mind is why so much of libertarianism, that seems to focus *purely* on opposition to government, really seems, well, off-center to me. Government is only an *example* of the problem. The more general problem is conformism/traditionalism/authoritarianism/group-ism/mob-ism/whatever you want to call it, and the attendant social pressure. Again, Mill covers this! But many libertarians seem to ignore this.
(I should also note here, that when I say "authoritarianism" above, I *don't* mean overbearing government. I've commented here before that I don't like the two-axis left/right libertarian/authoritarian model of politics-space. I *definitely* don't like the left-right axis, which to my mind is a total jumble. But libertarian-authoritarian I think is also unnatural; I don't think "more government vs less government" is a good axis along which to think, and I don't think it's useful for discussing any political positions other than thin libertarianism specifically. But the reason I bring it up is because of how the libertarian movement has taken this word "authoritarian" and used it over and over to mean "overbearing government", and now when I want to use it to mean something *else* I have to explain, no I don't mean that. Of course I haven't explained what I *do* mean, but, um, I'm just going to hope it's clear from context?)
(Also on that note Scott I'm hoping you got my recent email on this subject :P )
(Also, there seem to be some varieties of libertarianism that don't really feel classically liberal at all... like, the sort that's OK with a dictator so long as that dictator is corrupt and doesn't do very much, because, see, that's small government. OK, but it's not, like, rule of law, let alone democracy...)
I don't think talking about this in terms of "religion" is helpful, because I think the fundamental problem here is, well, more fundamental than that; religions are again merely an example.
...I guess I can't conclude this comment without making a note of my idea that we should really think in terms of a broad principle of *orthogonality* underlying this sort of liberalism, which I don't think many people have noted or made explicit, but, I'm not actually about to go on about this here, so I guess I'll just note this idea rather than explaining it. :P
Anyway yes good post, except for the part where you needlessly invent new terms. :P
I had to stare at the graphs for a little bit. Am I correct that the X-axis ranges from worse to better outcomes and the Y-axis ranges from less to more likely?
> Or more generally: in an area with frequent catastrophes, where the catastrophes have externalities on people who didn’t choose them, you want to lower variance, so that nothing ever gets bad enough to produce the catastrophe.
> In an area where people can choose whatever they want, and are smart enough to choose good things rather than bad ones, you want to raise variance, so that the best thing will be very good indeed, and then everybody can choose that and bask in its goodness.
I don't think this is exactly Taleb's point. Externalities vs internalities aren't really core to (anti)-fragility. Fragile systems are composed of large correlated risks. Anti-fragile systems are composed of uncorrelated small risks. So, in the nuclear example, even if the entire world could vote on and agree to building a giant nuclear plant that posed an existential risk to civilization, it would still be fragile to do so, even though we all consented to it and so on.
The same applies to the super Tesla example. Presumably an infinite variance Tesla could be infinitely negative, and somehow destroy the world. That would make it fragile. The key point that makes something anti-fragile is that, as a system, it survives its tail events and goes on to keep improving itself.
I don't know much about this, but I believe: Google is a big smartphone company because they push Android, which is actually free and open software and there are non-Google versions of it out theme, and because of the Play store, which actually can be replaced by other stores (see F-Droid). I am not sure there's such a huge barrier here.
No, I can argue against diversity libertarianism on non-catastrophic grounds. It's not that I think that having a high variation in car companies will cause society to break down (though I don't trust the owners of car companies as far as I can throw them, so I'd rather bury them under regulation just to slow them down). It's that having a high variation in car companies places an unacceptable burden of complexity on uninteresting choices.
If all cars are more or less the same amount of crappy, then if I need a car, I go out and buy a car that's the same amount of crappy as any other car. And I will know in advance how crappy that is, because I will have seen how crappy every other car is. If that's too crappy to be worth the expense, then I simply will not buy a car.
But if some cars are awesome and others are horrific, then I stand a significant risk of ending up with a horrific one. And then libertarians come and point the finger at me, going, "hah hah! That's your own fault for not researching different car models! Free market FTW!"
But the thing is, researching different car models would have taken time out of MY LIFE. That would be the one life I'll ever get. Every moment I am spending on soul-killing grunt work is time I am not spending on doing something that makes me happy. Libertarianism is an attempt to extort me into giving up large chunks of MY LIFE by threatening me with disaster if I don't. I find that entirely unacceptable.
So much about libertarianism makes more sense if you just assume that libertarians are freaks who enjoy drudgery and boredom. I personally wish they'd just take up stamp collection or something instead of trying to reshape society to force everyone else to be as dull as them.
> If you did want to argue against diversity libertarianism, you'd want to show that relevant systems are more fragile than anti-fragile; giving people more options will usually make them worse.
Sure. A world option to get a vaccine and not to get a vaccine is strictly worse than just the former.
Burden of proof? Why do I have to show that the systems in question are fragile; it is the libertarians who should demonstrate that anti-fragility holds. The risks due to fragility are much higher than those due to anti-fragility!
Now that I think about it, a lot of libertarian arguments can be summarized as "here are some reasons why X is actually pretty anti-fragile".
I'd love to hear your opinion on the psychology of libertarians. To me, it's like they knee-jerk protect the notion of "inner space" which they assume is freedom.
When I read the section with the car analogy, an objection immediately came to my mind: no, we don't want the widest possible distribution - at least not in the real world, where there are always tradeoffs, like between price and reliability (and it's these tradeoffs that justify the existence of the distribution in the first place - if there were a Super-Tesla that is strictly superior to Yugos, Yugos would go extinct right away, and the distribution would shrink). I do not want cars on the road that cost 1000 bucks brand-new, but spontaneously go up in flames at inopportune moments, such as driving through a tunnel right in front of me. That's why civilized countries have regulations regarding car safety, regular inspections to extend a vehicle's license, etc. - to cut off the left end of the distribution.
With regards to other industries, similar mechanisms may apply. Do clickbaity "fake-news" websites cause damage to the information landscape of a society? Do shady online pharmacies that sell tainted, counterfeit medications undermine trust in medicine (apart from directly hurting their customers)? Can banks that sell cleverly constructed derivative instruments spread hidden risk to a degree that an unexpected downturn wipes out the entire economy, rather than one unfortunate bank? No waaaay, right? I mean, your condition "and are smart enough to choose good things rather than bad ones" sounds great, but where in the real world would that apply? How many people even have an idea what's in the processed food they eat every day? They may be sure they like the taste, but how are they supposed to be sure it doesn't cause cancer?
I mean, these are objections of the kind you mention in the second-to-last paragraph, but I don't think you take them seriously enough.
I find this unconvincing because of what happens if you take the next step of mathematical sophistication and consider non-symmetric distributions.
Imagine a distribution with a long positive tail, and a sharp cutoff at some point on the negative side (we can accomplish this by including skew). In all of the cases described, it would be the best option. You get high efficiency nuclear power plants and Teslas alike, but eliminate the risk of meltdown or, say, buying a really crappy car.
Then the theory becomes "we always want lots of good options, and some way to limit the possible harm from bad options", which seems... tautological? And it also seems like it eliminates any obvious connection with the concept of fragility.
I think this makes the assumption that innovation is going to solve a lot of global issues, or make the world a better place and add to human flourishing. Systems that work often scale in a fractal way. If you apply government style bureaucracy to a small organization like a school, it produces a public school. But if you just let every teacher do what they want because it will lead to innovation, and creates some adaptable and anti fragile schools, then you will end up with atrocious outcome for some kids, and the few good ideas often wont get reproduced because the market is not setup to reward good ideas in this context. When you have skin in the game and the stakes are high, tried and true solutions make a lot of sense.
What does diversity libertarianism look like from a policy standpoint? Classical libertarians have it easy, they can just oppose anything that increases the reach of the government. But what do you do when all the big tech companies block access to your platform? Have some sort of anti-collusion law to prevent it? Or is it more of a guiding philosophy than a political driver?
A problem with this approach is that there are more axes along which we can measure good vs. bad than you can shake a stick at. We also can't expect to measure most of them accurately. Human measurements like working conditions are hard to pin down in an objective way. Even semi-objective measurements like groundwater pollution have too many internal variables (hundreds of different chemicals and their dispersion/absorption rates) and weighting issues — Is one gram of unobtanium per litre worse than five grams of cyanide deathsulphite? Is one cubic metre of lethal drinking water worse than a thousand cubic metres of merely debilitating drinking water? Is our understanding of human biology able to assign meaningful numbers to such questions, even given the exact chemical ratios, except in some rare situation with middling quantities of ubiquitous, well-understood chemicals?
In addition, why should we expect anyone, including really competent and benevolent government officials, to solve this impossibly complex equation to improve even a single variable without incurring a net loss via one, ten, or a hundred other variables? We're stumbling backwards through the thicket, hoping that the next move will be less painful than the last.
Re: the idea of Facebook choosing to censor BLM content, it's probably worth mentioning that this actually happened. During the first wave of BLM protests, it was found that Facebook's news feed was fairly consistently burying anything to do with the movement: Facebook had designed the algorithm to deliver users the sort of content that people generally like, and people tend to like stuff like the ice bucket challenge and cute animals more than burning cities or grieving families. All this without any "religiously"-motivated intervention on the part of Zuckerberg et al; the platform has a homogenising logic all its own. These days, of course, it runs the other way, but there's still a single uniform affect across pretty much every major social media platform; they deliver content that makes people angry, because that's what the fattest slice of the bell curve seems to want. It seems (to me, at least) that markets actually produce this kind of outcome fairly consistently across all fields. Diversity would be a nice idea, but I'm really not convinced that any kind of libertarianism is the right way to get us there.
You may want to investigate or do liberalism, which has some similar ideas.
That being said, I share many of your moral intuitions here, but wonder if you might be oversanguine about the degree to which market forces and competition produce diversity. You wrote a whole essay on this involving a certain demon king!
> The quickest way to enrage me is to criticize some group of weird people doing their own thing without harming anyone
Sure but that’s basically exactly the opposite of Parler’s mission. They exist to give a platform to weird people who were banned from Twitter because they _were_ trying to harm people.
Yes it’s bad that Apple and Google have all the power here but that doesn’t change the fact that Parler is terrible and everyone is better off without them.
(And I’d say the fix here is that the government needs to do... something. The free market is not going to solve QAnon by itself any more than it solved racism by itself.)
Wait a minute. In your model aren't Apple and Google merely Company As on your wide spectrum? If you think they are *very far* from what the average person wants, or what is optimum, then they're just representing one extreme of the wide "antifragile" spectrum that's desirable, and if we wait around a little, some entrepreneur will see the underserved demand and create the Company B/"Tesla" of the social media platforms and proceed to eat the lunch of those nasty tech oligarchs.
Or to put it another way, if your guiding principle as a diversity libertarian is tolerance for behaviors that are very far from your (or any one individual's) preference -- should you not be *celebrating* the existence of firms that take extremely negative (to you) positions, on the grounds that this must imply the existence (or probable emergence) of firms that will be extremely positive?
I think your two competing distributions take misses a key meta point: if your nuclear power plants are all "mediocre", all doing the same thing, then are they all *safe* or are they all *dangerous*? How do you know? Are you *sure*? The second distribution increases the chance that a nuclear power plant would have a meltdown. The first increase *systemic* risk of all of them melting down because somebody overlooked something and they all copied each other to be "safe". I realize this too is a bog-standard libertarian argument, but worth repeating.
In your second-to-last paragraph, you provide possible rebuttals; situations where diversity libertarianism shouldn't rule the day.
I think the case of Parler did meet some of these criteria, in a way that authentically swayed leaders (like Zuckerberg).
- Social networks are fragile as products. They are worse (the products themselves) when there are too many of them. There are natural network effects that tend towards monopolies.
- Online networks are also fragile in a social-malus way, we discovered, because if they harbor communities that rally into real-world violent actions, and the provide a sheen of acceptability that 4chan / 8chan could not
- People use social networks for perverse real-world outcomes, we realized
And finally, the government was subtly a factor. There was a lot of coverage about how, because the Democrats has just taken the senate with the Georgia elections, the social networks were trying to avoid being hauled in front of congress for allowing Parler to continue.
I agree that the key point about conformity pressure is not whether it comes from the government versus other sources. But I think you underrate regulation (including social conformity pressure).
First, it's not clear that variance reduction on a bell curve is the right way to think about it. An alternative model is that regulation can chop off the left tail of the distribution. For example, we do not in fact allow arbitrarily bad cars; is it clear that every possible car safety regulation affects both tails of the car quality curve equally? That would seem a contentious point, at minimum.
Second, bounded rationality and imperfect information. This blog arose from a community that started out exploring cognitive biases and why they're so persistent, endemic, and difficult to avoid. It's odd to then assert the power of personal choice to avoid bad outcomes. And the blog has explored ideas about the crucial role of culture and tradition to help people avoid repeating mistakes others have made (Heinrich, Chesterton, maybe Scott's metis). By what mechanism does culture do this, if not some form of conformity pressure?
Third, externalities. If we take seriously a consequentialism that weights potential future generations, limiting downside risk becomes extremely important. That should affect the balance with which we strike these variance trade-offs.
And further, I intuitively feel that our future selves deserve consequentalist weight. My present preferences and interests are not in complete alignment with those of past selves or future selves. Younger me made some choices that harmed present-me and future-me. I wish conformity pressure would have prevented him from making those choices. So poor choices have negative externalities even if they only affect the person making the choice.
Can someone give me an example of what Scott calls an "anti-state libertarian" expressing the point of view that we should not care about corporate censorship, mob attacks, or pressures towards social conformity because they do not come from the government? Because I've heard the historically ignorant argument that the First Amendment is all the free speech we need to have, thank you, but that argument doesn't seem to come from libertarians so much as from people who were skeptical of free speech to begin with.
I think I mentioned this last time but there are parallels between the subjective concept of jumping around ideas and the more formal concept of jumping around distributions using advanced monte carlo methods meant to explore bayesian probabilistic spaces thoroughly while reducing chances of getting stuck on local maxima. It's a complex subject with mind blowing math that takes you into higher dimensions inspired by physics in order for jumps to spend more time spiraling the peaks and looking around instead of hovering at the same place for too long. See https://twitter.com/betanalpha/status/1234576972132626445
I realize you're trying to make your examples simple for easy consumption, but none of the examples you showed are truly univariate. Trying to map things onto ill-defined "goodness" is very tricky. For example, I had thought price would be part of a car's "goodness", but John Schilling clearly thought it was not, and lower "goodness" cars would have lower price points, making them more affordable.
More generally, most manufacturers attempt to navigate a hugely-dimensional space (features, quality, formfactor, price point, etc.) and find a region in that space that is in high demand but remains poorly-served. For many products, you trade off one thing for another, a phenomenon a unidimensional illustration can't really illustrate.
Thinking about these things in multidimensional space may actually strengthen your argument for "diversity libertarianism", since the variance is no longer "crappy" and "good" but is instead about having more options available so that a single option is more likely to fit your unique preferences in that hugely multidimensional space.
> Economists have shown that this wasn't exactly organic - companies were afraid to give blacks good jobs, to pay them higher wages, or to sell them products they weren't supposed to have; they figured racist mobs would attack them, racist employees would desert them, or a racist government would crack down on them. I tend to think of this as a religious problem, by analogy to very religious towns where eg nobody will sell liquor even though there's no official law against it,
There was a law against it! Jim Crow was a series of laws by regimes that were incredibly hostile to corporations and business writ large. It's not a coincidence that Jim Crow happened right as 19th century progressivism took off. There was this new ethos that the state should take more control over society to make it better and for racists "better" meant less friendly to African Americans. In the absence of those laws, businesses did hire African Americans. In fact, in the presence of those laws they did hire them anyway. And that required enforcement in the form of crackdowns.
>I'm less likely to object to things like taxing the rich, redistributing wealth, or removing externalities on carbon - none of those decrease diversity very much.
This is logically inconsistent. Redistributing wealth is definitionally meant to decrease diversity in the sense you're using it here. It literally has the goal of changing the distribution from the second graph to the first.
Scott, if you haven't already, I recommend you read about Timur Kuran's theory of preference falsification. I think it solves your "religious problem". It could also explain why "people will perversely choose the worst option rather than the best if many options are made available" as you stated in your steelman paragraph. https://en.wikipedia.org/wiki/Preference_falsification
One other thought on your distribution models: I think regulation tends to push an industry from Distribution 2 to Distribution 1. Where public safety is a big concern you get an industry in Distribution 1, such as nuclear. The real challenge is how do you encourage the variance of Distribution 2 without the downside risk of Nuclear Plant A? How do you chop off the left tail? There's been a recent push from Bill Gates and others to fund innovative nuclear companies which is great but there's been stagnation in that industry since the Chernobyl disaster. Maybe venture capital is the answer but I don't think it gets us all the way there. Public support of safe an effective nuclear power is necessary. Just look at what happened when Gates announced that he was funding COVID vaccine R&D.
"Now Point B is Tesla, making revolutionary new environmentally-friendly cars. In fact, let's say it's some super-Tesla that's even better than the real Tesla"
Bopping off on a tangent here, but this is something I am curious about. Car ads over here are starting to push electric or hybrid vehicles, especially in light of the government's commitment to reducing carbon emissions and making grants and rebates for electric vehicles available.
Dumb question: if Zuckerberg owns 50%+1 of the shares in Facebook, how could anyone ever force him to stop being CEO? I get it, maybe the stock tanks, and maybe the company loses a lot of value but if he truly decided to die on the hill of stopping BLM... what could stop him?
I liked this article, but I think it has one glaring omission: people disagreeing (in good faith or bad) about the distribution associated with any given topic. The biggest proponents of constraining Big Tech (Scott's words, but what I interpret as Big Social Media) would say that their distribution is closer to that of the power plant example, where poor implementations result in catastrophic results (polarization, acts of sedition, persistent harassment, etc.). The proponents of regulation are not promoting a different distribution directly but attacking the perception of the existing distribution of outcomes.
Taleb has said "I am, at the Fed level, libertarian; at the state level, Republican; at the local level, Democrat; and at the family and friends level, a socialist. If that saying doesn’t convince you of the fatuousness of left vs. right labels, nothing will."
Which seems related to the idea of "diversity libertarianism" - the scale has an impact on how you feel about it (holds for nuclear plant vs. car example). Even the laws and rules of a town are very different from the laws and rules of a country. It is easy to move to another town with better rules, harder to move to another state, and much harder yet to leave a country.
Regarding your point about whether people will choose the worst vs best with many options available: I'd like to believe this is true, but I'm not sure how to nest phenomenon like anti-vaxers, q-anon, etc within it. It's seems at least plausible that while for some people or in some situations more options leads to better information / outcomes that doesn't apply in all cases.
A normal distribution (bell curve) does not even approximate the quality of a thing being produced (speech, cars, whatever) under unregulated conditions. To interrogate this idea, you have to start with the acknowledgement that the distribution of quality looks something like a negative binomial distribution (lots of shit and a long tail).
"there were times and places where all companies refused good service to blacks for several hundred years."
And yet railroads had to be forced to provide separate facilities because of the cost of having unused capacity. Your point may be different for railroads because they require huge amounts of capital.
Greg Lukianoff and Jonathan Haidt talk a lot about something they call antifragility in The Coddling of the American Mind. They explain it on page 23:
> Some [things], like china teacups, are fragile: they break easily and cannot heal themselves, so you must handle them gently and keep them away from toddlers. Other things are resilient: they can withstand shocks. Parents usually give their toddlers plastic cups precisely because plastic can survive repeated falls to the floor, although the cups do not benefit from such falls. But ... some things are antifragile. Many of the important systems in our economic and political life are like our immune systems: they require stressors and challenges in order to learn, adapt, and grow. Systems that are antifragile become rigid, weak, and inefficient when nothing challenges them or pushes them to respond vigorously. He notes that muscles, bones, and children are antifragile
The idea that there is a category of things that become stronger only when stressed (but break if stressed too much) is surprisingly useful. I've used "antifragile" in this sense since I read Lukianoff and Haidt's book. But it seems like this is actually pretty different from Taleb's definition?
I don’t think discriminatory car companies works that great as an example. For one, Tesla is impressive because it’s been so long since a new car company really broke though (after a decade of being very niche), and despite massive valuation hasn’t captured that much market share. Two, I think it sorta distracts from your point - like, yes, diversity/competition in providers provides protection against discrimination, but it’s also currently illegal. Given that you compare constraints on tech favorably to car companies, is the argument that companies should be allowed to not sell to particular races if the sector has plausible entry? Doesn’t banning discrimination improve diversity of choices for black people?
"Apple and Google both blocked Parler from their phones." This is, I am pretty sure, is very wrong. You can still download Parler to a Google phone, you just can't get it through their app store.
Anti-fragility seems like the kind of argument that could be levied against almost all things, to the point that you could argue having higher variations helps select things for survival. But it's too vague, too ill defined to be of any actual use. If you have enough variations, enough downside protection and enough natural selection to survive the tail events, you're anti fragile.
For instance, with car companies you actually don't want too much variance. That's why we have laws that limit things on safety and production. Because any increase in variance is a cost to you (the consumer) to figure out what needs to be bought. If you fob that decision off to a third party reviewer, guess what, you're back in "fragile" land, since explicitly or implicitly we're back with single point of failure. The reason we don't have high variance is because we started with high variance, it sucked for a lot of people, and we quickly figured out let's reduce the variance.
"If people are able to choose freely then you wan high variance" is a very high bar that actually doesn't get applied to very many things! It's a tautological argument - if something satisfies that criterion, by definition we've kind of figured out the problems. It's why we have hundreds of cereal options, but the actual variance isn't all that high.
Even diversity libertarianism doesn't stand up to this scrutiny. We don't like the fact that several companies, seemingly independently, reached a conclusion. Sure, this is fair. It's also the outcome of a system that is seemingly anti-fragile - the companies were not coordinating or coerced. If the argument is they all had a worldview, yes they did. But that's what evolving systems do, there's some points where they agree on the ecosystem they're in.
What we have here is the evidence that seemingly anti-fragile systems (with multiple options and competition) devolve on some issues to become unified. Which is also how almost everything works. We always draw lines around how much variance is acceptable. That's most of what our moral evolution has done for us. That's actual civilisation. We can discuss the genius of Swiss Cantons and Renaissance Italy all we want, but the reason most governance systems aren't like that is because those systems are incredibly fragile - to immigration, to cultural cohesion, to coordination failures and more. The US has relatively homogenous 50 quite independent states and it still can barely get anything done in this regard.
To me this is Taleb's genius and his downfall. He's like a poet, creating beautiful symphonies with words and allusions and history. Together it makes you feel as if you can almost reach something ineffable, a point of insight that explains the world. But turns out the inspiration is the actual point, and not an actual theory that you get from the books.
This is similar to a point I've mad about libertarian YIMBYism: There's a question of what the appropriate libertarian reaction is to "state prevents local government from preventing property owners from building more apartments on their land". From the diversity libertarian perspective though, the answer is obviously the YIMBY answer, since it gives us more diversity of land use and density types and gives people more options of what type of area to move to.
> The evangelical Christian town where nobody will tolerate gay people may not have laws against homosexuality, but gay people still won't be able to find a church, community, or business that meets their needs.
I think you've fallen into a trap, Scott. The whole point of the archipelago is that a community can run itself like this if it wants, as long as gay people who are unhappy with it can go somewhere they prefer. Thus a thousand flowers bloom. If your newly coined diversity libertarianism prohibits this, it de facto *flattens* the space of possible arrangements. And there is no telling how much more flattening consistency requires. Remember that one of your original examples was Aquinastopia, where Catholics could go to live in a Catholic community. Is this disallowed now? Does your libertarianism suddenly prohibit Catholics from living as Catholics?
Don't confuse what happens at the level of the evangelical town from what happens at the level of the archipelago. *Everybody in the archipelago* benefits from the fact that this evangelical town (or Aquinastopia, for that matter) exists. The people who want to live there can live there, the people who don't don't have to, and the people who aren't sure can learn one way or the other whether such a community works.
Granting all that, I don't know exactly how to apply this lesson to Big Tech, because we don't actually live in the archipelago. But I think it suggests that size matters. If a tiny company wants to run itself according to religious principles, that's fine, good even; there is a lot of room in the economy for other companies to run themselves differently, and perhaps outcompete it. But if a huge company with near-monopoly power in some industry or another wants to run itself according to religious principles, you have reason as a diversity libertarian to be concerned.
On another point:
> Right now there's religious pressure on tech companies to conform. Someone on Twitter pointed out that tech censoring Parler isn't a sign of their strength, but of their weakness. Imagine that Mark Zuckerberg decided he personally really disliked BLM, and he was going to censor BLM and any people/organizations/apps that promoted it from Facebook. Do you think he would succeed? Do you think he could stay CEO of Facebook after he was found to be doing this? Mark Zuckerberg and Big Tech in general are as much slaves to the prevailing religion as the rest of us; their "power" is the power to choose between medium vs. high levels of conformity.
I made a similar observation on your Erdogan post. Quoting myself:
> Erdogan's purges look to me like a particularly poignant subversion of this: hollowing out every other institution, until the only thing left is him and his party. It should really, really concern us if seemingly every semi-official institution in our country, from the top schools to the top newspapers to the top thinktanks, march in nearly complete uniformity: it raises the question whether they, and the people in them, can act of their own will at all.
THIS "Even beyond politics, I think this predicts a sort of ethos of what I like and what I don't like. The quickest way to enrage me is to criticize some group of weird people doing their own thing without harming anyone - to try to browbeat them into doing the same thing as everyone else. That group might be just a tiny bit of slack away from creating something amazing!"
This argument seems weaker than it should be due to using dumb examples. To take the less controversial one, car companies are not actually easy to create.
Yes, it’s true that many industrial nations have a car company, and some have more than one. They don’t seem to be nearly as difficult as building a competitive aerospace industry, which China has been trying to do for quite some time with little success so far.
But in the US, there were no new car companies for many decades. Tesla had a lot of near-death experiences along the way, and part of their survival strategy was selling luxury cars that most people can’t afford. Crappy 1920’s style cars would today be considered unsafe at any speed and the idea that a car company is something a few people can build in a garage is just nonsense. (Unless we are talking about grey-market golf-cart style electric vehicles, which are apparently a thing in China.)
By contrast, building a crappy cell phone seems much easier, at least in China where there seem to be many manufacturers building them using a common ecosystem of standard parts. Maybe that would have been a better example?
I'm not sure your car example checks out. In the first world cars are heavily regulated, perhaps even more than nuclear plants. The only reason anybody bothers to make cars at all is that the demand for them is so high that almost any regulatory burden is worth the effort.
On the other hand, you might be amused to find out about the unregulated electric golf cart in China. If the car is small, slow, cheap and electric, the externalities are minimal and government feels less pressure to regulate them.
This is an interesting framing! Game theory (or, well, math) can outline some situations where you would want to migrate your preferences between systems that are fragile and antifragile.
Generally if you're losing, you want high variance systems.
If you're winning, you want the status quo to sit like a diamond, more lock in is better.
So, Tim Wu's The Master Switch gives a history of communication companies all following this similar trajectory: scrappy startups finding workarounds that change all the rules and disrupt everything, then becoming large institutional players that want greater and greater regulation to prevent anyone else from inventing anything that would ever change the world in any way.
Do you want to roll a d20 or 3d6 to hit? Depends whether you're fighting kobolds or dragons. (There are even situations, in gambling and in life, where you would even want to take a lower "average" return, in order to shift the variance.)
So it's possible to entertain a meta-political perspective, where you think society should wander between libertarian (or, "50 labs federalism") vs. more central standardization, based on how well its currently meeting certain markers you've laid out for the health of society or the effectiveness of policy or whatever.
I don't know how you can make arguments like this without getting into object-level illustrative examples, but as soon as you do that, your argument becomes subject to the object-level examples being wrong. It is absolutely not true that market entry for new automakers is easier than new phone makers. Parler is accessible from a browser. Blocking them from native app stores doesn't block anyone's access. I have no idea if there even is a native app for Substack, yet none of us are blocked from reading your blog. What actually took Parler offline (for what, a week?) was Amazon Web Services turning off their servers. But even there, as coming back online showed, AWS isn't the only provider of managed infrastructure, and you can always self-host if literally no one will sell you service.
Until such time as a peer to peer radio network with TCP/IP capabilities exists, the only true monopoly service that can cut you off from the web are backbone ISP providers. If you can't hook into them, you can't hook into the network, whether your servers are running or not. And it stands out that a whole lot of people calling themselves libertarians have complained quite loudly about Google and Apple shutting Parler out of their app stores, but not about net neutrality, which actually addresses the only true monopoly providers controlling access to the web. It makes it seem more partisan than ideological.
Of course, I don't think you personally are partisan. You're an ideologue. But an inherent issue with trying to claim allegiance to any sort of named ideology is you immediately associate yourself with other people using the same name. "Diversity libertarianism" doesn't seem to me like it really captures what you're trying to say here. You're just arguing for classic welfare economics on the merits of the Pareto optimality achieved under conditions of pure competition. You're arguing against the existence of monopolies, whether those be warlords, mafias, governments, big businesses, religions, cultures, it doesn't matter.
I don't have any answer here on what you should call yourself, but you have to realize that insisting on "libertarian" even if you tag "diversity" in front of it alienates and marginalizes you from a lot of people who are going to have a knee jerk bad reaction to that term because of who else uses it. I can understand wanting to reclaim it. That seems to have worked for the word queer. I think it probably hasn't for slut. But is that really your goal? Do you really care what people call you? Why not just argue for specific positions without needing to identify and name an underlying guiding principle that may or may not really be there? This seems to be Taleb's problem, too. There are undoubtedly some things he is very right about, but as soon as he tries to go from those examples to generalize to a grand theory of everything, it doesn't work and he becomes trivially wrong about most of them. If your position is some regulations are good, some are bad, and the guiding principle is whether or not they promote or prevent monopolies, call yourself an "anti-monopolist." What does Matt Stoller call himself? Unfortunately, probably libertarian, based on the name of his organization.
I'm not even sure "against monopolies" can stay ideologically pure and coherent. How do you prevent monopolies from forming without a monopoly on law enforcement? I'm kind of with John Schilling on this. David Friedman can probably give some elegant sounding theory backed up with an example from 12th century Iceland being the only working implementation, but that doesn't leave me super confident it can really widely work.
"But other fields have higher entry barriers than cars do. Apple and Google both blocked Parler from their phones. But these are the only two major smartphone companies. "
Forgive me if this has been posted previously, but there are lots of smartphone companies. There are, however, only two major operating systems for smartphones, and most of the smartphone companies use one or the other.
Taleb isn't a particularly clear writer, so that's probably why it's hard to grasp the concept of antifragility (it's also stupidity named, but that is neither here nor there). The basic variable is the response to unpredictable stress. Fragile systems are weakened by it, robust systems are neither weakened bit strengthened by it, and antifragile systems are strengthened by it. Exercise is antifragile, because its best results occur when workouts are irregular and non-repetitive. Option trading is antifragile because it only really lucrative when you bet on ludicrous occurrences. It's not so much that the range of outcomes is highly variable, so much as that they aren't even calculable. The point isn't so much that "we" (as in humans generally) need to find a more accurate way to assess risk, as much as "we" need to realize that risk can't be calculated for everything.
"smart enough to choose good things rather than bad ones"
I think this falls into a mistake that Scott often makes - it overstates the importance of intelligence and understates the importance of information and of the cultural guidance to understand that you have a choice and on how to make the choice.
On an information example, the choice of medical providers. Most people are not medically qualified to tell which providers are good and which bad - low variance is good if you really can't tell a good doctor from a charlatan. Moreover, a lot of people default to using price to judge, assuming that more expensive medical choices are better. Even if the information is available to know which doctors are better than others (and the medical profession works hard to keep that information unavailable), lots of people don't know how to evaluate it.
On a cultural guidance example, the choice of colleges. Middle-class parents are much more likely to have opinions on which colleges are better than others, and those opinions are much more likely to be correct than working-class parents [I'm using class in the class-as-culture sense, not the class-as-economics sense]. Also, they're more likely to know about how to get scholarships, how to apply for and get income-based discounts to tuition, and so on. Knowing that you won't have to pay sticker price because you will get need-based aid and scholarships and so on is information that many working-class potential students do not have, resulting in them being put off by a price that they won't themselves have to pay.
I think you may be taking the premises of your argument for granted. The classic libertarian arguments ignore structural impediments to optimal (or rational) behavior. For example, the ideal of hearing as many opinions as possible so as to maximize your chance of hearing the true one seriously ignores reality. In reality, there are time constraints (e.g., does a single mother working two jobs have time to sufficiently read and evaluate every argument for and against gun control?), limitations on an individual's ability to process and filter information correctly, and demagogues that prey on human nature weaknesses (e.g., Peoples Temple 1978). Also, not every opinion is created equal. Is a plumber equally suited to evaluate the efficacy of a drug from experimental data as a research scientist? Qualifications matter. As do motivations - if two scientists reviewed the efficacy of a new drug and one was being paid by the pharmaceutical sponsor, would that impact your judgement of her credibility? If no forced disclosure, you wouldn't know. I'm not convinced by the premise that people are smart enough to choose good things rather than bad ones. Humans are flawed. Society is flawed. The way we govern society should reflect the flaws.
And then we move to free market fundamentalism, which you acknowledge in the last paragraph is flawed. In almost every important sector, there are hugely important externalities. That is not an argument for universal regulations and regulations should always be thoughtful and, to the extent possible, tested in order to avoid unintended consequences. But it does recognize that humans are inherently self-interested and short sighted. Oil magnates care about their economics, not the environment. The Sacklers cared about their fortune, not the risk of addiction. Volkswagen cared about their quarterly earnings, not the Clean Air Act. Zuckerberg cares about world domination (?), not the risk of facilitating genocides. Regulations need to reshape incentives to harness self-interest but also protect society from our inability to process risk and large scale ruin in addition to our bias of discounting the future.
The example of “nuclear plant” as antifragile bothers me but I’m having trouble nailing down why.
A single instance of tech is often fragile. One car crash does not make that car stronger. It can’t reorganize itself and sometimes it can’t economically be rebuilt individually (“totaled”). Many car crashes, when abstracted into data, can be used by companies to improve the fragile product and in this sense a company is more fragile or more antifragile (whether in service of better safety reputation attracting more buyers, or compliance with a regulatory environment) depending on company culture and how they leverage the data into better product.
Any single instance of tech is fragile, so any single nuclear plant is also fragile. Nuclear plants with the same design flaw don’t all blow up at once, either, when they are individually distant and subject to different circumstances.
However the nuclear engineering culture has to be antifragile; when an “accident” occurs, they scramble to learn how to more successfully create a safer plant. It has to learn from mistakes and respond to stress. From what I understand there were times historically when a potentially safer design was not chosen for political reasons; nuclear regulatory culture has to be less fragile than some regulatory cultures because they are trying to balance tech advancement, cost, and safety in a context of large externalities, and they have more actual influence (I think) on next-generation plant design than say the DOT has on the design of exhaust filters for idling trucks. It is a government-industry partnership in the US in ways other industries are not like car and truck companies.
The regulatory culture is fragile to financial pressure but antifragile to nuclear accidents. “The safer design is too expensive” is a barrier. But the pool of players that benefit from a functioning nuclear industry is large enough that the military, academia and national labs all get in on making the safer design cheaper, in an antifragile way.
In other words I don’t think it’s accurate to say the nuclear industry is like company A. For it to exist at all there is a lot of this industry-government design pressure to reject dangerous designs before they are ever built, and then learn as much as possible from accidents. I’d call it antifragile with a floor. If the industry was fragile they’d have given up and gone home after the zillionth unintentional radiation release. Any design process is allowing some variance and trying to capture the top end of it.
I’m trying to come up with an example of what is fragile but just got interrupted. I’m not saying nothing is a fragile system just that nuclear plants/power is not the best example.
Typo: "Tesla did something like this when Musk though the big dogs were neglecting electric, and that went great." I think "though" should have been "thought".
Scott, I love the clarity of laying this out with distributions and proper dissociation between different strains of libertarianism, but what I don't understand here is your position on the dynamic state of distribution 2.
You say distribution 2 is good in cases where everyone can and does make good choices, because everyone will buy from B and A will go out of business.
But at that point, distribution 2 no longer exists; everything on the left side of the graph is gone, and the reality looks like distribution 1 again!
(with a higher average value, thankfully)
So, my question is: as a diversity libertarian, do you actually want to be *living in* a world that looks like distribution 2, or do you want that distribution to have existed *at some time in the past* but be gone now?
If something that looks like distribution 2 has endured for a few decades or so, isn't that strong evidence that people in reality *don't* have either the ability or competence to make good choice in that space, since if they did all the As would be gone already and things would have converged on a distribution 2 made of all the Bs?
This gets to the point about Parler - in what way is Parler not just an A that the community recognized was bad and abandoned? Sure, there's the weirdness about boycotting it individually vs. boycotting its corporate partners, but in either case it seems like the majority of consumers decided it was awful and wanted it abolished, which is what you should expect to happen to As in distribution 2.
It seems to me like 'diversity libertarianism' as you describe it is too time-independent; it imagines a single point in time where you have a lot of options, and a single instance of decision-making where you choose the best option. In that singl instance, yese, you want tohe world to look like distribution 2.
But run that dynamic over time and measure the world at the limit, and it looks more like a world where all the bad choices are ignored and go extinct and only the good choices survive. Which looks like distribution 1, and possibly looks like 'ideological conformity.'
Given that, it feels to me like diversity libertarianism should be really really focused on making sure that *new* ideas have a chance to be generated and survive long enough to be fairly evaluated and take on their proper market share, and a lot less focused on old ideas vanishing when the market turns against them and moves on.
Ie, moving as rapidly as possible from distribution 2 to distribution 3 to distribution 4, rather than protecting the As from distribution 2 who cry that they're being canceled.
I don't think Tesla being started by Elon Musk, already a billionaire at the time, is very much at all like some guy creating a huge company by starting with a few crappy cars put together in his garage. Without getting in to all the regulatory and other barriers to entry that exist in car manufacturing, isn't plain old capital ultimately the biggest barrier to entry in pretty much everything?
People seem to be getting stuck on the "l" word a lot. The more important point is that large scale monopolies dominating an ecosystem reduce the variance that leads to positive innovations that improve human lives.
Insofar as most of the world is "governed" by legacy institutions ruled by stationary bandits (aka kleptocracies), the opportunity cost of not allowing "a thousand nations bloom" via new jurisdictions is immense. Hong Kong and Singapore showed the value of economic freedom more effectively than did any arguments, China copied them in designing their SEZs, and a billion people saw a 10x increase in their wages over twenty years, arguably one of the greatest moral achievements in history.
We don't need to agree or even know exactly what leads to success in an ecosystem open to innovation. As the Seasteaders say, "Don't argue, build." But the more we support an ecosystem of innovation with respect to the creation of new jurisdictions on land, the more rapidly we're likely to discover new and better forms of law and governance. Competition among Greek city states led to Athens; competition among medieval Italian city states led to the Renaissance; competition among the "Free Cities" of the Hanseatic League led to modern commercial society via the success of Holland, which in turn played a key role in the rise of British and American prosperity. More broadly, as David Landes has argued, Europe's fragmentation was key to its prosperity, including the fragmentation that allowed persecuted thinkers to move from jurisdiction to jurisdiction.
By contrast, large scale empires typically stagnated over time. Modern prosperity is thanks to competing jurisdictions, a sine qua non of the discovery process of the institutions required for prosperity.
Should we argue over the "l" word or get busy creating new and better institutions for human flourishing?
That said, although I'm also in favor of internalizing externalities and supporting the poor (and thus could pass as a "liberal"), many of those who have focused on the need for liberating innovators have at least some familiarity with Hayek and others who emphasize the stultifying effect of the state. Thus I'd rather be considered a bleeding heart libertarian than be part of the "progressive" tribe that rarely focuses on the morally urgent need for permissionless innovation (McCloskey).
From the outside view, increased variance increases both the high and the low ends of the range of possibilities. However I think an unfortunately common inside view, mapped onto this framework, goes as follows: my politics/tastes/whatever are basically Correct, and so variance is entirely bad: the high end is a fixed point (at me), so increased variance just cashes out as more people being increasingly Wrong.
The other place where the libertarian argument falls down is when time is critical. Sure, if social media companies censor certain info, you can start your own and migrate people there (if the AWS/Apple/Google thing doesn't happen). However, if the censored information happens to be highly pertinent to an election happening in a month, good luck
If the low variance scenario results from everyone happening to agree on a narrow range of desirable approaches, and everyone further agrees that exploration outside A and B will not be legally or socially punished, that seems fine. Obviously this doesn’t mean that nothing should ever be legally or socially punished. So what determines what we can prohibit legitimately?
Maybe the post was inspired by Taleb, but clearly the basic issue is different, so speaking in terms of antifragility confuses matters.
Quick counterintuitive thought: A narrow distribution of ideas may be beneficial in the public sphere. Now reflexively, I'm inclined to like the marketplace-of-ideas theory of discourse. I like markets, I like choices, I like diversity and I love ideas. So I'm offering this argument with a big grain of salt and perhaps against my own interests:
The "mainstream" of actionable public ideologies is akin to nuclear reactors: You want a narrow, safe range of variation more than you want a high-variance set that might include world-saving and world-breaking ideas. (Note: In practice, the resilience of the "world" at stake would matter here. High-resilience polities could brook more ideological variation.) Hence recent gatekeepers like publications, political parties, legislatures, universities. Before that, nobility, royal courts, hierarchical religions. And always social structures that reward and punish. Specific claim: In a population where X percent of the population believes in wrong-but-arguably-harmless ideas (angels or UFO? Your mileage will vary), an equal percentage of people is ready to believe wrong-and-potentially-catastrophic ideas (fill in your own blank here -- I'll put the crayon in your hand, but you have to draw your own monster). And maybe exposure to those ideas creates the potential for greater adoption. In such a case, gatekeepers that limit public discourse to "what the decent people think" might be performing a service, even if they limit exposure to fantastic new ideas. (While still allowing for hypothetical mechanisms for evolutionary fit ideas to grow in acceptance and no-longer-adaptive mainstream ideas to whither and die?)
Prediction: If this is true, societies/cultures/groups with greater stability (in terms of less vulnerability to ideological-triggered disaster?) brook more dissent/wrongthink.
The odds of this being true: I find this line of thought more interesting than persuasive. I'll admit to some nostalgia for a world without comment sections. But I can also think of ways that world was less good than this one. I do think we're in the long process of adjusting to the consequences of rapid advances in communication technology, but I need to be humble about my ability to assess what that means and where the process goes, and even less confident I can easily distinguish what's really new from what has always been thus.
"But other fields have higher entry barriers than cars do."
Er, I think car manufacturing has one of the highest barriers to entry of all consumer products? There's a reason that Tesla is the only successful new American car company to start in the last 50+ years.
> If Ford refuses to sell cars to black people, Toyota should see a profit opportunity and step in. If both Ford and Toyota ban blacks at the same time, some upstart like Tesla should step in. If Ford, Toyota, and Tesla all do it, some guy with a wrench who's always dreamed of making cars in his garage should notice a billion-dollar business opportunity lying on the ground, get seed capital from equally greedy investors, and solve the problem.
I don't think this actually works very well in practice. Suppose the hypothetical "racist car companies" world existed, and you wanted to start NRC, the Not Racist Car company. First, you have a hard time raising money from investors. It's not impossible, but in this hypothetical world the most successful car companies are super racist, and investors are skeptical that doing literally the opposite of every successful car company is going to work. But you find some and are able to raise some capital at extortionate terms. If everything goes well your company is going to make a zillion dollars even though your cost of capital is nuts.
Unfortunately, that's going to take years to get going, even if you're successful. In the meantime, minorities have no access to cars. Even after you start producing, for years or even decades, minorities are forced to Distribution 1 while everyone else has access to Distribution 2. When people talk about "privilege" this is the kind of thing they're talking about.
Now cars are pretty important. Not having a car, or having a crappy car, has lots of knock-on effects. Minorities basically have to live in major urban areas, while non-minorities can live mostly wherever they would like. This limits the jobs available to minorities, which in turn messes with the labor market. So now even equally qualified minorities will make less than non-minorities.
But can they become equally qualified? Not if people want do some extra racism on the sly. Just put your best schools, public services, and businesses far away from public transportation (Hi, Robert Moses!)
At the margin, some people who might have barely gotten a job now have no job at all. Some of those people go all Jean Valjean on you and your loaf of bread, and now your prisons are full of minorities, which reinforces Ford's & Toyota's position.
And of course this has long-range knock-on effects. The Car Inequality is going to take generations before it's dissipated.
My point here is that we can't rely on Diversity Libertarianism in cases like this because most negative externalities need to be regulated, and racism causes massive negative externalities.
Taleb doesn’t hate curves that look sort of “bell-shaped” like the ones you drew, which usually have “fat tails”. He hates GAUSSIAN curves. If you understand the difference between “power law” and “Gaussian”, you get to review Taleb’s work and be taken seriously, but if you make a mistake like this it doesn’t bode well.
The situations you describe are ones where a large majority of the population believe in the religion, in which case a democratic government will be helping them suppress the atheists not stopping them from doing it, as will almost any other form of government. So if you are a diversity libertarian I don't think there is a solution to your problem, except in the special case where the state is 99% fundamentalist (or racist or whatever) and the federal government isn't even majority fundamentalist.
So even in the situation you describe, I am against federal regulation of FB et. al. purported to keep them from censoring. You will note that all the supposed rules against racial discrimination got ignored when the discrimination took a form, affirmative action, that was politically popular.
The problem is proverbial — setting the fox to guard the hen house. The government has a stronger incentive to suppress (some) political speech than FB. FB is only one of several large (and many small) players in its market, so can only suppress speech to the extent that other players agree. The government doesn't have that limitation — it can regulate all of them.
I think the term "diversity libertarianism" makes sense as a type of libertarianism, but I think that the complexity argument for diversity is strictly superior to the libertarian argument, as it handles all the cases the libertarian argument does, and more.
John Stuart Mill's argument for free speech was that many unpopular opinions might contain elements of truth, and hence truth can be assembled from bits and pieces of many opinions. The complexity argument for diversity is that it takes many different kinds of pieces to make a complex machine. To build complex social mechanisms, we'd like many different units of any kind--opinions, ethnicities, personality types, and educational backgrounds.
In its original historical context, libertarianism effectively opposed ancient Platonist tradition, which says that there is one perfect, eternal model for everything, including one perfect opinion to hold on any subject, which is True in all times and places; and the job of society is to find that one perfect opinion and shove it down everybody's throat and forbid anybody from ever questioning it.
Locke used empiricist epistemology to argue (among other reasons) that we can't ever be sure that what we think is the one Truth really is the one Truth, so we better listen to all opinions. But (AFAIK) he didn't challenge the idea that there is just one Truth.
Mill's argument for freedom of speech doesn't convince people who are /really sure/ that they do, in fact, know The Truth. The complexity argument for diversity gives a different view of the same issue--not that we want more opinions for society to choose from, but that we want people with different opinions to remain in the population indefinitely, because this difference in opinion will produce division of labor, and a discrepancy in values which makes it more possible for everyone to get more of what they personally value most. It rejects the very notion of converging on a final Truth.
This post starts out riffing on the tradeoffs between risk and benefit variances and then pivots to the issue monopoly vs. "diversity" in the free market.
The first part just reflects the fact that an optimal balance of risk and return depends on the relative distribution of the various outcomes (as best you can forecast), multiplied by the costs and benefits associated with those outcomes. Like everything in economics, you are optimizing when the marginal benefits are at least equal to the marginal costs. (i.e., MB = MC).
Everything that follows on the risk-benefit topic is mostly a recognition that, even assuming arguendo that you can accurately predict the probabilities of various outcomes, different people will still have very different optimizing points (MB = MC) due to: (a) being differently impacted by the costs and benefits of particular outcomes (e.g. someone living close to an accident-prone plant, vs. far away); and (b) having different tolerances for risk (e.g., someone who can sleep like a baby knowing there is a 1/10,000 chance of a plant explosion vs. someone who can't stop worrying about it).
Bottom line: The theory of making decisions under uncertainty is uncomplicated. But ascertaining the true probabilities of outcomes, and then politically mediating the interests of lots and lots of people with different interests and preferences is pretty impossibly complicated.
I like this description. Mostly I just err on the side of freedom. This generally results in me being anti-regulation, but I also don't like when monopolies restrict speech. It's the same root principle, but people think there's some deep contradiction there.
Also the block being a result of big tech weakness is a good point not made often enough.
There is nothing inherently good about "Diversity." It's only potentially a good thing if: (a) the "diverse" things have different merits; and (b) people are allowed to judge which diverse options have the most merit and reject the others. Once this process has run its course, whatever is left over is the optimal amount of "diversity."
For example, having lots of restaurants creates a "diversity" of cuisines to choose from. That's great if you are allowed to choose to patronize the ones you like (say, Thai and Italian), and skip the ones you don't (say, the "House of Haggis" Scottish restaurant). The modern conception of "diversity" claims, however, that forcing people to eat Haggis enhances the "diversity" of food and is therefore improving them by definition.
Only when all the bad restaurants have been driven out of business is the remaining "diversity" of choice a "strength."
>(If you did want to argue against diversity libertarianism, you'd want to show that relevant systems are more fragile than anti-fragile; giving people more options will usually make them worse. Maybe the slightest failure could cause catastrophe - I think free speech opponents think this is true, but I’m less sure they’re right. Or people will perversely choose the worst option rather than the best if many options are made available - I think this is sometimes true for things like drugs and gambling. or there are overwhelming externalities. Or there are externalities, which can range from the very simple like pollution to the very complicated, like whether your working for a low wage has externalities because it forces me to compete with you. I think figuring out where to draw the lines here is really hard - but if you want to convince me, this would be more fruitful than the umpteenth essay about how the First Amendment only applies to government)
I think that one can object to absolute freedom of speech on the grounds not that people would be somehow "brainwashed" if exposed to ideas outside a particular approved band, but rather through understanding certain types of speech as primarily acts and the majority of people as basically unthinking followers of a particular flag. That is, someone like Richard Spencer is not contributing to a marketplace of ideas so much as raising a flag for people who already agree with him to rally around as a coordination point.
I also think it's somewhat of an errant frame to suggest that you can *only* adjust volatility-- like, in the car market example, it is not in fact the case that in order for there to be better cars there must be worse cars. The distribution of cars on the market on a scale of goodness to badness could, like, /move towards being gooder as a whole/. In relation to specifically cultural freedom of speech -- that is, /not/ mediated through large corporations but at the level of, say, a bunch of individuals denouncing someone for their speech --, I don't think it's a problem for bad ideas to become harder to access or for the "sellers" of bad ideas to go bankrupt or acquire correspondingly bad reputations.
Freedom of speech is important descendant from a general orientation towards maximizing eudaemonia and the ability of people to make real choices, and needs to be evaluted within that context, not as a separate magisterium.
>When might this not work? First, if cars are hard, and starting a new car company takes too much time. I'm not too concerned about this one. I think a lot of people could potentially make one or two crappy 1920s-style cars in their garage with a little work, and once they do that, black people with no other options will buy them, and that will give them enough money to bootstrap into a powerful car company that can compete with the big dogs. Tesla did something like this when Musk though the big dogs were neglecting electric, and that went great.
This is completely unrelated to the last thing I said as far as I can tell, but I think stuff like opening up a new crappy car company in your garage is likely to be harder than it ~ought to be in an ideally free market due to government regulations on car companies.
Sorry but this is a major theme running through ALL of Taleb’s books including this one. The canonical “fat tailed” distribution is the Pareto or “power law” distribution, and the canonical “thin tailed” distribution is the Gaussian or “normal” distribution, and treating the former as if it were the latter is what causes all the blowups and disasters he is trying to teach how to prevent.
This is the pons asinorum for understanding Taleb in any way other than impressionistically.
The trouble with corporate censorship is that "uncensored" forums tend to be on the "cars that immediately catch fire" side of the curve - moderation is a necessary part of building a forum that people actually want to use. Parler has already learned this the hard way - their policy rapidly shifted from "we're the good guys because we don't censor people" to "we're the good guys because we censor left-wing trolls, while everyone else censors right-wing trolls." Even under a "diversity libertarian" regime, you still have to be able to make money off your idea.
Also, factual correction: Apple and Google blocked Parler from their *app stores*, not their phones. You can install Parler on Android by downloading the .apk file and installing it manually, no need to go looking for another cell phone company.
If the profoundness of your articles was plotted it would probably be a bell curve. I'm not quite sure what would be around the mean, but "I can tolerant everyone except the outgroups" would be at least two sigmas above the mean. This article is on the long tail of the less insightful end.
Diversity libertarianism feels like a very strange concept boundary to me. Isn't it basically just capitalism, where you want lots of diversity and competition, but you also want to prevent companies or people from preventing competition?
How does libertarianism come into the picture? Is it just capitalism but you don't want government solving coordination problems because you want the free market to instead?
So what do you think then about the current deconstruction of Republicans? A single *major* party is left: this will eventually kill the competition for the "consumer of policies" (the voter). As in your example, when the super-car wins the competition created by the variance, it likely will just reap the profits and deteriorate on services and features. It will likely fill the niche of the bad car, not eliminate this niche. Because the elites are given the opportunity to think short-term but large rewards for themselves from such a process (with negligible negative consequences). Yet no better system is in evidence.
<< I'm less likely to object to things like taxing the rich, redistributing wealth, or removing externalities on carbon - none of those decrease diversity very much. And I’m more likely to care about conformist pressures from religions or mobs, even though technically those don’t involve government.
But, having such preferences, you are more likely to be left-libertarian: anarchist, socialist, etc., than right-libertarian, libertarian in the usual sense (as in the "Libertarian Party")
You seem to be treating “diversity” and “variance” as synonyms, where “variance” means a distribution along some axis of objective quality. But these are not actually synonyms, and it is important that they aren’t. Diversity just means “differences exist” and those differences may or may not result in variable objective quality. But it can still be valuable even if it doesn’t - people have preferences, some rational and others not, but the preferences are definitely diverse. Diversity, insofar as it meets a broader range of preferences, can therefore have utility even if it does not result in higher objective quality.
When Ford first made the Model T, you could have any color you wanted as long as it was black. But many people prefer cars in other colors, so eventually car companies started offering a range of color options. Offering cars in multiple colors does not improve objective quality in any way (if anything it slightly degrades it since now you’ve thrown an extra variable into the supply chain, plus there is probably some objectively best car color for most situations and people will nevertheless make different choices). So “car color” is a place where diversity is valuable even though it’s not really “variance” in the sense you use it, or the sense that creates anti-fragility.
People having irrational preferences may actually be important to keeping things diverse enough to produce useful variance. Many people have irrational brand loyalties - you might think this is bad, but if they didn’t, as soon as Ford trucks got a little objectively better than Dodge and Chevy trucks, 100% of people would buy Ford trucks, Dodge and Chevy would go out of business, and you’d be left with no diversity or variance. Irrational (maybe “arational” would be a better word) preferences add some friction and hysteresis to the system in a diversity-preserving way.
I think your lack of distinguishing between diversity and variance leaves something missing from your assessment of diversity libertarianism. Namely, libertarians believe that individual freedom is a terminal value. It’s a good thing even if it doesn’t make the world objectively better on average. Libertarians tend to believe that freedom WILL lead to better outcomes on average, but that’s not necessarily the point.
And this is partially why pragmatic libertarians might sometimes accept government intervention - if a coordination problem is limiting individual freedom, it may be better to let the government impose coordination by fiat than let the problem go unsolved. And why they might be skeptical of corporate censorship - sure, it might be a free choice by a corporation, but if the corporation has monopoly powers then their censorship is a significant imposition on individual liberty.
> Plant A suffers a meltdown after two days, killing everybody.
> a lot of people could potentially make one or two crappy 1920s-style cars in their garage [to sell]
Scott, do you have somebody review your pieces to make obvious objections before you publish?
The obvious objection to both of these points, especially the first, is that regulations won't allow these things. So long as we have an effective safety regulator, you absolutely want a high-variance nuclear industry, so that alongside the TEPCOs and V.C. Summers of the world, you have a Moltex/Thorcon/Terrestrial Energy/Elysium, one of whom is likely to be involved with building large numbers of reactors at some point.
P.S. Note my use of the word "regulator" on top of "regulations": you need somebody to notice that every single one of the emergency generators and batteries at Fukushima are located in the basement and ask themselves "should we allow this?" But see also: Meltdown World https://www.reddit.com/r/nuclear/comments/jtm6hm/how_bad_is_meltdown_world
"Diversity libertarianism" seems like the kind of philosophy of government that allows you to construct a justification for creating whatever kind of policy you want. Which is to say, a philosophy with actual legs.
For nuclear plants, when everything is fine, yes, we prefer the low variance distribution.
However! In a world where a nuclear catastrophe has occurred, the high variance distribution is better! If all nuclear power plants are basically the same and one explodes, we've gotta turn them all off ASAP until we figure out what went wrong. When there's a range of plant quality, one incident has an isolated, well, blast radius.
I have no idea how this preference inversion fits into the fragile/anti-fragile framework.
Doesn't distribution 2 inevitably lead to distribution 1 given enough time?
It's like you said - eventually all the bad car companies/bad power plants/bad or whatever will explode/go out of business because they're getting out-competed by the better ones. Once the absolute worst go away, they're followed quickly by the 2nd worst and so on an so forth until you're only left with the a distribution that looks like system 1.
As a rather doctrinaire libertarian, I am deeply unconvinced that government action is not the ultimate cause of social media censorship. Congress shall pass no law, sure, but very little of the administrative state actually involves Congress passing laws. After Operation Choke Point I see no reason to give the government the benefit of the doubt here.
Liberals saying that corporate censorship doesn't implicate free speech are arguing in bad faith. If the tech companies were to censor liberals then they would be up in arms about the chilling effects that would hinder free exchange of ideas. In the past, liberals often used to argue that corporations are more of a threat to free speech than government; they even wanted regulations of corporations on that ground. What are the common carrier and net neutrality laws other than government regulations to prevent corporate censorship? It's intensely hypocritical for liberals to champion net neutrality laws on one hand and support corporate censorship on the other. If ISPs don't get to control what users see then why do social media companies have that power?
> The quickest way to enrage me is to criticize some group of weird people doing their own thing without harming anyone - to try to browbeat them into doing the same thing as everyone else.
The fact that you characterize Parler and HBD enthusiasts as "some group of weird people doing their own thing without harming anyone" is the biggest self-own I've ever seen on this blog.
Thanks for the great blog! This whole antifragile thing isn't making sense to me. Car companies are not antifragile. They are fragile, if they make unsafe or unfashionable cars they fail.
What you are really saying is that if I as a consumer have choice then I am pleased to have a lot of variance. I choose which car to buy, to hell with the others. So I am happy to have a lot of varience in quality of cars.
Similarly I choose whether to exercise an option so I benefit if there is a lot of volatility/variance in the underlying asset.
I don't choose which nuclear power station effects me, they could all effect me if they create a global nuclear incident. So I would like little variance in nuclear power station safety.
In summary it is not the thing which is fragile/anti-fragile. Rather when people get to choose (options, products) they prefer high variance, and when I am effected by all individual items (nuclear power stations, earthquakes) they prefer low variance.
"If you did want to argue against diversity libertarianism, you'd want to show that relevant systems are more fragile than anti-fragile..."
The most relevant system here is the most fragile: the individual human life, of which we should all be fiercely protective, not on religious grounds, but on empirical evidence showing that every individual field of consciousness is finite, defined, and unique. (We could maybe debate this on religious grounds, but our concern in the modern world is necessarily political.)
The specific brand of libertarianism one advocates is not the problem. While I don't tend to object to libertarian impulses and actions per se, it's the ego commitment that blinds one to the damage, since intellectualisms reign supreme in big and small instances. It is simply not a morally defensible position to allow lives to become collateral damage to cerebral solutions, no matter how well thought out -- sorry MarxBro. (I know, I know, collateral damage happens all the time, but I'm discussing the extent to which a political philosophy should or should not embrace it.)
This is why the libertarian ego enjoys probably the most unearned sense of facile superiority of any political ego. Solutions can seem remarkably obvious to you if you are predisposed to shrug off the immediate risks and damages as "just part of life"; in which case you become free to ridicule all the "dumb" inefficiencies that plague democratic solutions as they attempt to attend to specific human rights, social justice issues, etc. Whether one is Ayn Rand, Hitler, or Scott Alexander, libertarianism (as a philosophy) is a symptom of diseased empathy among elitists.
It's entirely too easy to rattle off rationalizations that seem self-evident to you, such as saying everyone should just learn to rely more on Consumer Reports, without, apparently, the least bit of awareness that there are masses of real-life people who have never heard of Consumers Reports, and have no access to it. The confirmation biases based on one's own stable position in society are striking and omnipresent within libertarian circles. And if they aren't unconscious biases, they are, by definition, an embrace of Rand's abhorrent survival-of-the-fittest notions.
Perhaps our systems will evolve into new shapes in the future, but for now the only morally defensible political positions are the ones that tackle the MUCH more difficult, MUCH more complex task of tweaking existing systems progressively while protecting and preserving as many real flesh-and-blood individuals as possible. To say, well, 100 real people will suffer now, but if you implement our brilliant solutions millions of unknown, unborn future people will have freer, happier lives, is unacceptable in today's modern world. Y'all have to work harder than that. Still, many here see no problem with glibly sacrificing the uninformed to the consequences of deregulation, staggeringly unprepared or unable to place themselves in others' shoes.
And, by the way, the libertarian ego, I believe, is often susceptible to the really bad habit of rattling off false equivalencies without remotely adequate interrogation. I could go on and on about this, but just one current example: The Zuckerbergs of the world did NOT decide to "censor" Parler simply because they didn't like them. (Scott seems to think they could just have easily decided to "not like" Black Lives Matter.) No. They held out as long as they possibly could before eventually being forced to acknowledge the very real damage Parler was doing with explicit, philosophically based, calls to violence and insurrection based entirely on lies. Pressuring Facebook to restrict Parler (or Trump for that matter) is parallel to pressuring the government to regulate drugs and pesticides for the protection of even those people forced to work far too many hours each day to monitor the business pages. Enjoying Scott's mind as I generally do, I am repeatedly astonished by his, frankly, very lazy habit of spreading false equivalencies without adequate education -- a dead giveaway that trapped priors are in play.
I apologize for being pedantic, but this has been bothering me for the last two days.
The difference between the two distributions shown is not the amount of variance, but rather the kurtosis. Depending on the scales of the axes, the two might have exactly the same variance (or not). However, the second one has far different kurtosis from the first. In the language of statistics, it has fat tails. It is definitely not a Normal or Bell or gaussian distribution.
The complaint of people like Taleb is that we tend to assume that life looks more like Distribution 1 when in reality many events and processes look more like Distribution 2. Indeed, my graduate student advisor would become quite angry when people spoke of the Normal distribution, arguing that it was anything but normal.
My takeaway is that the world is a lot more random than most people think it is. We need to concentrate more on courses of action to deal with occurrences out in the far tails. That's what I call antifragile, and I think that has an echo in many of Taleb's writings. In this sense, Taleb has many precursors, albeit nowhere near as well known as Taleb.
I think you’re approaching Taleb’s point, but you’re still missing a big aspect of it; from his perspective, two different bell curves are very similar because they’re both effectively finite, even if they look different. If distribution B contains plants that melt down in a few days, distribution A will contain plants that melt down in a few years. The extra few years of power pales besides the hundreds or thousands of years of reduced land useability.
Buying a car takes advantage of variation, but it’s still at best basically robust; you the individual buyer may gain a small amount once from variation, but you probably won’t get more than twice or ten times as good a car as the average or anything. And you definitely could get a worse car than average!
Antifragility *compounds* on variation. Antifragility is like gravity, or science, or evolution, or a monopoly; it starts slowly and then it becomes all-encompassing and inevitable. It contains an argument about the single iteration game, but its true essence is about the repeated game.
Instead of your "religion" and "opinions" you should rather use "important and unimportant values".
The big innovation of liberalism is that maybe you should actually try to understand these evil people (that are against your own values) : either to have an easier time to beat them, to convert them, or to find a truce with them if the two previous ones aren't worth it or are not an option.
"Right now there's religious pressure on tech companies to conform. Someone on Twitter pointed out that tech censoring Parler isn't a sign of their strength, but of their weakness. Imagine that Mark Zuckerberg decided he personally really disliked BLM, and he was going to censor BLM and any people/organizations/apps that promoted it from Facebook. Do you think he would succeed? Do you think he could stay CEO of Facebook after he was found to be doing this? Mark Zuckerberg and Big Tech in general are as much slaves to the prevailing religion as the rest of us; their "power" is the power to choose between medium vs. high levels of conformity."
You neglect simple legalities. Companies couldn't desegregate in the 1930's because there were laws enforcing segregation. Facebook shareholders could ban BLM if they wanted, and the Woke couldn't do anything about it except grumble like conservatives are doing now. The CEO has to do what the board of directors wants, which is what the shareholders want. Zuckerberg owns 29% (https://www.google.com/search?q=zuckerberg+share+of+facebook&rlz=1C1GCEJ_enUS867US867&oq=zuckerberg+share+of+facebook&aqs=chrome..69i57j46.5480j0j3&sourceid=chrome&ie=UTF-8), which is enough to control, so he can do what he wants. I conclude he wants to ban conservatives but not BLM. One caveat: if woke users of facebook would leave but conservatives will not, then Zuckerberg must worry about the dollar cost of indulging any preferences he may have against BLM. Similarly, if he has Woke employees he values, it will cost him dollars if he offends them enough that they quit. But Zuckerberg is NOT a slave to conformity unless he chooses to be. He is a leader.
Assuming there is a "best" option in your distributions, then always picking the best ultimately results in fragility. Most people would get Teslas and then distribution 2 starts contracting to look like distribution 1.
However, Teslas are more fragile than a Ford model T, for instance. Teslas are only the "best" in the current environment where we have working infrastructure that can support them. A massive solar flare at the wrong time, and the Ford model T might then be the "best".
But then accepting the anti-fragile ethos would suggest that we must spend resources to preserve and maintain a healthy population of solutions that are a poor fit to current circumstances, just in case circumstances change and we might end up needing them. This is probably a good idea to some extent, but it must have limits.
It feels to me that the way one describes something leads to "obvious" conclusions about fragility and antifragility. If Scott had described the nuclear plants as "The good one produces enough energy to power all of humanity's needs in a second, the bad one not enough for a light bulb" and the the cars as "The good one gets you wherever you want to go in 5 minutes, the bad one explodes as soon as you turn on the ignition" then we would conclude that cars are fragile and nuclear plants are antifragile.
This type of shenanigan makes me deeply suspicious that the concept of antifragility exists at all - or, at the least, that it existing doesn't say anything profound about the world that we should be taking note of.
It's a myth that libertarians don't believe in rules and order. On the contrary, more rules are needed in a free society than an authoritarian one to keep people from stepping on each other's toes. For example, this comment thread needs a decent moderation system to stay on track and be pleasant for most participants.
The real problem is moving tech from a novelty into a reliable utility. If electric companies inquired about your politics before providing power, we would still have factories powered by donkeys running in a circle and miss out on massive improvements in living standards. In the same way, tech companies need to eventually realize that their mission of running civilization is more important than virtue signalling of selective access. Only then the benefits of tech revolution will be fully realized.
Even more onerous is social networks claiming ownership of your friendship. It's one thing to delete content and ban users. It's another to not give users a realistic way to move to another service and still stay in touch. If Facebook allowed a group owner to export e-mail or phone numbers and provide it to former members, that would be a start of a more appropriate relationship between business and customers.
It's possible to have a libertarian society with companies behaving in this manner, but it's not likely to be pleasant to live in or endure for long.
I mean, I don't know that we need to coin a new term "diversity libertarianism"... to me this just seems to be, well, classical liberalism. (Or if I'm wrong about that, because I might be, there's also already the existing therm "thick libertarianism".) I mean, go read "On Liberty", right? OK, "On Liberty" isn't so much about the value of diversity in the marketplace, but still, a lot of the other ideas mentioned here are there.
This to my mind is why so much of libertarianism, that seems to focus *purely* on opposition to government, really seems, well, off-center to me. Government is only an *example* of the problem. The more general problem is conformism/traditionalism/authoritarianism/group-ism/mob-ism/whatever you want to call it, and the attendant social pressure. Again, Mill covers this! But many libertarians seem to ignore this.
(I should also note here, that when I say "authoritarianism" above, I *don't* mean overbearing government. I've commented here before that I don't like the two-axis left/right libertarian/authoritarian model of politics-space. I *definitely* don't like the left-right axis, which to my mind is a total jumble. But libertarian-authoritarian I think is also unnatural; I don't think "more government vs less government" is a good axis along which to think, and I don't think it's useful for discussing any political positions other than thin libertarianism specifically. But the reason I bring it up is because of how the libertarian movement has taken this word "authoritarian" and used it over and over to mean "overbearing government", and now when I want to use it to mean something *else* I have to explain, no I don't mean that. Of course I haven't explained what I *do* mean, but, um, I'm just going to hope it's clear from context?)
(Also on that note Scott I'm hoping you got my recent email on this subject :P )
(Also, there seem to be some varieties of libertarianism that don't really feel classically liberal at all... like, the sort that's OK with a dictator so long as that dictator is corrupt and doesn't do very much, because, see, that's small government. OK, but it's not, like, rule of law, let alone democracy...)
I don't think talking about this in terms of "religion" is helpful, because I think the fundamental problem here is, well, more fundamental than that; religions are again merely an example.
...I guess I can't conclude this comment without making a note of my idea that we should really think in terms of a broad principle of *orthogonality* underlying this sort of liberalism, which I don't think many people have noted or made explicit, but, I'm not actually about to go on about this here, so I guess I'll just note this idea rather than explaining it. :P
Anyway yes good post, except for the part where you needlessly invent new terms. :P
I had to stare at the graphs for a little bit. Am I correct that the X-axis ranges from worse to better outcomes and the Y-axis ranges from less to more likely?
> Or more generally: in an area with frequent catastrophes, where the catastrophes have externalities on people who didn’t choose them, you want to lower variance, so that nothing ever gets bad enough to produce the catastrophe.
> In an area where people can choose whatever they want, and are smart enough to choose good things rather than bad ones, you want to raise variance, so that the best thing will be very good indeed, and then everybody can choose that and bask in its goodness.
I don't think this is exactly Taleb's point. Externalities vs internalities aren't really core to (anti)-fragility. Fragile systems are composed of large correlated risks. Anti-fragile systems are composed of uncorrelated small risks. So, in the nuclear example, even if the entire world could vote on and agree to building a giant nuclear plant that posed an existential risk to civilization, it would still be fragile to do so, even though we all consented to it and so on.
The same applies to the super Tesla example. Presumably an infinite variance Tesla could be infinitely negative, and somehow destroy the world. That would make it fragile. The key point that makes something anti-fragile is that, as a system, it survives its tail events and goes on to keep improving itself.
I don't know much about this, but I believe: Google is a big smartphone company because they push Android, which is actually free and open software and there are non-Google versions of it out theme, and because of the Play store, which actually can be replaced by other stores (see F-Droid). I am not sure there's such a huge barrier here.
No, I can argue against diversity libertarianism on non-catastrophic grounds. It's not that I think that having a high variation in car companies will cause society to break down (though I don't trust the owners of car companies as far as I can throw them, so I'd rather bury them under regulation just to slow them down). It's that having a high variation in car companies places an unacceptable burden of complexity on uninteresting choices.
If all cars are more or less the same amount of crappy, then if I need a car, I go out and buy a car that's the same amount of crappy as any other car. And I will know in advance how crappy that is, because I will have seen how crappy every other car is. If that's too crappy to be worth the expense, then I simply will not buy a car.
But if some cars are awesome and others are horrific, then I stand a significant risk of ending up with a horrific one. And then libertarians come and point the finger at me, going, "hah hah! That's your own fault for not researching different car models! Free market FTW!"
But the thing is, researching different car models would have taken time out of MY LIFE. That would be the one life I'll ever get. Every moment I am spending on soul-killing grunt work is time I am not spending on doing something that makes me happy. Libertarianism is an attempt to extort me into giving up large chunks of MY LIFE by threatening me with disaster if I don't. I find that entirely unacceptable.
So much about libertarianism makes more sense if you just assume that libertarians are freaks who enjoy drudgery and boredom. I personally wish they'd just take up stamp collection or something instead of trying to reshape society to force everyone else to be as dull as them.
> If you did want to argue against diversity libertarianism, you'd want to show that relevant systems are more fragile than anti-fragile; giving people more options will usually make them worse.
Sure. A world option to get a vaccine and not to get a vaccine is strictly worse than just the former.
Damn, talking of banning, Substack has now been blocked on my corporate network. :-(
Burden of proof? Why do I have to show that the systems in question are fragile; it is the libertarians who should demonstrate that anti-fragility holds. The risks due to fragility are much higher than those due to anti-fragility!
Now that I think about it, a lot of libertarian arguments can be summarized as "here are some reasons why X is actually pretty anti-fragile".
I'd love to hear your opinion on the psychology of libertarians. To me, it's like they knee-jerk protect the notion of "inner space" which they assume is freedom.
When I read the section with the car analogy, an objection immediately came to my mind: no, we don't want the widest possible distribution - at least not in the real world, where there are always tradeoffs, like between price and reliability (and it's these tradeoffs that justify the existence of the distribution in the first place - if there were a Super-Tesla that is strictly superior to Yugos, Yugos would go extinct right away, and the distribution would shrink). I do not want cars on the road that cost 1000 bucks brand-new, but spontaneously go up in flames at inopportune moments, such as driving through a tunnel right in front of me. That's why civilized countries have regulations regarding car safety, regular inspections to extend a vehicle's license, etc. - to cut off the left end of the distribution.
With regards to other industries, similar mechanisms may apply. Do clickbaity "fake-news" websites cause damage to the information landscape of a society? Do shady online pharmacies that sell tainted, counterfeit medications undermine trust in medicine (apart from directly hurting their customers)? Can banks that sell cleverly constructed derivative instruments spread hidden risk to a degree that an unexpected downturn wipes out the entire economy, rather than one unfortunate bank? No waaaay, right? I mean, your condition "and are smart enough to choose good things rather than bad ones" sounds great, but where in the real world would that apply? How many people even have an idea what's in the processed food they eat every day? They may be sure they like the taste, but how are they supposed to be sure it doesn't cause cancer?
I mean, these are objections of the kind you mention in the second-to-last paragraph, but I don't think you take them seriously enough.
I find this unconvincing because of what happens if you take the next step of mathematical sophistication and consider non-symmetric distributions.
Imagine a distribution with a long positive tail, and a sharp cutoff at some point on the negative side (we can accomplish this by including skew). In all of the cases described, it would be the best option. You get high efficiency nuclear power plants and Teslas alike, but eliminate the risk of meltdown or, say, buying a really crappy car.
Then the theory becomes "we always want lots of good options, and some way to limit the possible harm from bad options", which seems... tautological? And it also seems like it eliminates any obvious connection with the concept of fragility.
I think this makes the assumption that innovation is going to solve a lot of global issues, or make the world a better place and add to human flourishing. Systems that work often scale in a fractal way. If you apply government style bureaucracy to a small organization like a school, it produces a public school. But if you just let every teacher do what they want because it will lead to innovation, and creates some adaptable and anti fragile schools, then you will end up with atrocious outcome for some kids, and the few good ideas often wont get reproduced because the market is not setup to reward good ideas in this context. When you have skin in the game and the stakes are high, tried and true solutions make a lot of sense.
What does diversity libertarianism look like from a policy standpoint? Classical libertarians have it easy, they can just oppose anything that increases the reach of the government. But what do you do when all the big tech companies block access to your platform? Have some sort of anti-collusion law to prevent it? Or is it more of a guiding philosophy than a political driver?
A problem with this approach is that there are more axes along which we can measure good vs. bad than you can shake a stick at. We also can't expect to measure most of them accurately. Human measurements like working conditions are hard to pin down in an objective way. Even semi-objective measurements like groundwater pollution have too many internal variables (hundreds of different chemicals and their dispersion/absorption rates) and weighting issues — Is one gram of unobtanium per litre worse than five grams of cyanide deathsulphite? Is one cubic metre of lethal drinking water worse than a thousand cubic metres of merely debilitating drinking water? Is our understanding of human biology able to assign meaningful numbers to such questions, even given the exact chemical ratios, except in some rare situation with middling quantities of ubiquitous, well-understood chemicals?
In addition, why should we expect anyone, including really competent and benevolent government officials, to solve this impossibly complex equation to improve even a single variable without incurring a net loss via one, ten, or a hundred other variables? We're stumbling backwards through the thicket, hoping that the next move will be less painful than the last.
Re: the idea of Facebook choosing to censor BLM content, it's probably worth mentioning that this actually happened. During the first wave of BLM protests, it was found that Facebook's news feed was fairly consistently burying anything to do with the movement: Facebook had designed the algorithm to deliver users the sort of content that people generally like, and people tend to like stuff like the ice bucket challenge and cute animals more than burning cities or grieving families. All this without any "religiously"-motivated intervention on the part of Zuckerberg et al; the platform has a homogenising logic all its own. These days, of course, it runs the other way, but there's still a single uniform affect across pretty much every major social media platform; they deliver content that makes people angry, because that's what the fattest slice of the bell curve seems to want. It seems (to me, at least) that markets actually produce this kind of outcome fairly consistently across all fields. Diversity would be a nice idea, but I'm really not convinced that any kind of libertarianism is the right way to get us there.
You may want to investigate or do liberalism, which has some similar ideas.
That being said, I share many of your moral intuitions here, but wonder if you might be oversanguine about the degree to which market forces and competition produce diversity. You wrote a whole essay on this involving a certain demon king!
> The quickest way to enrage me is to criticize some group of weird people doing their own thing without harming anyone
Sure but that’s basically exactly the opposite of Parler’s mission. They exist to give a platform to weird people who were banned from Twitter because they _were_ trying to harm people.
Yes it’s bad that Apple and Google have all the power here but that doesn’t change the fact that Parler is terrible and everyone is better off without them.
(And I’d say the fix here is that the government needs to do... something. The free market is not going to solve QAnon by itself any more than it solved racism by itself.)
Wait a minute. In your model aren't Apple and Google merely Company As on your wide spectrum? If you think they are *very far* from what the average person wants, or what is optimum, then they're just representing one extreme of the wide "antifragile" spectrum that's desirable, and if we wait around a little, some entrepreneur will see the underserved demand and create the Company B/"Tesla" of the social media platforms and proceed to eat the lunch of those nasty tech oligarchs.
Or to put it another way, if your guiding principle as a diversity libertarian is tolerance for behaviors that are very far from your (or any one individual's) preference -- should you not be *celebrating* the existence of firms that take extremely negative (to you) positions, on the grounds that this must imply the existence (or probable emergence) of firms that will be extremely positive?
Don't worry Scott, those aren't bell curves ;)
I think your two competing distributions take misses a key meta point: if your nuclear power plants are all "mediocre", all doing the same thing, then are they all *safe* or are they all *dangerous*? How do you know? Are you *sure*? The second distribution increases the chance that a nuclear power plant would have a meltdown. The first increase *systemic* risk of all of them melting down because somebody overlooked something and they all copied each other to be "safe". I realize this too is a bog-standard libertarian argument, but worth repeating.
I propose coining the term diversitarianism to describe diversity libertarianism as the ideologies feel very distinct to me.
I favor large redistribution and strong local government, viewpoints few would describe as particularly libertarian.
Yet I definitely would describe myself as a diversitarian, as I find large federal bureaucracies (EU, US, etc.) to be at the root of much evil.
having a distinct word for this ideology would definitely feel helpful!
In your second-to-last paragraph, you provide possible rebuttals; situations where diversity libertarianism shouldn't rule the day.
I think the case of Parler did meet some of these criteria, in a way that authentically swayed leaders (like Zuckerberg).
- Social networks are fragile as products. They are worse (the products themselves) when there are too many of them. There are natural network effects that tend towards monopolies.
- Online networks are also fragile in a social-malus way, we discovered, because if they harbor communities that rally into real-world violent actions, and the provide a sheen of acceptability that 4chan / 8chan could not
- People use social networks for perverse real-world outcomes, we realized
And finally, the government was subtly a factor. There was a lot of coverage about how, because the Democrats has just taken the senate with the Georgia elections, the social networks were trying to avoid being hauled in front of congress for allowing Parler to continue.
I agree that the key point about conformity pressure is not whether it comes from the government versus other sources. But I think you underrate regulation (including social conformity pressure).
First, it's not clear that variance reduction on a bell curve is the right way to think about it. An alternative model is that regulation can chop off the left tail of the distribution. For example, we do not in fact allow arbitrarily bad cars; is it clear that every possible car safety regulation affects both tails of the car quality curve equally? That would seem a contentious point, at minimum.
Second, bounded rationality and imperfect information. This blog arose from a community that started out exploring cognitive biases and why they're so persistent, endemic, and difficult to avoid. It's odd to then assert the power of personal choice to avoid bad outcomes. And the blog has explored ideas about the crucial role of culture and tradition to help people avoid repeating mistakes others have made (Heinrich, Chesterton, maybe Scott's metis). By what mechanism does culture do this, if not some form of conformity pressure?
Third, externalities. If we take seriously a consequentialism that weights potential future generations, limiting downside risk becomes extremely important. That should affect the balance with which we strike these variance trade-offs.
And further, I intuitively feel that our future selves deserve consequentalist weight. My present preferences and interests are not in complete alignment with those of past selves or future selves. Younger me made some choices that harmed present-me and future-me. I wish conformity pressure would have prevented him from making those choices. So poor choices have negative externalities even if they only affect the person making the choice.
Can someone give me an example of what Scott calls an "anti-state libertarian" expressing the point of view that we should not care about corporate censorship, mob attacks, or pressures towards social conformity because they do not come from the government? Because I've heard the historically ignorant argument that the First Amendment is all the free speech we need to have, thank you, but that argument doesn't seem to come from libertarians so much as from people who were skeptical of free speech to begin with.
I think I mentioned this last time but there are parallels between the subjective concept of jumping around ideas and the more formal concept of jumping around distributions using advanced monte carlo methods meant to explore bayesian probabilistic spaces thoroughly while reducing chances of getting stuck on local maxima. It's a complex subject with mind blowing math that takes you into higher dimensions inspired by physics in order for jumps to spend more time spiraling the peaks and looking around instead of hovering at the same place for too long. See https://twitter.com/betanalpha/status/1234576972132626445
I realize you're trying to make your examples simple for easy consumption, but none of the examples you showed are truly univariate. Trying to map things onto ill-defined "goodness" is very tricky. For example, I had thought price would be part of a car's "goodness", but John Schilling clearly thought it was not, and lower "goodness" cars would have lower price points, making them more affordable.
More generally, most manufacturers attempt to navigate a hugely-dimensional space (features, quality, formfactor, price point, etc.) and find a region in that space that is in high demand but remains poorly-served. For many products, you trade off one thing for another, a phenomenon a unidimensional illustration can't really illustrate.
Thinking about these things in multidimensional space may actually strengthen your argument for "diversity libertarianism", since the variance is no longer "crappy" and "good" but is instead about having more options available so that a single option is more likely to fit your unique preferences in that hugely multidimensional space.
> Economists have shown that this wasn't exactly organic - companies were afraid to give blacks good jobs, to pay them higher wages, or to sell them products they weren't supposed to have; they figured racist mobs would attack them, racist employees would desert them, or a racist government would crack down on them. I tend to think of this as a religious problem, by analogy to very religious towns where eg nobody will sell liquor even though there's no official law against it,
There was a law against it! Jim Crow was a series of laws by regimes that were incredibly hostile to corporations and business writ large. It's not a coincidence that Jim Crow happened right as 19th century progressivism took off. There was this new ethos that the state should take more control over society to make it better and for racists "better" meant less friendly to African Americans. In the absence of those laws, businesses did hire African Americans. In fact, in the presence of those laws they did hire them anyway. And that required enforcement in the form of crackdowns.
>I'm less likely to object to things like taxing the rich, redistributing wealth, or removing externalities on carbon - none of those decrease diversity very much.
This is logically inconsistent. Redistributing wealth is definitionally meant to decrease diversity in the sense you're using it here. It literally has the goal of changing the distribution from the second graph to the first.
Scott, if you haven't already, I recommend you read about Timur Kuran's theory of preference falsification. I think it solves your "religious problem". It could also explain why "people will perversely choose the worst option rather than the best if many options are made available" as you stated in your steelman paragraph. https://en.wikipedia.org/wiki/Preference_falsification
One other thought on your distribution models: I think regulation tends to push an industry from Distribution 2 to Distribution 1. Where public safety is a big concern you get an industry in Distribution 1, such as nuclear. The real challenge is how do you encourage the variance of Distribution 2 without the downside risk of Nuclear Plant A? How do you chop off the left tail? There's been a recent push from Bill Gates and others to fund innovative nuclear companies which is great but there's been stagnation in that industry since the Chernobyl disaster. Maybe venture capital is the answer but I don't think it gets us all the way there. Public support of safe an effective nuclear power is necessary. Just look at what happened when Gates announced that he was funding COVID vaccine R&D.
"Now Point B is Tesla, making revolutionary new environmentally-friendly cars. In fact, let's say it's some super-Tesla that's even better than the real Tesla"
Bopping off on a tangent here, but this is something I am curious about. Car ads over here are starting to push electric or hybrid vehicles, especially in light of the government's commitment to reducing carbon emissions and making grants and rebates for electric vehicles available.
So how does the competition stack up against Tesla? This article gives a rundown but since I know nothing about cars, is Tesla still the best or are there any potential rivals? https://www.irishtimes.com/life-and-style/motors/the-irish-times-guide-to-new-electric-cars-for-2020-and-2021-1.4285450
Dumb question: if Zuckerberg owns 50%+1 of the shares in Facebook, how could anyone ever force him to stop being CEO? I get it, maybe the stock tanks, and maybe the company loses a lot of value but if he truly decided to die on the hill of stopping BLM... what could stop him?
I liked this article, but I think it has one glaring omission: people disagreeing (in good faith or bad) about the distribution associated with any given topic. The biggest proponents of constraining Big Tech (Scott's words, but what I interpret as Big Social Media) would say that their distribution is closer to that of the power plant example, where poor implementations result in catastrophic results (polarization, acts of sedition, persistent harassment, etc.). The proponents of regulation are not promoting a different distribution directly but attacking the perception of the existing distribution of outcomes.
Taleb has said "I am, at the Fed level, libertarian; at the state level, Republican; at the local level, Democrat; and at the family and friends level, a socialist. If that saying doesn’t convince you of the fatuousness of left vs. right labels, nothing will."
Which seems related to the idea of "diversity libertarianism" - the scale has an impact on how you feel about it (holds for nuclear plant vs. car example). Even the laws and rules of a town are very different from the laws and rules of a country. It is easy to move to another town with better rules, harder to move to another state, and much harder yet to leave a country.
Regarding your point about whether people will choose the worst vs best with many options available: I'd like to believe this is true, but I'm not sure how to nest phenomenon like anti-vaxers, q-anon, etc within it. It's seems at least plausible that while for some people or in some situations more options leads to better information / outcomes that doesn't apply in all cases.
A normal distribution (bell curve) does not even approximate the quality of a thing being produced (speech, cars, whatever) under unregulated conditions. To interrogate this idea, you have to start with the acknowledgement that the distribution of quality looks something like a negative binomial distribution (lots of shit and a long tail).
"there were times and places where all companies refused good service to blacks for several hundred years."
And yet railroads had to be forced to provide separate facilities because of the cost of having unused capacity. Your point may be different for railroads because they require huge amounts of capital.
Greg Lukianoff and Jonathan Haidt talk a lot about something they call antifragility in The Coddling of the American Mind. They explain it on page 23:
> Some [things], like china teacups, are fragile: they break easily and cannot heal themselves, so you must handle them gently and keep them away from toddlers. Other things are resilient: they can withstand shocks. Parents usually give their toddlers plastic cups precisely because plastic can survive repeated falls to the floor, although the cups do not benefit from such falls. But ... some things are antifragile. Many of the important systems in our economic and political life are like our immune systems: they require stressors and challenges in order to learn, adapt, and grow. Systems that are antifragile become rigid, weak, and inefficient when nothing challenges them or pushes them to respond vigorously. He notes that muscles, bones, and children are antifragile
The idea that there is a category of things that become stronger only when stressed (but break if stressed too much) is surprisingly useful. I've used "antifragile" in this sense since I read Lukianoff and Haidt's book. But it seems like this is actually pretty different from Taleb's definition?
Is there some other word for this concept?
I don’t think discriminatory car companies works that great as an example. For one, Tesla is impressive because it’s been so long since a new car company really broke though (after a decade of being very niche), and despite massive valuation hasn’t captured that much market share. Two, I think it sorta distracts from your point - like, yes, diversity/competition in providers provides protection against discrimination, but it’s also currently illegal. Given that you compare constraints on tech favorably to car companies, is the argument that companies should be allowed to not sell to particular races if the sector has plausible entry? Doesn’t banning discrimination improve diversity of choices for black people?
"Apple and Google both blocked Parler from their phones." This is, I am pretty sure, is very wrong. You can still download Parler to a Google phone, you just can't get it through their app store.
Anti-fragility seems like the kind of argument that could be levied against almost all things, to the point that you could argue having higher variations helps select things for survival. But it's too vague, too ill defined to be of any actual use. If you have enough variations, enough downside protection and enough natural selection to survive the tail events, you're anti fragile.
For instance, with car companies you actually don't want too much variance. That's why we have laws that limit things on safety and production. Because any increase in variance is a cost to you (the consumer) to figure out what needs to be bought. If you fob that decision off to a third party reviewer, guess what, you're back in "fragile" land, since explicitly or implicitly we're back with single point of failure. The reason we don't have high variance is because we started with high variance, it sucked for a lot of people, and we quickly figured out let's reduce the variance.
"If people are able to choose freely then you wan high variance" is a very high bar that actually doesn't get applied to very many things! It's a tautological argument - if something satisfies that criterion, by definition we've kind of figured out the problems. It's why we have hundreds of cereal options, but the actual variance isn't all that high.
Even diversity libertarianism doesn't stand up to this scrutiny. We don't like the fact that several companies, seemingly independently, reached a conclusion. Sure, this is fair. It's also the outcome of a system that is seemingly anti-fragile - the companies were not coordinating or coerced. If the argument is they all had a worldview, yes they did. But that's what evolving systems do, there's some points where they agree on the ecosystem they're in.
What we have here is the evidence that seemingly anti-fragile systems (with multiple options and competition) devolve on some issues to become unified. Which is also how almost everything works. We always draw lines around how much variance is acceptable. That's most of what our moral evolution has done for us. That's actual civilisation. We can discuss the genius of Swiss Cantons and Renaissance Italy all we want, but the reason most governance systems aren't like that is because those systems are incredibly fragile - to immigration, to cultural cohesion, to coordination failures and more. The US has relatively homogenous 50 quite independent states and it still can barely get anything done in this regard.
To me this is Taleb's genius and his downfall. He's like a poet, creating beautiful symphonies with words and allusions and history. Together it makes you feel as if you can almost reach something ineffable, a point of insight that explains the world. But turns out the inspiration is the actual point, and not an actual theory that you get from the books.
This is similar to a point I've mad about libertarian YIMBYism: There's a question of what the appropriate libertarian reaction is to "state prevents local government from preventing property owners from building more apartments on their land". From the diversity libertarian perspective though, the answer is obviously the YIMBY answer, since it gives us more diversity of land use and density types and gives people more options of what type of area to move to.
> The evangelical Christian town where nobody will tolerate gay people may not have laws against homosexuality, but gay people still won't be able to find a church, community, or business that meets their needs.
I think you've fallen into a trap, Scott. The whole point of the archipelago is that a community can run itself like this if it wants, as long as gay people who are unhappy with it can go somewhere they prefer. Thus a thousand flowers bloom. If your newly coined diversity libertarianism prohibits this, it de facto *flattens* the space of possible arrangements. And there is no telling how much more flattening consistency requires. Remember that one of your original examples was Aquinastopia, where Catholics could go to live in a Catholic community. Is this disallowed now? Does your libertarianism suddenly prohibit Catholics from living as Catholics?
Don't confuse what happens at the level of the evangelical town from what happens at the level of the archipelago. *Everybody in the archipelago* benefits from the fact that this evangelical town (or Aquinastopia, for that matter) exists. The people who want to live there can live there, the people who don't don't have to, and the people who aren't sure can learn one way or the other whether such a community works.
Granting all that, I don't know exactly how to apply this lesson to Big Tech, because we don't actually live in the archipelago. But I think it suggests that size matters. If a tiny company wants to run itself according to religious principles, that's fine, good even; there is a lot of room in the economy for other companies to run themselves differently, and perhaps outcompete it. But if a huge company with near-monopoly power in some industry or another wants to run itself according to religious principles, you have reason as a diversity libertarian to be concerned.
On another point:
> Right now there's religious pressure on tech companies to conform. Someone on Twitter pointed out that tech censoring Parler isn't a sign of their strength, but of their weakness. Imagine that Mark Zuckerberg decided he personally really disliked BLM, and he was going to censor BLM and any people/organizations/apps that promoted it from Facebook. Do you think he would succeed? Do you think he could stay CEO of Facebook after he was found to be doing this? Mark Zuckerberg and Big Tech in general are as much slaves to the prevailing religion as the rest of us; their "power" is the power to choose between medium vs. high levels of conformity.
I made a similar observation on your Erdogan post. Quoting myself:
> Erdogan's purges look to me like a particularly poignant subversion of this: hollowing out every other institution, until the only thing left is him and his party. It should really, really concern us if seemingly every semi-official institution in our country, from the top schools to the top newspapers to the top thinktanks, march in nearly complete uniformity: it raises the question whether they, and the people in them, can act of their own will at all.
THIS "Even beyond politics, I think this predicts a sort of ethos of what I like and what I don't like. The quickest way to enrage me is to criticize some group of weird people doing their own thing without harming anyone - to try to browbeat them into doing the same thing as everyone else. That group might be just a tiny bit of slack away from creating something amazing!"
This argument seems weaker than it should be due to using dumb examples. To take the less controversial one, car companies are not actually easy to create.
Yes, it’s true that many industrial nations have a car company, and some have more than one. They don’t seem to be nearly as difficult as building a competitive aerospace industry, which China has been trying to do for quite some time with little success so far.
But in the US, there were no new car companies for many decades. Tesla had a lot of near-death experiences along the way, and part of their survival strategy was selling luxury cars that most people can’t afford. Crappy 1920’s style cars would today be considered unsafe at any speed and the idea that a car company is something a few people can build in a garage is just nonsense. (Unless we are talking about grey-market golf-cart style electric vehicles, which are apparently a thing in China.)
By contrast, building a crappy cell phone seems much easier, at least in China where there seem to be many manufacturers building them using a common ecosystem of standard parts. Maybe that would have been a better example?
I'm not sure your car example checks out. In the first world cars are heavily regulated, perhaps even more than nuclear plants. The only reason anybody bothers to make cars at all is that the demand for them is so high that almost any regulatory burden is worth the effort.
On the other hand, you might be amused to find out about the unregulated electric golf cart in China. If the car is small, slow, cheap and electric, the externalities are minimal and government feels less pressure to regulate them.
https://restofworld.org/2021/tesla-vs-tiny-cars/
This is an interesting framing! Game theory (or, well, math) can outline some situations where you would want to migrate your preferences between systems that are fragile and antifragile.
Generally if you're losing, you want high variance systems.
If you're winning, you want the status quo to sit like a diamond, more lock in is better.
So, Tim Wu's The Master Switch gives a history of communication companies all following this similar trajectory: scrappy startups finding workarounds that change all the rules and disrupt everything, then becoming large institutional players that want greater and greater regulation to prevent anyone else from inventing anything that would ever change the world in any way.
Do you want to roll a d20 or 3d6 to hit? Depends whether you're fighting kobolds or dragons. (There are even situations, in gambling and in life, where you would even want to take a lower "average" return, in order to shift the variance.)
So it's possible to entertain a meta-political perspective, where you think society should wander between libertarian (or, "50 labs federalism") vs. more central standardization, based on how well its currently meeting certain markers you've laid out for the health of society or the effectiveness of policy or whatever.
I don't know how you can make arguments like this without getting into object-level illustrative examples, but as soon as you do that, your argument becomes subject to the object-level examples being wrong. It is absolutely not true that market entry for new automakers is easier than new phone makers. Parler is accessible from a browser. Blocking them from native app stores doesn't block anyone's access. I have no idea if there even is a native app for Substack, yet none of us are blocked from reading your blog. What actually took Parler offline (for what, a week?) was Amazon Web Services turning off their servers. But even there, as coming back online showed, AWS isn't the only provider of managed infrastructure, and you can always self-host if literally no one will sell you service.
Until such time as a peer to peer radio network with TCP/IP capabilities exists, the only true monopoly service that can cut you off from the web are backbone ISP providers. If you can't hook into them, you can't hook into the network, whether your servers are running or not. And it stands out that a whole lot of people calling themselves libertarians have complained quite loudly about Google and Apple shutting Parler out of their app stores, but not about net neutrality, which actually addresses the only true monopoly providers controlling access to the web. It makes it seem more partisan than ideological.
Of course, I don't think you personally are partisan. You're an ideologue. But an inherent issue with trying to claim allegiance to any sort of named ideology is you immediately associate yourself with other people using the same name. "Diversity libertarianism" doesn't seem to me like it really captures what you're trying to say here. You're just arguing for classic welfare economics on the merits of the Pareto optimality achieved under conditions of pure competition. You're arguing against the existence of monopolies, whether those be warlords, mafias, governments, big businesses, religions, cultures, it doesn't matter.
I don't have any answer here on what you should call yourself, but you have to realize that insisting on "libertarian" even if you tag "diversity" in front of it alienates and marginalizes you from a lot of people who are going to have a knee jerk bad reaction to that term because of who else uses it. I can understand wanting to reclaim it. That seems to have worked for the word queer. I think it probably hasn't for slut. But is that really your goal? Do you really care what people call you? Why not just argue for specific positions without needing to identify and name an underlying guiding principle that may or may not really be there? This seems to be Taleb's problem, too. There are undoubtedly some things he is very right about, but as soon as he tries to go from those examples to generalize to a grand theory of everything, it doesn't work and he becomes trivially wrong about most of them. If your position is some regulations are good, some are bad, and the guiding principle is whether or not they promote or prevent monopolies, call yourself an "anti-monopolist." What does Matt Stoller call himself? Unfortunately, probably libertarian, based on the name of his organization.
I'm not even sure "against monopolies" can stay ideologically pure and coherent. How do you prevent monopolies from forming without a monopoly on law enforcement? I'm kind of with John Schilling on this. David Friedman can probably give some elegant sounding theory backed up with an example from 12th century Iceland being the only working implementation, but that doesn't leave me super confident it can really widely work.
"But other fields have higher entry barriers than cars do. Apple and Google both blocked Parler from their phones. But these are the only two major smartphone companies. "
Forgive me if this has been posted previously, but there are lots of smartphone companies. There are, however, only two major operating systems for smartphones, and most of the smartphone companies use one or the other.
Equating Parler and BLM, as this piece does, is beyond ludicrous, and reveals a huge blind spot.
Taleb isn't a particularly clear writer, so that's probably why it's hard to grasp the concept of antifragility (it's also stupidity named, but that is neither here nor there). The basic variable is the response to unpredictable stress. Fragile systems are weakened by it, robust systems are neither weakened bit strengthened by it, and antifragile systems are strengthened by it. Exercise is antifragile, because its best results occur when workouts are irregular and non-repetitive. Option trading is antifragile because it only really lucrative when you bet on ludicrous occurrences. It's not so much that the range of outcomes is highly variable, so much as that they aren't even calculable. The point isn't so much that "we" (as in humans generally) need to find a more accurate way to assess risk, as much as "we" need to realize that risk can't be calculated for everything.
"smart enough to choose good things rather than bad ones"
I think this falls into a mistake that Scott often makes - it overstates the importance of intelligence and understates the importance of information and of the cultural guidance to understand that you have a choice and on how to make the choice.
On an information example, the choice of medical providers. Most people are not medically qualified to tell which providers are good and which bad - low variance is good if you really can't tell a good doctor from a charlatan. Moreover, a lot of people default to using price to judge, assuming that more expensive medical choices are better. Even if the information is available to know which doctors are better than others (and the medical profession works hard to keep that information unavailable), lots of people don't know how to evaluate it.
On a cultural guidance example, the choice of colleges. Middle-class parents are much more likely to have opinions on which colleges are better than others, and those opinions are much more likely to be correct than working-class parents [I'm using class in the class-as-culture sense, not the class-as-economics sense]. Also, they're more likely to know about how to get scholarships, how to apply for and get income-based discounts to tuition, and so on. Knowing that you won't have to pay sticker price because you will get need-based aid and scholarships and so on is information that many working-class potential students do not have, resulting in them being put off by a price that they won't themselves have to pay.
I think you may be taking the premises of your argument for granted. The classic libertarian arguments ignore structural impediments to optimal (or rational) behavior. For example, the ideal of hearing as many opinions as possible so as to maximize your chance of hearing the true one seriously ignores reality. In reality, there are time constraints (e.g., does a single mother working two jobs have time to sufficiently read and evaluate every argument for and against gun control?), limitations on an individual's ability to process and filter information correctly, and demagogues that prey on human nature weaknesses (e.g., Peoples Temple 1978). Also, not every opinion is created equal. Is a plumber equally suited to evaluate the efficacy of a drug from experimental data as a research scientist? Qualifications matter. As do motivations - if two scientists reviewed the efficacy of a new drug and one was being paid by the pharmaceutical sponsor, would that impact your judgement of her credibility? If no forced disclosure, you wouldn't know. I'm not convinced by the premise that people are smart enough to choose good things rather than bad ones. Humans are flawed. Society is flawed. The way we govern society should reflect the flaws.
And then we move to free market fundamentalism, which you acknowledge in the last paragraph is flawed. In almost every important sector, there are hugely important externalities. That is not an argument for universal regulations and regulations should always be thoughtful and, to the extent possible, tested in order to avoid unintended consequences. But it does recognize that humans are inherently self-interested and short sighted. Oil magnates care about their economics, not the environment. The Sacklers cared about their fortune, not the risk of addiction. Volkswagen cared about their quarterly earnings, not the Clean Air Act. Zuckerberg cares about world domination (?), not the risk of facilitating genocides. Regulations need to reshape incentives to harness self-interest but also protect society from our inability to process risk and large scale ruin in addition to our bias of discounting the future.
The example of “nuclear plant” as antifragile bothers me but I’m having trouble nailing down why.
A single instance of tech is often fragile. One car crash does not make that car stronger. It can’t reorganize itself and sometimes it can’t economically be rebuilt individually (“totaled”). Many car crashes, when abstracted into data, can be used by companies to improve the fragile product and in this sense a company is more fragile or more antifragile (whether in service of better safety reputation attracting more buyers, or compliance with a regulatory environment) depending on company culture and how they leverage the data into better product.
Any single instance of tech is fragile, so any single nuclear plant is also fragile. Nuclear plants with the same design flaw don’t all blow up at once, either, when they are individually distant and subject to different circumstances.
However the nuclear engineering culture has to be antifragile; when an “accident” occurs, they scramble to learn how to more successfully create a safer plant. It has to learn from mistakes and respond to stress. From what I understand there were times historically when a potentially safer design was not chosen for political reasons; nuclear regulatory culture has to be less fragile than some regulatory cultures because they are trying to balance tech advancement, cost, and safety in a context of large externalities, and they have more actual influence (I think) on next-generation plant design than say the DOT has on the design of exhaust filters for idling trucks. It is a government-industry partnership in the US in ways other industries are not like car and truck companies.
The regulatory culture is fragile to financial pressure but antifragile to nuclear accidents. “The safer design is too expensive” is a barrier. But the pool of players that benefit from a functioning nuclear industry is large enough that the military, academia and national labs all get in on making the safer design cheaper, in an antifragile way.
In other words I don’t think it’s accurate to say the nuclear industry is like company A. For it to exist at all there is a lot of this industry-government design pressure to reject dangerous designs before they are ever built, and then learn as much as possible from accidents. I’d call it antifragile with a floor. If the industry was fragile they’d have given up and gone home after the zillionth unintentional radiation release. Any design process is allowing some variance and trying to capture the top end of it.
I’m trying to come up with an example of what is fragile but just got interrupted. I’m not saying nothing is a fragile system just that nuclear plants/power is not the best example.
Typo: "Tesla did something like this when Musk though the big dogs were neglecting electric, and that went great." I think "though" should have been "thought".
> it's won Power Plant Magazine's Reactor Of The Year award five times in a row.
I wanted to see if this is a real thing.
In fact, this is fake news. The magazine is called POWER, and the award is called "Plant of the year."
https://www.powermag.com/awards/plant-of-the-year/
Scott, I love the clarity of laying this out with distributions and proper dissociation between different strains of libertarianism, but what I don't understand here is your position on the dynamic state of distribution 2.
You say distribution 2 is good in cases where everyone can and does make good choices, because everyone will buy from B and A will go out of business.
But at that point, distribution 2 no longer exists; everything on the left side of the graph is gone, and the reality looks like distribution 1 again!
(with a higher average value, thankfully)
So, my question is: as a diversity libertarian, do you actually want to be *living in* a world that looks like distribution 2, or do you want that distribution to have existed *at some time in the past* but be gone now?
If something that looks like distribution 2 has endured for a few decades or so, isn't that strong evidence that people in reality *don't* have either the ability or competence to make good choice in that space, since if they did all the As would be gone already and things would have converged on a distribution 2 made of all the Bs?
This gets to the point about Parler - in what way is Parler not just an A that the community recognized was bad and abandoned? Sure, there's the weirdness about boycotting it individually vs. boycotting its corporate partners, but in either case it seems like the majority of consumers decided it was awful and wanted it abolished, which is what you should expect to happen to As in distribution 2.
It seems to me like 'diversity libertarianism' as you describe it is too time-independent; it imagines a single point in time where you have a lot of options, and a single instance of decision-making where you choose the best option. In that singl instance, yese, you want tohe world to look like distribution 2.
But run that dynamic over time and measure the world at the limit, and it looks more like a world where all the bad choices are ignored and go extinct and only the good choices survive. Which looks like distribution 1, and possibly looks like 'ideological conformity.'
Given that, it feels to me like diversity libertarianism should be really really focused on making sure that *new* ideas have a chance to be generated and survive long enough to be fairly evaluated and take on their proper market share, and a lot less focused on old ideas vanishing when the market turns against them and moves on.
Ie, moving as rapidly as possible from distribution 2 to distribution 3 to distribution 4, rather than protecting the As from distribution 2 who cry that they're being canceled.
I don't think Tesla being started by Elon Musk, already a billionaire at the time, is very much at all like some guy creating a huge company by starting with a few crappy cars put together in his garage. Without getting in to all the regulatory and other barriers to entry that exist in car manufacturing, isn't plain old capital ultimately the biggest barrier to entry in pretty much everything?
People seem to be getting stuck on the "l" word a lot. The more important point is that large scale monopolies dominating an ecosystem reduce the variance that leads to positive innovations that improve human lives.
Insofar as most of the world is "governed" by legacy institutions ruled by stationary bandits (aka kleptocracies), the opportunity cost of not allowing "a thousand nations bloom" via new jurisdictions is immense. Hong Kong and Singapore showed the value of economic freedom more effectively than did any arguments, China copied them in designing their SEZs, and a billion people saw a 10x increase in their wages over twenty years, arguably one of the greatest moral achievements in history.
We don't need to agree or even know exactly what leads to success in an ecosystem open to innovation. As the Seasteaders say, "Don't argue, build." But the more we support an ecosystem of innovation with respect to the creation of new jurisdictions on land, the more rapidly we're likely to discover new and better forms of law and governance. Competition among Greek city states led to Athens; competition among medieval Italian city states led to the Renaissance; competition among the "Free Cities" of the Hanseatic League led to modern commercial society via the success of Holland, which in turn played a key role in the rise of British and American prosperity. More broadly, as David Landes has argued, Europe's fragmentation was key to its prosperity, including the fragmentation that allowed persecuted thinkers to move from jurisdiction to jurisdiction.
By contrast, large scale empires typically stagnated over time. Modern prosperity is thanks to competing jurisdictions, a sine qua non of the discovery process of the institutions required for prosperity.
Should we argue over the "l" word or get busy creating new and better institutions for human flourishing?
That said, although I'm also in favor of internalizing externalities and supporting the poor (and thus could pass as a "liberal"), many of those who have focused on the need for liberating innovators have at least some familiarity with Hayek and others who emphasize the stultifying effect of the state. Thus I'd rather be considered a bleeding heart libertarian than be part of the "progressive" tribe that rarely focuses on the morally urgent need for permissionless innovation (McCloskey).
Q: Why does a Yugo have a rear-window defroster?
A: Keeps your hands warm while you're pushing it.
From the outside view, increased variance increases both the high and the low ends of the range of possibilities. However I think an unfortunately common inside view, mapped onto this framework, goes as follows: my politics/tastes/whatever are basically Correct, and so variance is entirely bad: the high end is a fixed point (at me), so increased variance just cashes out as more people being increasingly Wrong.
The other place where the libertarian argument falls down is when time is critical. Sure, if social media companies censor certain info, you can start your own and migrate people there (if the AWS/Apple/Google thing doesn't happen). However, if the censored information happens to be highly pertinent to an election happening in a month, good luck
If the low variance scenario results from everyone happening to agree on a narrow range of desirable approaches, and everyone further agrees that exploration outside A and B will not be legally or socially punished, that seems fine. Obviously this doesn’t mean that nothing should ever be legally or socially punished. So what determines what we can prohibit legitimately?
Maybe the post was inspired by Taleb, but clearly the basic issue is different, so speaking in terms of antifragility confuses matters.
Quick counterintuitive thought: A narrow distribution of ideas may be beneficial in the public sphere. Now reflexively, I'm inclined to like the marketplace-of-ideas theory of discourse. I like markets, I like choices, I like diversity and I love ideas. So I'm offering this argument with a big grain of salt and perhaps against my own interests:
The "mainstream" of actionable public ideologies is akin to nuclear reactors: You want a narrow, safe range of variation more than you want a high-variance set that might include world-saving and world-breaking ideas. (Note: In practice, the resilience of the "world" at stake would matter here. High-resilience polities could brook more ideological variation.) Hence recent gatekeepers like publications, political parties, legislatures, universities. Before that, nobility, royal courts, hierarchical religions. And always social structures that reward and punish. Specific claim: In a population where X percent of the population believes in wrong-but-arguably-harmless ideas (angels or UFO? Your mileage will vary), an equal percentage of people is ready to believe wrong-and-potentially-catastrophic ideas (fill in your own blank here -- I'll put the crayon in your hand, but you have to draw your own monster). And maybe exposure to those ideas creates the potential for greater adoption. In such a case, gatekeepers that limit public discourse to "what the decent people think" might be performing a service, even if they limit exposure to fantastic new ideas. (While still allowing for hypothetical mechanisms for evolutionary fit ideas to grow in acceptance and no-longer-adaptive mainstream ideas to whither and die?)
Prediction: If this is true, societies/cultures/groups with greater stability (in terms of less vulnerability to ideological-triggered disaster?) brook more dissent/wrongthink.
The odds of this being true: I find this line of thought more interesting than persuasive. I'll admit to some nostalgia for a world without comment sections. But I can also think of ways that world was less good than this one. I do think we're in the long process of adjusting to the consequences of rapid advances in communication technology, but I need to be humble about my ability to assess what that means and where the process goes, and even less confident I can easily distinguish what's really new from what has always been thus.
"But other fields have higher entry barriers than cars do."
Er, I think car manufacturing has one of the highest barriers to entry of all consumer products? There's a reason that Tesla is the only successful new American car company to start in the last 50+ years.
> If Ford refuses to sell cars to black people, Toyota should see a profit opportunity and step in. If both Ford and Toyota ban blacks at the same time, some upstart like Tesla should step in. If Ford, Toyota, and Tesla all do it, some guy with a wrench who's always dreamed of making cars in his garage should notice a billion-dollar business opportunity lying on the ground, get seed capital from equally greedy investors, and solve the problem.
I don't think this actually works very well in practice. Suppose the hypothetical "racist car companies" world existed, and you wanted to start NRC, the Not Racist Car company. First, you have a hard time raising money from investors. It's not impossible, but in this hypothetical world the most successful car companies are super racist, and investors are skeptical that doing literally the opposite of every successful car company is going to work. But you find some and are able to raise some capital at extortionate terms. If everything goes well your company is going to make a zillion dollars even though your cost of capital is nuts.
Unfortunately, that's going to take years to get going, even if you're successful. In the meantime, minorities have no access to cars. Even after you start producing, for years or even decades, minorities are forced to Distribution 1 while everyone else has access to Distribution 2. When people talk about "privilege" this is the kind of thing they're talking about.
Now cars are pretty important. Not having a car, or having a crappy car, has lots of knock-on effects. Minorities basically have to live in major urban areas, while non-minorities can live mostly wherever they would like. This limits the jobs available to minorities, which in turn messes with the labor market. So now even equally qualified minorities will make less than non-minorities.
But can they become equally qualified? Not if people want do some extra racism on the sly. Just put your best schools, public services, and businesses far away from public transportation (Hi, Robert Moses!)
At the margin, some people who might have barely gotten a job now have no job at all. Some of those people go all Jean Valjean on you and your loaf of bread, and now your prisons are full of minorities, which reinforces Ford's & Toyota's position.
And of course this has long-range knock-on effects. The Car Inequality is going to take generations before it's dissipated.
My point here is that we can't rely on Diversity Libertarianism in cases like this because most negative externalities need to be regulated, and racism causes massive negative externalities.
Taleb doesn’t hate curves that look sort of “bell-shaped” like the ones you drew, which usually have “fat tails”. He hates GAUSSIAN curves. If you understand the difference between “power law” and “Gaussian”, you get to review Taleb’s work and be taken seriously, but if you make a mistake like this it doesn’t bode well.
The situations you describe are ones where a large majority of the population believe in the religion, in which case a democratic government will be helping them suppress the atheists not stopping them from doing it, as will almost any other form of government. So if you are a diversity libertarian I don't think there is a solution to your problem, except in the special case where the state is 99% fundamentalist (or racist or whatever) and the federal government isn't even majority fundamentalist.
So even in the situation you describe, I am against federal regulation of FB et. al. purported to keep them from censoring. You will note that all the supposed rules against racial discrimination got ignored when the discrimination took a form, affirmative action, that was politically popular.
The problem is proverbial — setting the fox to guard the hen house. The government has a stronger incentive to suppress (some) political speech than FB. FB is only one of several large (and many small) players in its market, so can only suppress speech to the extent that other players agree. The government doesn't have that limitation — it can regulate all of them.
I think the term "diversity libertarianism" makes sense as a type of libertarianism, but I think that the complexity argument for diversity is strictly superior to the libertarian argument, as it handles all the cases the libertarian argument does, and more.
John Stuart Mill's argument for free speech was that many unpopular opinions might contain elements of truth, and hence truth can be assembled from bits and pieces of many opinions. The complexity argument for diversity is that it takes many different kinds of pieces to make a complex machine. To build complex social mechanisms, we'd like many different units of any kind--opinions, ethnicities, personality types, and educational backgrounds.
In its original historical context, libertarianism effectively opposed ancient Platonist tradition, which says that there is one perfect, eternal model for everything, including one perfect opinion to hold on any subject, which is True in all times and places; and the job of society is to find that one perfect opinion and shove it down everybody's throat and forbid anybody from ever questioning it.
Locke used empiricist epistemology to argue (among other reasons) that we can't ever be sure that what we think is the one Truth really is the one Truth, so we better listen to all opinions. But (AFAIK) he didn't challenge the idea that there is just one Truth.
Mill's argument for freedom of speech doesn't convince people who are /really sure/ that they do, in fact, know The Truth. The complexity argument for diversity gives a different view of the same issue--not that we want more opinions for society to choose from, but that we want people with different opinions to remain in the population indefinitely, because this difference in opinion will produce division of labor, and a discrepancy in values which makes it more possible for everyone to get more of what they personally value most. It rejects the very notion of converging on a final Truth.
This post starts out riffing on the tradeoffs between risk and benefit variances and then pivots to the issue monopoly vs. "diversity" in the free market.
The first part just reflects the fact that an optimal balance of risk and return depends on the relative distribution of the various outcomes (as best you can forecast), multiplied by the costs and benefits associated with those outcomes. Like everything in economics, you are optimizing when the marginal benefits are at least equal to the marginal costs. (i.e., MB = MC).
Everything that follows on the risk-benefit topic is mostly a recognition that, even assuming arguendo that you can accurately predict the probabilities of various outcomes, different people will still have very different optimizing points (MB = MC) due to: (a) being differently impacted by the costs and benefits of particular outcomes (e.g. someone living close to an accident-prone plant, vs. far away); and (b) having different tolerances for risk (e.g., someone who can sleep like a baby knowing there is a 1/10,000 chance of a plant explosion vs. someone who can't stop worrying about it).
Bottom line: The theory of making decisions under uncertainty is uncomplicated. But ascertaining the true probabilities of outcomes, and then politically mediating the interests of lots and lots of people with different interests and preferences is pretty impossibly complicated.
I like this description. Mostly I just err on the side of freedom. This generally results in me being anti-regulation, but I also don't like when monopolies restrict speech. It's the same root principle, but people think there's some deep contradiction there.
Also the block being a result of big tech weakness is a good point not made often enough.
There is nothing inherently good about "Diversity." It's only potentially a good thing if: (a) the "diverse" things have different merits; and (b) people are allowed to judge which diverse options have the most merit and reject the others. Once this process has run its course, whatever is left over is the optimal amount of "diversity."
For example, having lots of restaurants creates a "diversity" of cuisines to choose from. That's great if you are allowed to choose to patronize the ones you like (say, Thai and Italian), and skip the ones you don't (say, the "House of Haggis" Scottish restaurant). The modern conception of "diversity" claims, however, that forcing people to eat Haggis enhances the "diversity" of food and is therefore improving them by definition.
Only when all the bad restaurants have been driven out of business is the remaining "diversity" of choice a "strength."
>(If you did want to argue against diversity libertarianism, you'd want to show that relevant systems are more fragile than anti-fragile; giving people more options will usually make them worse. Maybe the slightest failure could cause catastrophe - I think free speech opponents think this is true, but I’m less sure they’re right. Or people will perversely choose the worst option rather than the best if many options are made available - I think this is sometimes true for things like drugs and gambling. or there are overwhelming externalities. Or there are externalities, which can range from the very simple like pollution to the very complicated, like whether your working for a low wage has externalities because it forces me to compete with you. I think figuring out where to draw the lines here is really hard - but if you want to convince me, this would be more fruitful than the umpteenth essay about how the First Amendment only applies to government)
I think that one can object to absolute freedom of speech on the grounds not that people would be somehow "brainwashed" if exposed to ideas outside a particular approved band, but rather through understanding certain types of speech as primarily acts and the majority of people as basically unthinking followers of a particular flag. That is, someone like Richard Spencer is not contributing to a marketplace of ideas so much as raising a flag for people who already agree with him to rally around as a coordination point.
I also think it's somewhat of an errant frame to suggest that you can *only* adjust volatility-- like, in the car market example, it is not in fact the case that in order for there to be better cars there must be worse cars. The distribution of cars on the market on a scale of goodness to badness could, like, /move towards being gooder as a whole/. In relation to specifically cultural freedom of speech -- that is, /not/ mediated through large corporations but at the level of, say, a bunch of individuals denouncing someone for their speech --, I don't think it's a problem for bad ideas to become harder to access or for the "sellers" of bad ideas to go bankrupt or acquire correspondingly bad reputations.
Freedom of speech is important descendant from a general orientation towards maximizing eudaemonia and the ability of people to make real choices, and needs to be evaluted within that context, not as a separate magisterium.
>When might this not work? First, if cars are hard, and starting a new car company takes too much time. I'm not too concerned about this one. I think a lot of people could potentially make one or two crappy 1920s-style cars in their garage with a little work, and once they do that, black people with no other options will buy them, and that will give them enough money to bootstrap into a powerful car company that can compete with the big dogs. Tesla did something like this when Musk though the big dogs were neglecting electric, and that went great.
This is completely unrelated to the last thing I said as far as I can tell, but I think stuff like opening up a new crappy car company in your garage is likely to be harder than it ~ought to be in an ideally free market due to government regulations on car companies.
Sorry but this is a major theme running through ALL of Taleb’s books including this one. The canonical “fat tailed” distribution is the Pareto or “power law” distribution, and the canonical “thin tailed” distribution is the Gaussian or “normal” distribution, and treating the former as if it were the latter is what causes all the blowups and disasters he is trying to teach how to prevent.
This is the pons asinorum for understanding Taleb in any way other than impressionistically.
The trouble with corporate censorship is that "uncensored" forums tend to be on the "cars that immediately catch fire" side of the curve - moderation is a necessary part of building a forum that people actually want to use. Parler has already learned this the hard way - their policy rapidly shifted from "we're the good guys because we don't censor people" to "we're the good guys because we censor left-wing trolls, while everyone else censors right-wing trolls." Even under a "diversity libertarian" regime, you still have to be able to make money off your idea.
Also, factual correction: Apple and Google blocked Parler from their *app stores*, not their phones. You can install Parler on Android by downloading the .apk file and installing it manually, no need to go looking for another cell phone company.
If you're in favour of viewpoint diversity why have you avoided all socialist argumentation and even go so far as to ban socialists from your blog?
If the profoundness of your articles was plotted it would probably be a bell curve. I'm not quite sure what would be around the mean, but "I can tolerant everyone except the outgroups" would be at least two sigmas above the mean. This article is on the long tail of the less insightful end.
Diversity libertarianism feels like a very strange concept boundary to me. Isn't it basically just capitalism, where you want lots of diversity and competition, but you also want to prevent companies or people from preventing competition?
How does libertarianism come into the picture? Is it just capitalism but you don't want government solving coordination problems because you want the free market to instead?
So what do you think then about the current deconstruction of Republicans? A single *major* party is left: this will eventually kill the competition for the "consumer of policies" (the voter). As in your example, when the super-car wins the competition created by the variance, it likely will just reap the profits and deteriorate on services and features. It will likely fill the niche of the bad car, not eliminate this niche. Because the elites are given the opportunity to think short-term but large rewards for themselves from such a process (with negligible negative consequences). Yet no better system is in evidence.
<< I'm less likely to object to things like taxing the rich, redistributing wealth, or removing externalities on carbon - none of those decrease diversity very much. And I’m more likely to care about conformist pressures from religions or mobs, even though technically those don’t involve government.
But, having such preferences, you are more likely to be left-libertarian: anarchist, socialist, etc., than right-libertarian, libertarian in the usual sense (as in the "Libertarian Party")
You seem to be treating “diversity” and “variance” as synonyms, where “variance” means a distribution along some axis of objective quality. But these are not actually synonyms, and it is important that they aren’t. Diversity just means “differences exist” and those differences may or may not result in variable objective quality. But it can still be valuable even if it doesn’t - people have preferences, some rational and others not, but the preferences are definitely diverse. Diversity, insofar as it meets a broader range of preferences, can therefore have utility even if it does not result in higher objective quality.
When Ford first made the Model T, you could have any color you wanted as long as it was black. But many people prefer cars in other colors, so eventually car companies started offering a range of color options. Offering cars in multiple colors does not improve objective quality in any way (if anything it slightly degrades it since now you’ve thrown an extra variable into the supply chain, plus there is probably some objectively best car color for most situations and people will nevertheless make different choices). So “car color” is a place where diversity is valuable even though it’s not really “variance” in the sense you use it, or the sense that creates anti-fragility.
People having irrational preferences may actually be important to keeping things diverse enough to produce useful variance. Many people have irrational brand loyalties - you might think this is bad, but if they didn’t, as soon as Ford trucks got a little objectively better than Dodge and Chevy trucks, 100% of people would buy Ford trucks, Dodge and Chevy would go out of business, and you’d be left with no diversity or variance. Irrational (maybe “arational” would be a better word) preferences add some friction and hysteresis to the system in a diversity-preserving way.
I think your lack of distinguishing between diversity and variance leaves something missing from your assessment of diversity libertarianism. Namely, libertarians believe that individual freedom is a terminal value. It’s a good thing even if it doesn’t make the world objectively better on average. Libertarians tend to believe that freedom WILL lead to better outcomes on average, but that’s not necessarily the point.
And this is partially why pragmatic libertarians might sometimes accept government intervention - if a coordination problem is limiting individual freedom, it may be better to let the government impose coordination by fiat than let the problem go unsolved. And why they might be skeptical of corporate censorship - sure, it might be a free choice by a corporation, but if the corporation has monopoly powers then their censorship is a significant imposition on individual liberty.
> Plant A suffers a meltdown after two days, killing everybody.
> a lot of people could potentially make one or two crappy 1920s-style cars in their garage [to sell]
Scott, do you have somebody review your pieces to make obvious objections before you publish?
The obvious objection to both of these points, especially the first, is that regulations won't allow these things. So long as we have an effective safety regulator, you absolutely want a high-variance nuclear industry, so that alongside the TEPCOs and V.C. Summers of the world, you have a Moltex/Thorcon/Terrestrial Energy/Elysium, one of whom is likely to be involved with building large numbers of reactors at some point.
P.S. Note my use of the word "regulator" on top of "regulations": you need somebody to notice that every single one of the emergency generators and batteries at Fukushima are located in the basement and ask themselves "should we allow this?" But see also: Meltdown World https://www.reddit.com/r/nuclear/comments/jtm6hm/how_bad_is_meltdown_world
"Diversity libertarianism" seems like the kind of philosophy of government that allows you to construct a justification for creating whatever kind of policy you want. Which is to say, a philosophy with actual legs.
For nuclear plants, when everything is fine, yes, we prefer the low variance distribution.
However! In a world where a nuclear catastrophe has occurred, the high variance distribution is better! If all nuclear power plants are basically the same and one explodes, we've gotta turn them all off ASAP until we figure out what went wrong. When there's a range of plant quality, one incident has an isolated, well, blast radius.
I have no idea how this preference inversion fits into the fragile/anti-fragile framework.
Doesn't distribution 2 inevitably lead to distribution 1 given enough time?
It's like you said - eventually all the bad car companies/bad power plants/bad or whatever will explode/go out of business because they're getting out-competed by the better ones. Once the absolute worst go away, they're followed quickly by the 2nd worst and so on an so forth until you're only left with the a distribution that looks like system 1.
As a rather doctrinaire libertarian, I am deeply unconvinced that government action is not the ultimate cause of social media censorship. Congress shall pass no law, sure, but very little of the administrative state actually involves Congress passing laws. After Operation Choke Point I see no reason to give the government the benefit of the doubt here.
Liberals saying that corporate censorship doesn't implicate free speech are arguing in bad faith. If the tech companies were to censor liberals then they would be up in arms about the chilling effects that would hinder free exchange of ideas. In the past, liberals often used to argue that corporations are more of a threat to free speech than government; they even wanted regulations of corporations on that ground. What are the common carrier and net neutrality laws other than government regulations to prevent corporate censorship? It's intensely hypocritical for liberals to champion net neutrality laws on one hand and support corporate censorship on the other. If ISPs don't get to control what users see then why do social media companies have that power?
> The quickest way to enrage me is to criticize some group of weird people doing their own thing without harming anyone - to try to browbeat them into doing the same thing as everyone else.
The fact that you characterize Parler and HBD enthusiasts as "some group of weird people doing their own thing without harming anyone" is the biggest self-own I've ever seen on this blog.
Thanks for the great blog! This whole antifragile thing isn't making sense to me. Car companies are not antifragile. They are fragile, if they make unsafe or unfashionable cars they fail.
What you are really saying is that if I as a consumer have choice then I am pleased to have a lot of variance. I choose which car to buy, to hell with the others. So I am happy to have a lot of varience in quality of cars.
Similarly I choose whether to exercise an option so I benefit if there is a lot of volatility/variance in the underlying asset.
I don't choose which nuclear power station effects me, they could all effect me if they create a global nuclear incident. So I would like little variance in nuclear power station safety.
In summary it is not the thing which is fragile/anti-fragile. Rather when people get to choose (options, products) they prefer high variance, and when I am effected by all individual items (nuclear power stations, earthquakes) they prefer low variance.
"If you did want to argue against diversity libertarianism, you'd want to show that relevant systems are more fragile than anti-fragile..."
The most relevant system here is the most fragile: the individual human life, of which we should all be fiercely protective, not on religious grounds, but on empirical evidence showing that every individual field of consciousness is finite, defined, and unique. (We could maybe debate this on religious grounds, but our concern in the modern world is necessarily political.)
The specific brand of libertarianism one advocates is not the problem. While I don't tend to object to libertarian impulses and actions per se, it's the ego commitment that blinds one to the damage, since intellectualisms reign supreme in big and small instances. It is simply not a morally defensible position to allow lives to become collateral damage to cerebral solutions, no matter how well thought out -- sorry MarxBro. (I know, I know, collateral damage happens all the time, but I'm discussing the extent to which a political philosophy should or should not embrace it.)
This is why the libertarian ego enjoys probably the most unearned sense of facile superiority of any political ego. Solutions can seem remarkably obvious to you if you are predisposed to shrug off the immediate risks and damages as "just part of life"; in which case you become free to ridicule all the "dumb" inefficiencies that plague democratic solutions as they attempt to attend to specific human rights, social justice issues, etc. Whether one is Ayn Rand, Hitler, or Scott Alexander, libertarianism (as a philosophy) is a symptom of diseased empathy among elitists.
It's entirely too easy to rattle off rationalizations that seem self-evident to you, such as saying everyone should just learn to rely more on Consumer Reports, without, apparently, the least bit of awareness that there are masses of real-life people who have never heard of Consumers Reports, and have no access to it. The confirmation biases based on one's own stable position in society are striking and omnipresent within libertarian circles. And if they aren't unconscious biases, they are, by definition, an embrace of Rand's abhorrent survival-of-the-fittest notions.
Perhaps our systems will evolve into new shapes in the future, but for now the only morally defensible political positions are the ones that tackle the MUCH more difficult, MUCH more complex task of tweaking existing systems progressively while protecting and preserving as many real flesh-and-blood individuals as possible. To say, well, 100 real people will suffer now, but if you implement our brilliant solutions millions of unknown, unborn future people will have freer, happier lives, is unacceptable in today's modern world. Y'all have to work harder than that. Still, many here see no problem with glibly sacrificing the uninformed to the consequences of deregulation, staggeringly unprepared or unable to place themselves in others' shoes.
And, by the way, the libertarian ego, I believe, is often susceptible to the really bad habit of rattling off false equivalencies without remotely adequate interrogation. I could go on and on about this, but just one current example: The Zuckerbergs of the world did NOT decide to "censor" Parler simply because they didn't like them. (Scott seems to think they could just have easily decided to "not like" Black Lives Matter.) No. They held out as long as they possibly could before eventually being forced to acknowledge the very real damage Parler was doing with explicit, philosophically based, calls to violence and insurrection based entirely on lies. Pressuring Facebook to restrict Parler (or Trump for that matter) is parallel to pressuring the government to regulate drugs and pesticides for the protection of even those people forced to work far too many hours each day to monitor the business pages. Enjoying Scott's mind as I generally do, I am repeatedly astonished by his, frankly, very lazy habit of spreading false equivalencies without adequate education -- a dead giveaway that trapped priors are in play.
I apologize for being pedantic, but this has been bothering me for the last two days.
The difference between the two distributions shown is not the amount of variance, but rather the kurtosis. Depending on the scales of the axes, the two might have exactly the same variance (or not). However, the second one has far different kurtosis from the first. In the language of statistics, it has fat tails. It is definitely not a Normal or Bell or gaussian distribution.
The complaint of people like Taleb is that we tend to assume that life looks more like Distribution 1 when in reality many events and processes look more like Distribution 2. Indeed, my graduate student advisor would become quite angry when people spoke of the Normal distribution, arguing that it was anything but normal.
My takeaway is that the world is a lot more random than most people think it is. We need to concentrate more on courses of action to deal with occurrences out in the far tails. That's what I call antifragile, and I think that has an echo in many of Taleb's writings. In this sense, Taleb has many precursors, albeit nowhere near as well known as Taleb.
I think you’re approaching Taleb’s point, but you’re still missing a big aspect of it; from his perspective, two different bell curves are very similar because they’re both effectively finite, even if they look different. If distribution B contains plants that melt down in a few days, distribution A will contain plants that melt down in a few years. The extra few years of power pales besides the hundreds or thousands of years of reduced land useability.
Buying a car takes advantage of variation, but it’s still at best basically robust; you the individual buyer may gain a small amount once from variation, but you probably won’t get more than twice or ten times as good a car as the average or anything. And you definitely could get a worse car than average!
Antifragility *compounds* on variation. Antifragility is like gravity, or science, or evolution, or a monopoly; it starts slowly and then it becomes all-encompassing and inevitable. It contains an argument about the single iteration game, but its true essence is about the repeated game.
Seems like there's an extra unintentional "or there are overwhelming externalities." in there.
Instead of your "religion" and "opinions" you should rather use "important and unimportant values".
The big innovation of liberalism is that maybe you should actually try to understand these evil people (that are against your own values) : either to have an easier time to beat them, to convert them, or to find a truce with them if the two previous ones aren't worth it or are not an option.
>But these are the only two major smartphone companies.
This should say something like:
But these are the only two major smartphone APP VENDORS.
"Right now there's religious pressure on tech companies to conform. Someone on Twitter pointed out that tech censoring Parler isn't a sign of their strength, but of their weakness. Imagine that Mark Zuckerberg decided he personally really disliked BLM, and he was going to censor BLM and any people/organizations/apps that promoted it from Facebook. Do you think he would succeed? Do you think he could stay CEO of Facebook after he was found to be doing this? Mark Zuckerberg and Big Tech in general are as much slaves to the prevailing religion as the rest of us; their "power" is the power to choose between medium vs. high levels of conformity."
You neglect simple legalities. Companies couldn't desegregate in the 1930's because there were laws enforcing segregation. Facebook shareholders could ban BLM if they wanted, and the Woke couldn't do anything about it except grumble like conservatives are doing now. The CEO has to do what the board of directors wants, which is what the shareholders want. Zuckerberg owns 29% (https://www.google.com/search?q=zuckerberg+share+of+facebook&rlz=1C1GCEJ_enUS867US867&oq=zuckerberg+share+of+facebook&aqs=chrome..69i57j46.5480j0j3&sourceid=chrome&ie=UTF-8), which is enough to control, so he can do what he wants. I conclude he wants to ban conservatives but not BLM. One caveat: if woke users of facebook would leave but conservatives will not, then Zuckerberg must worry about the dollar cost of indulging any preferences he may have against BLM. Similarly, if he has Woke employees he values, it will cost him dollars if he offends them enough that they quit. But Zuckerberg is NOT a slave to conformity unless he chooses to be. He is a leader.
Assuming there is a "best" option in your distributions, then always picking the best ultimately results in fragility. Most people would get Teslas and then distribution 2 starts contracting to look like distribution 1.
However, Teslas are more fragile than a Ford model T, for instance. Teslas are only the "best" in the current environment where we have working infrastructure that can support them. A massive solar flare at the wrong time, and the Ford model T might then be the "best".
But then accepting the anti-fragile ethos would suggest that we must spend resources to preserve and maintain a healthy population of solutions that are a poor fit to current circumstances, just in case circumstances change and we might end up needing them. This is probably a good idea to some extent, but it must have limits.
It feels to me that the way one describes something leads to "obvious" conclusions about fragility and antifragility. If Scott had described the nuclear plants as "The good one produces enough energy to power all of humanity's needs in a second, the bad one not enough for a light bulb" and the the cars as "The good one gets you wherever you want to go in 5 minutes, the bad one explodes as soon as you turn on the ignition" then we would conclude that cars are fragile and nuclear plants are antifragile.
This type of shenanigan makes me deeply suspicious that the concept of antifragility exists at all - or, at the least, that it existing doesn't say anything profound about the world that we should be taking note of.
It's a myth that libertarians don't believe in rules and order. On the contrary, more rules are needed in a free society than an authoritarian one to keep people from stepping on each other's toes. For example, this comment thread needs a decent moderation system to stay on track and be pleasant for most participants.
The real problem is moving tech from a novelty into a reliable utility. If electric companies inquired about your politics before providing power, we would still have factories powered by donkeys running in a circle and miss out on massive improvements in living standards. In the same way, tech companies need to eventually realize that their mission of running civilization is more important than virtue signalling of selective access. Only then the benefits of tech revolution will be fully realized.
Even more onerous is social networks claiming ownership of your friendship. It's one thing to delete content and ban users. It's another to not give users a realistic way to move to another service and still stay in touch. If Facebook allowed a group owner to export e-mail or phone numbers and provide it to former members, that would be a start of a more appropriate relationship between business and customers.
It's possible to have a libertarian society with companies behaving in this manner, but it's not likely to be pleasant to live in or endure for long.