There are bad events which cannot in principle be predicted.
However, Taleb can already predict which systems will benefit from those events. /s
This is my general problem with Taleb: it feels like his books keep telling you repeatedly that no one can actually predict or understand something, only to suggest that Taleb has some kind of knowledge beyond knowledge that allows him to predict the unpredictable and explain the incomprehensive. Sorry, I don’t buy this. If no one can predict stuff, then Taleb can’t either; if Taleb can predict a thing or two about stuff, so can possibly someone else.
Of course, the “motte” is that institutions which are inflexible and their success is based on too many dubious assumptions, will break when something important changes, and such changes happen once in a while.
But beyond this, I think it is more likely to be a trade off. A bet on things remaining the same, versus a bet on things changing quickly enough that we can actually benefit from being prepared for the change. A huge empire may gradually fall apart as a result of its own complexity and bureaucracy; but in the meanwhile, it will destroy hundreds of communities that weren’t large enough and coordinated enough to resist the attack of a huge army of a centralized state. Other hundreds of communities will avoid the attention of the empire and survive. It is not obvious that being a member of a randomly selected community is better that being a citizen of the centralized state. Even a reliable prophecy that one day—an unspecified moment between today and 500 years later—the empire will fall apart, will not make the choice easier. Or maybe one day, Microsoft Windows will be completely replaced by thousands of competing flavors of Linux; I just don’t believe that Bill Gates should lose his sleep over that. One day, Java will be a new Cobol, and all Python and Ruby developers will have a good laugh about it (that is, until Python and Ruby become new Cobols, too), but in the meanwhile, my Java skills are paying my bills. Etc.
So, one problem is that unless the changes come soon enough, your anti-fragility features are going to be just dead weight. (If they are providing some benefit in the meanwhile, it means you could have designed them for the purpose of that benefit, even without worrying about anti-fragility.) Another problem is that a genuinely unpredictable bad event can wipe off your anti-fragile solution, too. (Maybe the “anti-fragile” features you designed make it actually more susceptible to the event, not less. That’s what genuine unpredictability means.)
tl;dr—robust systems are usually more desirable than fragile ones, but “anti-fragility” is a pipe dream
Antifragile systems do exist. The ecosystem of restaurants quickly respond to demands, many go under within 5 years, but some survive. At the individual level, the restaurants are fragile, but the system of restaurants is antifragile and not going anywhere barring a major catastrophe. True, one cannot be antifragile to everything. Nonetheless, one can determine what types of disasters a system is antifragile to.
If all restaurants were run by the government, either that system would collapse or it would be some hell-hole equilibrium of high-costs and low quality.
Your point about large centralized states is well-taken, though.
The current system of restaurants could suffer greatly, if (1) some company would start providing cheap delivery of high quality food by drones, or (2) some epidemic would make it dangerous for people to eat in public. Well, neither of these would wipe out the whole system, but it’s just what I though in a few seconds; worse things could probably happen. Also, luck would play a great role, e.g. if first we would have the food delivery by drones, and a few months later the epidemic, with proper timing the combined impact could be much greater than either of these individually. A machine that could cook an (almost) arbitrary recipe automatically at home (plus a convenient delivery for the raw materials) -- at least the recipes usually found in the restaurants—could also change a lot.
Yes, having many parallel solutions that work slightly differently, makes things more robust. This is a lesson I would love to see implemented in the school system: have hundreds of different types of schools, each providing education in a different way.
On schools: I agree that it would be interesting to have many types of schools. However, it doesn’t seem like an ecosystem that could work. Schools don’t have strong feedback mechanisms, and for there to be different types of schools there needs to be different types of communities that these schools are serving. The current model of schools is not merely a top-down imposition, but is the evolved form the school given a larger social system.
There can only be different types of schools, when we form communities with different sets of needs.
Charter, Private, and Public schools are usually very similar. Sometimes schools do something interesting with their curriculum… usually not though. I think the place to look for different types of education are Special Education, Gifted Education, Online Learning, Homeschooling, Boarding School, and Hybrid schools. Each of these form themselves to serve a particular kind of people.
However, Taleb can already predict which systems will benefit from those events. /s
This is my general problem with Taleb: it feels like his books keep telling you repeatedly that no one can actually predict or understand something, only to suggest that Taleb has some kind of knowledge beyond knowledge that allows him to predict the unpredictable and explain the incomprehensive. Sorry, I don’t buy this. If no one can predict stuff, then Taleb can’t either; if Taleb can predict a thing or two about stuff, so can possibly someone else.
Of course, the “motte” is that institutions which are inflexible and their success is based on too many dubious assumptions, will break when something important changes, and such changes happen once in a while.
But beyond this, I think it is more likely to be a trade off. A bet on things remaining the same, versus a bet on things changing quickly enough that we can actually benefit from being prepared for the change. A huge empire may gradually fall apart as a result of its own complexity and bureaucracy; but in the meanwhile, it will destroy hundreds of communities that weren’t large enough and coordinated enough to resist the attack of a huge army of a centralized state. Other hundreds of communities will avoid the attention of the empire and survive. It is not obvious that being a member of a randomly selected community is better that being a citizen of the centralized state. Even a reliable prophecy that one day—an unspecified moment between today and 500 years later—the empire will fall apart, will not make the choice easier. Or maybe one day, Microsoft Windows will be completely replaced by thousands of competing flavors of Linux; I just don’t believe that Bill Gates should lose his sleep over that. One day, Java will be a new Cobol, and all Python and Ruby developers will have a good laugh about it (that is, until Python and Ruby become new Cobols, too), but in the meanwhile, my Java skills are paying my bills. Etc.
So, one problem is that unless the changes come soon enough, your anti-fragility features are going to be just dead weight. (If they are providing some benefit in the meanwhile, it means you could have designed them for the purpose of that benefit, even without worrying about anti-fragility.) Another problem is that a genuinely unpredictable bad event can wipe off your anti-fragile solution, too. (Maybe the “anti-fragile” features you designed make it actually more susceptible to the event, not less. That’s what genuine unpredictability means.)
tl;dr—robust systems are usually more desirable than fragile ones, but “anti-fragility” is a pipe dream
Antifragile systems do exist. The ecosystem of restaurants quickly respond to demands, many go under within 5 years, but some survive. At the individual level, the restaurants are fragile, but the system of restaurants is antifragile and not going anywhere barring a major catastrophe. True, one cannot be antifragile to everything. Nonetheless, one can determine what types of disasters a system is antifragile to.
If all restaurants were run by the government, either that system would collapse or it would be some hell-hole equilibrium of high-costs and low quality.
Your point about large centralized states is well-taken, though.
I guess we mostly agree here.
The current system of restaurants could suffer greatly, if (1) some company would start providing cheap delivery of high quality food by drones, or (2) some epidemic would make it dangerous for people to eat in public. Well, neither of these would wipe out the whole system, but it’s just what I though in a few seconds; worse things could probably happen. Also, luck would play a great role, e.g. if first we would have the food delivery by drones, and a few months later the epidemic, with proper timing the combined impact could be much greater than either of these individually. A machine that could cook an (almost) arbitrary recipe automatically at home (plus a convenient delivery for the raw materials) -- at least the recipes usually found in the restaurants—could also change a lot.
Yes, having many parallel solutions that work slightly differently, makes things more robust. This is a lesson I would love to see implemented in the school system: have hundreds of different types of schools, each providing education in a different way.
On schools: I agree that it would be interesting to have many types of schools. However, it doesn’t seem like an ecosystem that could work. Schools don’t have strong feedback mechanisms, and for there to be different types of schools there needs to be different types of communities that these schools are serving. The current model of schools is not merely a top-down imposition, but is the evolved form the school given a larger social system.
There can only be different types of schools, when we form communities with different sets of needs.
Charter, Private, and Public schools are usually very similar. Sometimes schools do something interesting with their curriculum… usually not though. I think the place to look for different types of education are Special Education, Gifted Education, Online Learning, Homeschooling, Boarding School, and Hybrid schools. Each of these form themselves to serve a particular kind of people.