Nassim Taleb: Small is Beautiful - but Also Less Fragile

well the the worst most tired cliché and introducing someone is to say they need no introduction and I would slightly modify it I hope now see needs no introduction because it's impossible to introduce him now seen Nicholas Taleb is unclassifiable it is this combination has never been seen before now see who actually to give my own personal story of virtues in New York City one of the pleasures of moving to New York City in 2003 was that soon after I got to meet with the author of fooled by randomness Nicholas nuttin Nassim Nicholas Taleb and he said since written two other very well-known books and other books that are not as much for mass market with the team he's written the Black Swan and most recently anti fragile he's ran his own two derivatives firm he was a pit trader at one point he's written thirty five academic papers he's possibly competing in the Olympics and gymnastics for all that I know so it's really a pleasure to have him to close our conference and please give nothing Nicholas Tyler a warm welcome thanks I'm extremely honored to be here and it was a great presentation on the power of bottom-up systems and some of the properties please no flash I may lose consciousness and then we may have to use another speaker so I'm gonna what I'm going to talk about is something very precise but I'm going to try to link it to research on cities okay because there is a connection via nonlinearities so of the plat next I guess forty minutes I'm going to talk about fragility and then we're going to see how sighs at some point makes you fragile at some point we're at that point how why a collection of distributed city-states are more robust than a nation-state why for example elephants are not as robust as mice and but these are physical properties and then we're going to see how bottom-up systems make a lot of small errors and you want to make multiplied errors and actually you probe through complex system via errors so it's very precise focused on this with a link to a non-romantic version of smallest beautiful there's a romantic version small as beautiful and then they give you explanation and it sounds good but it's not sighs I'm gonna and this is actually not empirical science so much as necessary mathematical troops with properties of object what what convexity nonlinear response necessarily implies as far as size is concerned anyway so let's start with this coffee cup here this coffee cup has one property fragility okay we know it's fragile nobody had define fragility okay before there's some versions but nothing really I would say not fragile there's no non fragile definition of fragility other the one then that it took a while but twenty years to figure out which is fragile is what doesn't like disorder simple okay in other words there's a random event hitting an earthquake in New York I think yeah earthquakes are not cities are not good with earthquakes that's one disadvantage well this coffee cup doesn't have any upside so it's events are either neutral or negative for for these things and can define facility based on that and then we're going to see what doesn't like fragility why it doesn't like a lot of other things like time and and so on can you hear me well with this ok now sometimes people because I know economists don't like me sometimes I turn it off completely so I thought but so you never know all right okay so one thing the myth some research on cities particularly by the people at Santa Fe people who deal with complex system please I beg you no no no sir no flash all right this is the negative externality professor of discussing before there is something this is what I found on a web as far as elementary that effectively when things grow something economists have yet to discover things don't grow in the same way so when you have a relative size of the head of a child as compared to that of that all okay the second thing I want you to notice is this curve as you see there is a curve there the curve of the relationship between metabolic rate and a size okay Express say in mass okay and that and as you see it's just the only thing you have to return from this is that it's nonlinear that's the only thing you need to notice from it is that it's not a straight line if you unless you do it log log plot in which case some people claim it's a straight line actually they try to fit straight line but you can see here easily that it is non linear okay it's not the raw relationships are not straight and then the other thing we could do is effectively we can figure out the shape of animals based on very small number of pieces simply because they scale they have to scale the in a certain way so if you have a small pieces you can really really figure out how this animal looks like okay and and within you know you can get get it right within you know small errors anyway so let me restart talking about fresh it's a lot of people at Santa Fe have done work on cities based on what we call elementary West and and Baton Corps and and and these people admirable work so just to tell you that a lot of work one can use to fit these ideas but I'm going to talk about just my word right what I do is what I bring to the picture is that notion of fragility that I picked up as an option trader in what does and doesn't like volatility so and then also there's a property that's the opposite of fragile when you ask people what's the opposite of fragile they say robust robust isn't quite the opposite of fragile because if you ask people even here at NYU in the Economics Department what's the opposite of negative they don't say neutral or the opposite negative is positive so negative fragility is not you know nothing so the opposite would be something that gains from disorder okay this is effectively the shape of something fragile in time-series space it has no upside from random events pretty much like a banquet for you if you're short volatility and this is what is not fragile robust it doesn't care and this is the opposite where the errors are small and gains cannot be large another's gain sorry harm cannot be large the opposite of the other one and this we tax atomized as anti fragile and effectively this is the only thing that have words on it that I put in here there's a major property which is anything that likes one of these will like the others if you like uncertainty as a property then you necessarily will like time okay if you don't like uncertainty take we humans don't like uncertainty we have no upside from uncertainty which is that which is the reason why if we extend life forever say medication whatever it is your God or anything extend like you still would not live forever simply because a small probability of an accident random events cannot help you as much of the harm you and that's sufficient to make you you know leave the gene pool eventually from an accident okay excellent free is not possible so what doesn't like uncertainty doesn't like variability what like saw certainty like variability and then we can you know see that in a second order effect so this is pretty much what my work is now one intuition in the book anti fragile and and that would help us get an idea something likes this order doesn't like this order or okay or likes moderate amount of disorder or not is this chapter I put an anti fragile given that I hate the press because they try like professor easterly they don't have this want to categorize you categorize what you're right about put it this is assistance is that you know you things you don't really care about so they completely destroyed my Black Swan idea by by conveying the wrong message Black Swan is about robustness two classes of events that we cannot compute it's not about animals you know and sexy images and try to predict random events of large consequences so to to retaliate I wrote that chapter and make sure that no chapter in anti fragile can no chapter title can give you any indication about this contents all right unless you've read it and I say okay or yeah yeah that's true so just to punish the journalist and effectively they couldn't read the book because usually there is a table of content and they stop there and then make a theory of where you're wrong about and stuff like that okay this is typical so I just wrote something on precautionary principle against GMOs and I notice that people start commenting automatically scientist without reading what's inside there because their comments were already addressed as a fallacy in the text means they didn't read it so it's typical of most people who comment the more people comment the less they read what they commented on and it's pretty much one inverse law anyway so the main idea is on a main difference between a cat and washing machine okay and so but it so but that's the central idea what's the main difference between the organic and non-organic between the engineered non engineer this is a watch okay and this watch will never benefit from random event you agree okay if you bank on it's not going to get better my bones get better if I bang on the table right may break the table you know if you do too too much you need a stronger table okay overtime so there's a difference between the organic and the engineered the organic communicates with its environment via stressors and only via stressors okay your skin doesn't communicate with the environment via your brain if you go on a Sun it will overshoot thinking there's going to be more Sun tomorrow and will darken you agree same with bone density same with stuff like that shots improve this why people go to the gym we don't quite get that idea that some things we classify as engineered because we read too many books about them okay effectively are organic and I call that the Soviet Harvard fallacy or top-down design thinking something they're organic and I'm sure it applies to the previous speakers you know intuitions and that fallacy is very big take the economy right is the economy something to be engineered or something organic that has a bottom-up process and only probe because uncertainty cannot be probed by top-down systems if this a certain class and effectively forest fires if you repress forest fires for it's something organic complex systems if you repress forest fires the risks accumulate and the minute you have a big fire it collapses alright so same thing with the economy but mr. Greenspan I think he got his fee flee here and why you I mean it's not the university isn't a harmless thing alright so he wanted no more boom and bust he didn't understand that because your washing machine doesn't need ok variation variability in the environment to get better ok the economy needs it to clean up the junk the bad stuff the bad company so people can fail early and that's a property of a good complex system and a perfectly good organic bottom-up system it should fail early and frequently to clean up continuously and that's how you probe a certain TVI you don't pro it via planning you probably for your experimentation so contact experimentation as you'll see so there this brings also a mistake in the perception of complex system or stuff like that is that people mistake variability for risk mostly because we have Beck terminology they're just victims of the terminology you say risk is not variability all right except I think it started in the econ Department and indecision science and then people say let's control risks therefore you control variability you're not the only two are together in some class for some narrowly defined classes distribution you control very variance of the Gaussian distribution you lower tail risk automatically they're two different animals and effectively if you eliminate variability from systems they blow up and this was known in control theory since Maxwell all right so people a lot of people know about these things but it so happened that those involved in economic stuff have a simplified view of things okay so if you remove and in what I was writing the Black Swan I quizzed people I'll tell them there are two countries two types of countries one country had forty governments the other country had only two governments of the same period which one is more stable actually the two countries on the right and one on the left the one on the left was what which one is more stable most people said that one that only had two governments was more stable which ones the country that had forty governments was not hard to guess Italy all right they're still around last time I'm going to actually right now I have my suitcase I'm going there so I hope it's there all right still standing right the the what is which country was on the right the country that had few governments hence was perceived by unanimously by the people I asked mostly economists unanimously has more stable Syria right so this is the problem all right over intervention if you organs being a system you make it seem a lot more stable but eventually it collapses and now I have an article coming out in foreign affairs triggered by Professor easterly actually who asked me to talk about city-states and I can't talk about it when I was a trader I could argue was a lamppost about anything right but now that I'm a scientist I can only argue about things and very well-defined stuff so I wrote a paper in foreign affairs tell them what you know the early is the way you can figure out if something is going to blow up and one of them is over stability excess stability low variability and stuff like that so low dimensionality now now the big mistake also is in real life we are we have effectively too many psychologists on the planet because when you ask people you know you tell hey have you heard of post-traumatic disorder everybody knows about it we cure people stuff they know but you ask him hey have you heard of something called post-traumatic growth how many of you have heard was traumatic growth and for those how many times have you heard of it right are you psychologists okay all right so okay shares my book so post-traumatic growth is that things from the Romans people know that things overcompensate by jumping higher right from compensation so post-traumatic growth I think I think I'm the researchers one hundred to one mentions in psychology literature and I think the prevalence is about 1 to 100 right post-traumatic growth your hundred times more likely to bounce back from a small trauma than behind by it but then again why do this part of that what I call Soviet Harvard mentality that started with the modernity but also there's another element nobody's ever going to make any money trying to cure you of post-traumatic growth right which is exactly where you have an agency problem so city states one thing about city stakes effectively there's a mechanism of over compensation okay but this is not my research so I've say to your casually don't tell anyone I said it but if you look you notice that what how countries get rich the phoenicians way supposing my assistant that I don't know you know the visitors to the area but the the Phoenicians had no resources so they went to Cyprus named after copper right was to import copper alright and they say hey we're importing it's an easy thing now we have a boat what do you do with it okay so look like it's a case of most small units that have absolutely no resources so we start trading and trading becomes a business so overcompensation again it may work for venice then it has another thing that made you notice of interest that that is my specialty are called the Lindy effect but it could be that that was also the case of Venice I wasn't there it definitely is a case of Singapore okay so if you need something if you have a little bit of post-traumatic growth makes you you know overcompensate by being rich but also there's a matter of scale scale is a problem and my Foreign Affair article vented frustration because I went to a conference with people doing political science and say how on earth can you compare you know political systems without attaching a size to it okay Singapore is different okay from has a different governance system China all right but the boss has similar political systems okay Singapore is not even Dubai right it's closer to Singapore but it has a you know it's not a Saudi Arabia it's not Sawyer is smaller Saudi Arabia things don't scale there is something that governance doesn't scale a lot of things don't scale things when they're distributed they L differently okay so now let me talk about non-linearity and then we'll get to the point of that's my specialty or a small specialty is any curve that has this in it okay as an option trader automatically we associate with it with money P&L we're going to lose make so automatically a different part of my brain gets engaged and then I can talk about it with some confidence and produce theorems but the rest is again anecdotal all right the cities is before was anecdotal this is not anecdotal there's something about fragility right that really makes it very easy to detect as follows if I harm someone with this big stone in my book I discuss something in from Midrash Tehillim where a king had to punish his son and the punishment was this big stone on his head right but the king was not very excited about the the regulation and come change regulation right so it was not the Bush family or something like there so he had to go with it but and then the rabbi came with a great a great idea and the here was to break that in small pebbles and you can execute the punishment and and keep the Sun okay and so what is so from that we can figure out that the ancients weren't that stupid they understood very well that if I hunt if I throw a 20 pound block of concrete on your head I'm not threatening again admit a thought experiment okay it have to say that now people may show you the stuff there it's assumed as the thought experimented that I saw someone's head okay someone from Isis station okay so 20 pound concrete do I harm them more than twice that if I hit him with a ten pound block of concrete yes no is it more than twice or if it's yes then he then then we have a very easy universality to fragility that's pretty much a something that actually by theorem we can pull out what doesn't like uncertainty has to have that and the interesting thing is anything everything alive has to have this this reaction so the way we define fragility was a little complicated is sensitivity to uncertainty as expressed by some measure of uncertainty that we found is quite universal right and incidentally I don't know I mean let me move the graph from here so people don't faint but in incidentally people here do economics or do statistical metric to statistics stuff like that okay can you ever see Sigma with that time okay Sigma as always has time attached to it alright so you can see time without Sigma for interest rates and discounting right so time and and Sigma and scale of distribution are the same thing in physics and an economic except for a few exceptions and these exceptions are usually approximations so and so it tells us the following that we can define fragility a lot of things and and sharks based on some dislike of some measure of uncertainty if another national trade in the room he would say Vega okay I get a Vega that's what its name is and it's not a Greek word but I don't know how it came about the word Vega but it's something how something doesn't like change in volatility and or some kind of a degree of uncertainty in a system so based on that measure I'll remove this very quickly we can do things to continue the same the our process and let's look at a simple story which is elusive have you heard of Kerrville the French rogue trader ok so you've heard he would have was hiding 50 billion dollars 50 billion euros in the drawer okay and that would show you why distributed systems usually feel better they discovered the error okay it was a Martin Luther King holiday in New York but the French didn't know about it right so they thought the markets were okay and they tried to dump in the market 50 billion euros and it was a French government type bank methods which means one trader on the phone selling and ten people tell them what to do there we go really so at the end of the day the liquidation cost is what costume you know is 100 and it was 4 billion euros of losses Marcos ended up down 12% when you're dumping 50 billion dollars you lose 12% all right the market end up Isola does an average 8% that's a average of what they lost so this now let's think about it what if instead of having a large banks our situation in Iran the employer of KVL you had tens smaller banks and each one of them would have in turn a Kerviel problem okay independently that can be the same day even doesn't matter was a different random variable different things you know there's a lot of things you to trade or you can treat with care if y'all treated with with some S&P futures the other person can cheat with coconut oil whatever it is so you have ten of them and you're trying to liquidate now five billion euros and even if you did it the Soviet way the Soviet Way is 50 people telling one person what to do all right okay so we have really hierarchy so and then you have on top of them 500 people tell them what so this amuses the Soviet way how much will it cost you to liquidate five billion nothing not close to nothing today plus or nothing so we have a curve here very simple curve you see that acceleration curve it's nonlinear all right the linear is the red one and the nonlinear is the vertical one that is a vertical one the nonlinear is a blue one for large deviation the nonlinear is hard much more than a linear one you say but for small deviation it doesn't show any harm and on a vertical access you have harm on horizontal axis you have event size think of Stone how much is stone with harm you okay versus weight of the stone so we have what we call convexity effect and that is pretty much love everything that test survived it's necessary because think about it if you jump 10 meters you die no okay except that then were you but then the other special case sorry but people die if you jump 10 meters but then if you jump 100 thousand times one millimeter you don't die there's it's not linear okay why because you've many many small sharks compared to large sharks so anything that survived is necessarily sensitive to large deviations would have the multiplicity of small deviation would have killed the thing early to give you an idea other in example a coffee cup there four million earthquakes per annum depends what you call an earthquake between two and four million all right I forgot what what the threshold the coffee cup filled all of them but doesn't care but you go above 300 their scales gone you see so you have non-linearity so it means larger deviations because they're rare okay haven't been experienced by the item so item sensitive to the large deviations have you know survived so this I mean the graph before we can explain to us a lot of things you know why when you grow there's something called diseconomies stochastic this economies of scale you are you have economies of scale we all know about it anybody who studied anything in Business School has heard the word economy the scale but then there's a logical fallacy right there they can observe the first thing I realize when if you think dynamically is when the professor was talking about the economies of scale say why on earth and we have like two companies on planet earth and effectively so you don't have to be blind and at a time when I was a school there was a very old alright so the there was a theory about that mergers didn't work the companies doubled in size and but they didn't deliver the predicted thank the predicted benefits okay and then call the hubris hypothesis so people still talked about the economies of scale and then they explained yeah because they got arrogant type of okay so that people get arrogant stuff like that all these romantic explanations don't work the only one is an elephant breaks a leg if Falls from this all right or Falls from heat from this height an elephant breaks a leg a mouse as they say across were I teach in Brooklyn and Brooklyn they say it don't care all right so really so so you realize that the scale so you have something called this economy Zisk and this we can measure we can measure you take bridges okay cost overruns how you handle our certainty cost overruns per size the same curve what else execution and markets same curve what else dams same thing project in the UK hundred million dollar pound projects okay have 30 percent more cost overruns than 20 million pound projects alright so you can measure all these things and effectively see it why a political unit that small has an advantage okay aside from robustness has has even financial advantage that's not taken into it so it's simply from the way they handle uncertainty so and here I have convexity effect in French and English so in English is don't cross a river that's on average four feet deep always bear in mind second-order effects in French is a complex function of an average not the average of a convex function right and the Jensen's inequality or variations around it that I work with alright so this is pretty much the response of the coffee cup it has it is a concave until it breaks and then you know you can't break a coffee cup that's broken all right so it doesn't it don't care as I say so you have everything it has nonlinearities at our local up to a point you benefit from uncertainty in a convex part and a concave one you don't this another biblical thing is the Tower of Babel all right Bush now is Baghdad suburbs of Baghdad or they have Isis and so anthalie alright so they had the idea that when you build something very big it's more fragile than a lot of small houses okay so we continue why fragility is necessarily second order effect with some intuition and I am pulling this from the book because it seems to me that what type interest people the most is that you carry you work you worry more of a second-order effect than the first-order effect when you very fragile okay by definition what's concave and here we have an example if you have grandmother and you know she spent 48 hours at 70 degrees temperature you're very happy for her no but if someone tells you 70 degrees on average but one day was zero degrees and the next day 140 degrees you're not going to be very happy for your grandmother okay so you realize that the second-order effect dominates the first-order effect on your fragile and that's one thing that's not very present in this course okay that when you start fragility and there's a second-order effect and we can see immediately what dominates the first-order effect okay so the the this is the s-curve in nature okay where you start at some level you go up and notice what that everything in nature that has an s-curve those response has a phase in which is convex and a phase where it's concave and you can have it you know convex concave and then convex again like the one at the bottom for example going to a lecture at NYU is pleasant okay great you can invite a friend it's a great thing but two lectures are alright and twenty lectures it's a negative you see so you have this is those response okay what is interesting is that the convex part of the response means you want to have volatility right there people don't quite get for example that you should vary within the place you should have some randomness doctors aren't getting it because they can't I'm just discovering from my fight over GMOs that they're not too much into into equations all right but then you can show it that anything in medicine that has nonlinear response all right should respond to uncertainty and we have evidence from III for interfacial I I found 200 papers showing evidence but they're not grouped together for example lung ventilators instead of giving right there instead of giving someone eighty one hundred percent of the dose you very eighty percent one hundred twenty percent you have a complex response okay you do better and effectively people survive better it's the same thing with your your heart the way you can predict someone's going to die is if their heart heart beats too regular so same thing with with a gym same thing with feeding alright if you don't start people religions know that of protein they get cancer so it's a second-order effect that matters more than first-order effect when people studied at Crete and diet and medicine for example they look at what people eat no morons look at how frequently they eat right they only had meets on on celebration days and that was not just a Greek Orthodox Church that was from you read any book on on ancient Greece how they ate they ate meat on festivals okay but then they ate a lot of it so they have definitely there is definitely convexity effect in that and then days when you don't eat me to caramelize some of your bad thing and you lower your cancer rate it's in the literature but they don't get the universality of the point of everything that's convex like variability so okay so I finished with with one went one idea that that I mentioned various earlier is a system that lasted a thousand years okay city-states have lost that's the system forever collection of city-states under some kind of mafia dawn survived much better than centralized nation state these two failed twice the first time in Upper Egypt the second time in in China okay when and effectively people think that the growth of China came from the centralized state with a bunch of nerds you know around the Emperor's Palace no no that's not how it worked this one is started decline they got rich and then that's how they spent their money and now first time third time would be America probably 2018 when large Washington would make the whole thing clap I know though okay I know I'm anyway so there's one interesting thing property of time and fragility you remember said that fragile what does like uncertainty the coffee cup doesn't like variability doesn't like sharks effectively so what's a survived time is giving us indication as to its ability to handle this order agree so you take anything non perishable means has passed the hurdle doesn't have a natural hurdle you can easily take the perishable to take something non perishable and you see an interesting property look at the graph you can see a child and the grandfather all right statistically you can expect the great-grandfather you can expect the child to outlive the great-grandfather bearing some kind of you know outlier you agree all right but now if your technology is a young technology or an old technology who's likely to outlive the other sorry the old it doesn't mean that the world doesn't have a lot of technologies it means simply that new technologies don't survive they're more likely to be supplanted by another new technology okay so that's we call the Lindy effect and I've sort of modeled it within fragility and it came out right okay so I'm sort of happy with this that effectively you're getting conditional information on on fragility from time and at end okay something there's a very simple rule when you know every year the old person lives his life expectancy goes down but a little under a year you agree every year technology lives this life expectancy goes up by more than one year so if it's a hundred years old it has hundred years to go at least statistically for someone who's going to call me with some exception we know there are exceptions but but the interesting thing is this is statistical and and it's quite liberating to see that the ones that really tend to stick are the ones and new technologies that replicate the old just simple example take the car nothing has been able to replace a car all right you know we tried you know doesn't work the plane that's how old the car 100 years the plane about the same all right the the bicycle is older now it's what's growing okay computers all right people say well we have the internet computers ah complete let's talk about computers all right the rule is invariant to definition of fragility you see so as a car we can define the car as something that has a box with wheels in which case is what 5,000 years old so Hyksos okay or you define it as a red convertible 51 years old all right you see so it's very interesting how you can define so now let's talk about computers right you've heard you know we had computers all right with something called mainframes what replace the mainframes PC what replace a PC this dying thing called laptop what's your placing laptops tablets aha now to find a tablet that's something you write on and you carry all right and you write with your hand on and how old is that technology all right okay is it so and it's very interesting that to see it fact if you look around glass you know this is plastic copy you know real thing the the all these things at the chair the chair actually is a you know people don't know it's it's very old but it was reserved this is thrown for four kings in the upper den Egyptian kings in the first dynasties right so you realize what else a table the the view her demeanor they where they have muscle here with a muscle ninive a that's where the fork was this was discovered okay so you can imagine that a lot of this knowledge we used today are very old and you can predict 3,000 years from now we'll use Forks we're not going to use some kind of fancy contraption and stuff like that so this is interesting because you can judge political system that way you can charge size you can scale you can figure out what has worked okay and effectively city-states are very interesting to look and that that way okay in something that has survived and there's something even more inch I mean and they come back okay you destroy him they come back and effectively something what I figured out the second order of the second order effect if I take two sample paths they both end up at St pass one pass to pass one went down straight-line path to or doesn't have to go down past two went through horrible lows and bus back the second path alright something that one through second pass is giving us information as to its fragility it's telling us how much turmoil it can handle you had to clean it up by something called survivorship bias but a coffee cup that survived the 3.0 on the Richter scale is telling us it can sustain at least 2.0 okay so that kind of information we're getting from cities so I'm going to add one thing for you to think not part of cities but sort of other cities which is where humans like dimensionality and modernity along with Houseman and ugly big ant I was named James Jacobs Jacobs the reverse of Jane Jacobs alright except modernity likes to do things that are of low dimensionality with a lot of corners but but but you look at this as aesthetically Pleasant no very high dimensionality but it is disorder no it's much more pleasant to look at then a highway you agree ok but why is it that everything in nature is absolutely except for if you're on a boat the horizon is probably threatening there very little in nature that's ugly and I think that look at this curve crop right this is very utterly right so there's something about geometry and dimension and I noticed one thing in Rome is I was an evacuated taken Museum when I took this thing and they have there's not a single square inch without ornaments and in a museum even outside the museum when I discover pieces it's not just that the good pieces made it to the museum and if you go and I thought that the pyramids were smooth till I went I went to verify and went to the New York you know the Met where they stole that big pyramid you know that was a big temple temple of yeah done door and that you go there I thought it was smooth get closer it's only smooth because of time you see we like high dimensionality like moldings we like stuff so while you walk around New York City we're the same thing mathematically can be mapped to your preferences living in an environment that is full of follow people high fractal nests to the environment and other and that is your preference your antifragility preference for some class of disorder so I'm going to conclude here I spoke about the property called fragility I try to apply it to city states to the notion of city because its scale is smaller and to the fact that we like some classes of the solar and maybe that's what we find in cities thank you for listening to me okay we have some time left for Q&A if you have questions please come up to the mic do I stay here yes yes I I wanted to thank you in 2008 I was working at the Federal Reserve and the research department and I was very tortured about the decision to pursue my PhD or not and it was when reading the Black Swan that I decided not to which was really an excellent position so okay thank you for that and my question is sort of on that topic of economics which you know you it's not exactly the topic but you alluded to problems with economics I would argue it's a fragile discipline and I've been spending a lot of time talking to groups and how I'm thinking how to shift it and I wonder if you think you just have to wait for this next crash maybe that's coming to really discredit it or if you think that I mean it's not going to come from the professors they're too invested no no okay so the economics so in other words summarize is you don't want to discredit scholars I've discovered from you read history of scholarship scholarship has always had a tendency to like this parallel word I called it in this book teaching the lecturing birds how to fly all right the event the world for example anti fragile tinkering bottom-up thinking process is behind growth and about anything okay but it's masqueraded you have masquerades like it's top-down science so that you have to look at it history a little more closely to realize that I call it the Soviet Harvard illusion again that you have a tendency by academia to think that the invented ways for people you know so when you look at the finish of technology in the dictionary what does it tell you technology means application of science to practical problems now in fact if you effect its exact opposite science tends to come from technology and what science didn't come from technology like Euclidean geometry this ugly thing right it came later we were building cathedrals without Euclidian geometry and we will start applying that we get this ugly squares you know that the ancients know better not to have okay so so the idea is so not to teach economists to be wise although we have some friends like bill easterly who are of course improving the field greatly but don't rely on them they're not they're never going to get wise so the idea is to build system that can deal with mistakes made by economists all right that's one one way I mean I personally say okay try to make some money by missing from by you know from the mistakes they make and that's fun but you can't really is hard and a lot of work and you have to argue with brokers and stuff so it takes but but so eventually we're going to have to live in a world that has some robustness built into it and you have to separate us from errors made by professionals city-states have the smaller the the distributed decision making all right allows us the said that that the robustness okay so don't but don't think that economies are going to be smart overnight okay that's not going to happen question yes yes sir iterate on that it says that you to in order to see in the future you know is to go back to the past exact know it tells you the following this is very interesting take the present as it is today and I'm trying to predict the future twenty years from now the the most robust thing you could do is you take anything of the past 20 years and you to remove it from the present you say you think that you project the present you don't know what's going to happen but you know what's not going to be there okay and this works statistically very well you say it I'm saying I'm not trying to predict account of Technology I'm predicting what technology will not survive okay you cannot predict it so it will be fools to predict what's going to happen you know what to remove from the future you see so the car has been 100 years you know with us don't expect it to go away next 20 years right the bicycle even less is more robust than a car you get the idea so and but you can expect this how systems work and even in architecture this is modern X vector I can it we can expect things to go back to natural things with a little more dimensionality yes you think positively in some way of Jane Jacobs yes in the sense that she helped preserve the West Village and from like neo mania modern yes but in like the result of that day was some sense the the neighborhood became frozen in time because it got land by okay transform anymore so I mean there's a difference between freezing something in time okay or keeping something alive organically you see the problem is when you start to post top-down solution you end up with avenues that others are there's this you see you end up with the Houseman style big avenues and Jane Jacobs probably should be it's good to be conservative but nature is not conservative okay nature is not going to dish or destroys every day but nature progresses by locally and by thinker it doesn't progress by having an Alan Greenspan co-opt nature and say we're not going to have seasons anymore all right just without a cycle it doesn't happen that ways it's an organic thing but so you have to put things in the condition to organically move and if you look at pictures actually building you see our buildings very short time hello I can't see them there in terms of fragility is it better to rent or own in the city I have okay if you I have I have I have no idea you're more convex if you own say you're more convex you have more upside of your own you have less downside less left tail I would say they're equally fragile all right anything to do with the city is not very smart let me I mean New York is too big it's efficient okay but listen New York is too big plus there's too much of the wrong noise you get in New York plus I mean buying on something when I see squares I don't buy you need cities built more walk around Rome and try to find a wall that is smooth that doesn't have even low-income housing in Ostia which is low-income housing by the Romans okay then had had ornaments you see so I wouldn't buy in the city that's like this yes universal human thing that we just can't see it or something you detecting then we tell you you detect it and let me tell you what one one one error I'm writing actually I wrote a book after anti fragile called silent risk and I'm splitting it into book one called the mathematical theory of fragility which is a short book and the other book on on risk okay and in that book I have a chapter on why psychology every time psychologists all right go against human nature by saying these are biases made by humans like black Ebola all right psychologists were all saying that we all reacting to Ebola no why because psychologists focus on first-order effect and your intuition incorporates second-order effect if you go with your first intuition it has there's something called dread risk which overestimate large deviation but in fact you needed to stay alive you say which is why when went we'll catalog when I wrote the Black Swan people say oh he's making people scared of these deviations it's very healthy to be scared a deviation but it's very interesting that was Ebola both your grandmother and good complexity scientist would have seen the same thing because one the grandmother detects intuitively second-order effect you the complexity scientists understand non-linearity but psychologists haven't yet gotten there in another two hundred twenty years okay to understand when they analyze human biases whether that their model okay may not be very complete okay it's the rational kind of analysis that doesn't that they're no so-called rational analysis is not rational just have the wrong models in that chapter I show you for example hyperbolic discounting why the thing is irrational in fact is very rational all right if you just make the model a little richer or models that would that have precisely second-order effect not detected by baby model or things like saying you know you understand you know that book nudge everything I mean largely what they say is completely you should do the opposite all right most people do the opposite like because they think they have a model that irrational not doing that when in fact there is a meta reason if you have a much richer more developed fact a of modeled things that fit the empirical reality a little more better and the mathematical okay reality you know is compatible with a good mathematical structure for example I took biases or people we stupid for for not investing that something called the equity premium bias people were stupid people's came out in 2000 saying people were stupid to invest in bonds or stocks and then you look at their model it uses a some kind of very invariance distribution all right use a different model he's saying is completely stupid to invest in stocks all right you see the idea so but and the switch between these models comes through convexity and then I explained in silent race so the book is called silent risk it's available for free I yeah I don't see the because of light yes I yeah I think that your book sort of give as a student all of us an important sense of skepticism about economic forecasting and central planning in general I guess I was wondering what you had to say about sort of all this stuff about big data and sort of a changing landscape towards vision that's very interesting because you can fit convexity to Big Data and it was following way and actually did that in full by randomness you have a covariance matrix okay all right now when the larger they have two things going for you all right when you do your number one is n the number of observation per variable you agree and two is the number of variables you have you see okay the failures of Big Data were known for a long time and effectively in 2007 we had half a billion correlations okay so and and and when you have correlation odds are something's gonna be spirits you agree okay so now there's something very interesting there's number of fake correlation you generate in a random matrix okay gross non linearly to the number of variables you're adding you see so there's the P as you're adding the number of variables making the cover for example if I have correlation matrix hundred by 100 it's almost impossible not to find twenty correlations that are significant you see you see there and this is so as you're increasing your covariance matrix the number of you know the dimensionality okay then this increases nonlinearly you see if I double my correlation matrix all right how many correlation I get n times n minus one divided by two no so you really growing up in a convex way it goes up it's sort of limited by the fact that there's something called the substructure is limited but it grows still very fast now anybody who's took a class in statistics knows that your statistical significance grows at what rate root end no okay which is concave so one is convex one is concave which tells you the following the more variables you have okay for a given set of n okay we only we don't have a lot of ends the more variables you have in finance we have like 35,000 variables all right but we hardly have twenty thirty years of data all right maximum I mean some we have but that's okay the larger the number of variables you have the more you're going to observe noise who detected it with the detected it was fooled by randomness and started meeting with people and then there's your one needies who was part of that group organized by fellow who died trying to find the in statistics okay and he wrote a paper in nineteen 2005 that's well known why most published results are wrong because as you have a lot of variables epidemiologists are going to take the upper bound you agree so you're taking the upper bound of variables so with time we have more and more data you're going to have more more spurious papers showing link between any kind of stuff epidemiologically and the FDA in 2005 there's a lady from the FDA who told us that they banned any statistical information they want clinical trials all right okay so that was 2005 by then 80% of papers that are statistical in medicine could not replicate and the prediction we could make all at a time the logical conclusion is with the speed of growth of data that 10 years later we'd be at 95% and guess where we are right there you see so pretty much because you have a growth of data and you have researchers and you have convexity and at the time it was not framing convexity I framed it recently in terms of convex versus concave so big data is going to eat us to the point but you can use big data for one thing if you decrease dimensionality and the only people who seem to understand big data properly are the the NSA guys because you're only interested is he a terrorist yes no if you're only asking that question you're not harmed but if you saw the data out someone's going to come in I'll make connection between the colors of the eyes of people in certain town okay and a number of positive events in stock markets that kind of that kind you see if you're looking if you're not pre specifying the thing by reducing dimensionality we have more time no sorry a three minutes okay so give it oh I just wanted to ask a question about one city you probably find really ugly to buy now if you if you map it on dubai yeah yeah Bradley so if you map it over time to you do you think certain cities have the stable property of fragility your anti future I don't know I I came late because of the travel and I looked at a reset in Paris when I was writing anti fragile and I was completely ignorant of these things and and in fact and I wondered where was the expensive real estate and expensive real estate turned out to be in neighborhoods that were village II in looking you see and then and then where you effectively had some kind of life alright and you can look at it in terms of dimensionality namely the fifth sixth in Paris and and and the Marais for example and all these areas that were completely spared by modernity you say jokes tells you that we were on a process before modernity of building neighborhoods in a certain way so I think that if you look at downtown Beirut to the area of the old neighborhoods and and we're not talking like Dubai it was a bunch of tents we're talking about a city that's five thousand years old right continuously inhabited or take Damascus eight thousand years old continuously inhabited so we have streets have been around 8,000 years Beirut has a few earthquakes but was rebuilt really along you know repatched up along what they had before so if you look at these places I would say that that that these cities are not doing something wrong okay because of Linde effect so I would say I would commit some kind of naturalistic fallacy by mixing my own aesthetics with with that Lindy rule by saying what people have like the longest is likely to make me happy for the longest you say you can be happy looking at this ugly this ugly building over there might have been pretty attractive for someone for about two hours when it was built right but you want something worse to eat you see the idea so preservation is understand that very well but they don't have that framework we have one more yes thank you I want to ask and this is maybe too soft of a question but is there in our society at least Western or American a cultural bias against variability or a certain shame associated with I looked at your path one path - yeah exactly so what happened is that there's a growth of allergy for variability that grows with a stat ISM modernity down control C so anything that stopped down control has that allergy for modernity for for for variability so you think you'd want to control everything and you start compressing variation that sort of happens was with the state to give you an idea where the state when I look at the area I'm from to conclude here I come and this is the reason I took this but I am not yet an expert on cities it's because I've made the following discovery in in 2003 years ago Aleppo start getting into war and I discovered this Aleppo had been a city that was really richer than Western cities in Europe and that part of the world from Aleppo to northern Lebanon to including homes that owned the Bekaa Valley had produced for Roman Emperor stuff like that but it was rich for about 5000 years right and the system of city-states and then of course you take Phoenician city handled invaders very well so they survived you know usually making a deal with the local mafia dawn and I was wondering why was it so rich and when when did the things fall apart it fell apart when that state entered and the metaphor is replaced the souk was it's all its mess with the office building and if you see Egypt for example Egypt if you see the prosperity Egypt was lending to the UK all right and Egypt was episodically Alexandria was not part of Egypt but exactly Solon City Beirut was not part of Lebanon its own spoke its own language and didn't want to be part of Lebanon so but these cities were very prosperous very mercantile doing very well but also they had something going for them okay that there were it was there was no status of the Olympic state and the state came to completely struggle you know to completely stifle these cities anyway thank you for listening to me

Channel: NYU MARRON INSTITUTE OF URBAN MANAGEMENT

Views: 90,927

Keywords: Nassim Nicholas Taleb (Author), Antifragile, City-State, cities, New York University (College/University), Risk, Volatility, New York City

Id: xyXwrpmt0zk

Length: 62min 4sec (3724 seconds)

Published: Thu Dec 18 2014