Nassim Nicholas Taleb at ICCS 2018

our first speaker is the distinguished professor of risk engineering at NYU Bloomberg named him as among the 50 most influential people in the world in finance one of his books was described by The Sunday Times as one of the twelve most influential books since World War two and The Times described him as quote the hottest tinkerer in the world unquote so please help me welcome Nassim Nicholas kala usually when you talk about risk you hope for failure because much more illustrates a point much better than that success so we have a there's a paper co-written with a bunch of people in this room john norman yummy a brawny arm we even have a philosopher involved who'll give us a metaphor that if you have a blind horse you want it you want your horse to be slow okay and and other people this was put on a web in 2014 and some events of course prevented us from publishing it mainly a smear campaign okay by Monsanto something like 1,500 letters to NYU and other things so but of course we're gonna publish it I'm not going to talk about genetics I'm going to talk about path dependence mostly so what happens when you look at risk dynamically not statically and something very trivial there practically every trader knows but somehow people who don't engage in risk-taking professionally don't know it because they haven't had to face the problem and also traders don't write books so typically so you don't have that problem so let me start with a very simple concept and and and with the following point if you iron your clothes first then wash them do you get a different result from if you first wash and then ironed you think okay so I mean I have not tested it myself but but you can feel free okay so now we have a pest dependence that's the plan a sequence of events matters hence any form of analysis needs to be dynamic of the outcome not static okay all right so now we take another point that sort of been every single surviving trader or surviving risk-taking person knows is that in order to succeed you must first survive it's not like two different additive components okay so simple but let's see why it has non-trivial mathematical consequences and it was missed supposedly by the decision science literature for the past 300 years I was very lucky to be ignorant of the decision size literature will help me write my first book dynamic hedging not knowing something and and and there was Murray gell-mann who discovered a couple of years ago that was missed by the decision sauce literature let's take the following case you have a bunch of people who want to go to a casino okay sample and of 100 you senator casino was constant betting rules and and same a lot allocation for one day by the end of the day how do you get the return from the casino how do you get the alpha of the casino assuming your samples large enough to capture under that gambling policy is a casino return so how do you get what do you do your average no so the average the return so if on day number if this fellow here on day sorry number 28 who index and number 28 he's ruined number 29 doesn't care no will even laugh at them okay so you get the return the casino VRS and very simple averaging and that's what we know it's something called the law of large numbers what happens to the buffered stochastic properties now if you decide instead to go to the casino for a hundred days with the total allocation same total location and engagement gammak policy okay on if on day number 28 you go to P oka to use the lingo what happens today number 29 there is no day number 29 so you cannot get what we call ensemble probabilities on sound leveraging from over time for a single individual I mean it's non-trivial why because this average here is computed as an arithmetic average and this is a multiplication multiplicative and Einstein said the hardest thing to understand is compound interest okay spending having spent 34 years and Finance I'm old so it's 33 all right so you realize the first thing people don't get this compound interest okay so if you have a zero somewhere okay it matters okay so it tells you the following something very simple if you look if you invest in stock market under any policy you will never get the return of a step market okay under under you know except there's only one and only one it's something called a Kelly criterion dynamical because if you have an uncle point in other words you you 2009 you're lower your allocation if something happens to you you're not gonna get the return of the market so our sound probabilities are very hard to compare two time probability so let me continue Peterson Gelman and 2016 showed something I showed it in 1997 I called it never cross the river if it's on average four feet deep okay and then we formalize it 2018 with a paper called the mathematical foundation of the precautionary principle which are summarized very simply that time averages have different dynamics and then you're subjected to absorbing barrier okay so you have to worry about taylor risk and whitetail risk are complete different animal from regular variation what Gillman figured out is something vastly worse and he called it non ergo the city is the fact that if you you know people have the illusion that over the long run things average out for you they will not Everage out for any single individuals they were average out for a collective so let's say here the sample pass you stop you you you you you you you over time your various of outcomes will increase you say it will not decrease you individually so in other words things will never average out for you so when you take a decision you gotta make sure you're gonna make it to the end you cannot use so all these theories of probabilities about you know multiple world that you're selecting from sample all that doesn't work under that you know rigid structure where you're follow only one sample pass okay and that was Mary what Gelman and Peterson Gelman actually it was Peters idea and bellman or the co-author sorry for for problem okay so now we understand one thing that practically every trader knows goldman sachs had been around where i hate him with them in around 149 years okay how do they make it because they have a rule they don't want small tail risk you know what they want zero tail risk okay zero not small and I was a bit trader you know you go stand and you destroy your vocal cords you lot of germs you learn new language you know how to trade with hands you know how to get attention and a bit when when there's a crash so a lot of things but the first thing you learn is what every single old trader tells you is the following don't worry about losing money lose all the money you want just worry about making being here tomorrow having surviving and complete different animal okay and that's basically different dynamics the tail risk because very simple if you have a small probability ruined on the repetition what will happen to you well you know what happens to you things add up so the problem of not developing things dynamically okay if you take the decision size literature and there is a lot of literature gonna look at it the literature of qualitative behavioral economics or all these things you haven't figured out that your risk should never be evaluated as one-shot experiment you see as a series and then therefore think that appear completely irrational such as paranoid paranoid behavior see paranoia are necessary and not even the rations are necessary and then we can be a very simple example let's say that why your evaluation needs to be done in a certain way smoking okay if you do one shot experiment smoking okay is it's irrational to not smoke if you enjoy a cigarette and the reduction of rushed pregnancy from a cigarette is very small no okay why don't we smoke because there is a correlation between smoking not dying after a cigarette that's smoking yet so so therefore you don't really look at think dynamically that I'm please different think if they say oh he's paranoid by giving you a payoff structure and looking at your risk aversion and thinking they can measure it that's what psyche always did and on top of that with students what if you're parked outside what if you're going through a divorce what do you have the risk or have children you may have to put through college all these things in fact your decision I mean so you cannot isolate decision you know any single decision from your lifetime saying so what we see if I read my lifespan is very simple every action you do it is how many times you gotta repeat it over your lifetime how much it reduces life expectancy so you cross the street you're reduce your life span by one and forty seven thousand years not a big deal alright unless you plan to live forever it's not a big deal okay but there are things that reduce life expectancy particularly if you repeat it so you take your life you practice your search effectively is an unknown variable but for that so you have to integrate in a different way so the other problem is it's not let's talk a little with scientific consensus because people think that you have science and then on top of that statisticians deciding on hey this is this is our K statistically and then on top of that for risk you cannot go by consensus and let's say you have a 0.1 error rate which effectively you know whatever you know very statistically significant safety of the airplane how many pilots would still be alive okay you realize that the probability of crash probability of deaths on a plane is less than 1 and something like 55 million per something that or something obscenely small okay so we're saying that in risk management different criterion than science you can say they're scientists scientific established that because gotta go higher and in my session tomorrow don't talk about p-values anybody who sings p-value or scientific has a problem we're going to show the following that if a p-value you have a p-value the phenomenon has a p-value of 0.12 which point 12 25% of realization of that phenomenon will produce a p-value of less than point or want okay so all of the p-value is not an observable phenomena it's a stochastic number that people don't realize particularly when the researcher does many experiments till you get to be valuable 0.01 whatever you want okay so so p-value not tell me this scientific now finally now comes the precautionary principle we're talking about life span no okay how do you look at it there's you you would like mmm my life span is I don't know it's particularly if you eat yogurt and worked out at the gym stuff like that your life spans what about ah come on 20 people in this room maybe under 25 right so you so you reduce from that but there's something because of your family friends and pets okay so your des plus their death as up so you have a bigger reduction of life expectancy if you involve more than one person and in that group okay and then you keep keep going now it's scaled up till you hit on something like self defined extended tribe okay however you define it okay like squid ink you have a squid ink ears on Twitter for example group itself and your descendants alright so and then here is humanity okay so if the whole I cross the street I'm not reducing humanity's lifespan because that's an independent thing but if everybody all of humanity cross the street at the same time you may have a problem okay and then of course the ecosystem so you realize that the higher you go the more of a penalty because very simple counting things you're reducing life expectancy if you're going to do that policy forever so it's why we got to be very careful now another thing I learned from training and something learned from Warren Buffett is always say no all right you say no Warren Buffett says no systematically all right or he says very politely I think about it which means no all right always say no because there's you know you don't want returnless risk okay unless you paid a lot for something so you notice people have survived they don't think Taylorism and they have the the benchmark to invest or do anything is very very high it's people who invest other people's money you'll get in trouble typically now one small little thing about behavioral finance and we're quantitative finance with that theory of nudging is very simple let's go back to the casino I remember my policy in the casino all right I go ahead go bust the exist one on one policy to not go bust in the casino and in entails playing with the house money okay in other words you invest you start taking more risk as you're making more money okay the the that's the only one that will allow you to capture the as what we call the return the alpha of the casino now it so happens that such a policy is pathologized by Richard Taylor just got the Nobel Prize no sorry pseudo Nobel Prize in Economics that he says plain of the house money he called it is irrational because why because it's irrational to treat money the but if it comes from the casino or says your initial endowment for him it's the same money it's not the same money dynamically okay so no wonder why all these much people don't understand that there's a lot of things that considered irrational such as paranoia I mean listen which survives 350 million years so it's not for some schmuck with a p-value to come argue with you that is a rational all right okay sweetheart 50 million years you gotta be doing something right okay whatever we can decide 350 million I don't know I mean maybe five depends how the fact you have geneticists here who define humans differently alright so let's say as mammals or whatever or as something that's alive for millions of years okay so now which let me ease into something which is my specialty fat tails okay what I'm saying isn't particularly intelligent nor particularly new okay it was voiced 1903 by a fellow Luneberg and might formalized in 1930 by another fellow crime heir who invented insurance mathematics and he has this state a theorem all right the crime is something called the crime air condition if you start insurance company you collect fees from people you see so you have the thing it's like a bathtub throwing that was water and once in a while you got claims agree what happens for you to never to to have a lifespan okay that's no it doesn't have a certain bankruptcy certain world's called gamma deluxe download when certain ruin the distribution all these claims need to follow it in certain class called the exponent subjects outside the subjects financial class okay sometimes called Super eight financial class what does it mean let's see with me this catastrophe principle means that if you're drawing from the solution that is hat tail forget about it forget about you cannot insure this unless you have bounded losses you have you have a lot of restrictions okay you can and and and that's a crime here in other words you get absorbed once you lose money beyond a certain your calendar your endowment here boom you're out you're froze you've lost it okay and otherwise you have a certain bankruptcy regardless what your return is which is why cost-benefit analysis don't work under fat tails hence fat tails so I'm going to explain fat tails and the least technical terms were the following quiz I have two gentlemen here and and the total height is four meters and 20 centimeters okay randomly selected from the population what's the most likely allocation 4 meters and 20 centimeters no what's the most likely combination sorry okay I pulled sorry 4 meters 2 meters 10 to me to stand it's very very unlikely to get 2 meters 10 okay but it's vastly more likely to have so in other words take it let's say the following what's more likely in the Gaussian to have one Six Sigma event or two 3 Sigma events 2 3 Sigma events you know you multiply okay no matter how how low that's gonna be will still be considerably higher odds than six sigma events okay and the more you go into the tail the more that right now distribution that don't have this property are calls in a sub-exponential class so let's continue now with another quiz you run into two people I said you know randomly selected from population and it so happened that the total net worth is 36 million dollars what's the most likely combination exactly so this is their you know for a fact tail distribution if you have seen now that has applied to returns for fat tail distribution if you have series of returns okay and there's a huge loss odds are it came from one for a thin tail distribution other it came from a combination of a lot of negative events hence insurance is split thanks for crime air he said you split the insurance regular insurance like heart attack stuff like that into one block which is Gaussian and you move the rest to reinsurance which is very very hard which is fat tail okay what is catastrophe principle so now we know the difference between tail fat tail it so happened that a lot of relevant variables are fat there so and mistake people tend to commit often and yeah near here we had to fight the State Department and to fight all these guys couldn't get it when they start using empiricism say a I'm a psychologist at Chicago or this and this so irrational people worry about Ebola which killed Klum between zero and one person in America and this that killed four hundred thousand year - no you idiot let's think about it you Bill Maher's on Elance thing as a guest so on and there's no computation and then the minute you establish communication with earth you here well 180 million people died conditional on hearing that was it gonna come from alcohol diarrhea obesity tobacco or Ebola course 100% chance that came from Ebola you say so when you look at the tails it's a complete different story you got to analyze things from the tail okay and you should never never ever compare the means of the fat tail distribution with entail this music you cannot compare me you should never do that okay so this is the Economist and I have a trick any time you want to read something wrong in other words you put a negative sign in front of it for the truth read The Economist because they ridicule people who worried about Ebola all right and of course we bola as you can see is a multiplicative process okay well as all these other diseases are not as multiplicative okay maybe but not as a tilt so and statisticians unfortunately so I'm I work now for called fat tail project has fine estimators that work for fat tails it's completely different class of things and a lot of its not changes color of the dress see we're gonna use right for distribution instead of the Gaussian the whole story changes so and so the whole story becomes very different now let's let me finish one thing before I moved out to fragility concept of fragility that when we compare policies we have a tendency of saying okay for example there's some idiot sorry so some person wants to treat em rice right modifying rice why because they look at the benefits that's fine but the problem is if you look from the tails it's just like story of Ebola okay when you look from the tails all right sometimes benefits don't change the tail probabilities the more you go on to the tale the more when you look at particularly if you introduce an uncertainty with your policy slight amount of moral certainty percolates big time in the tail so so saying I'm gonna improve the system by adding by improving the mean okay you're not moving away from the tail you're actually getting further into the tail I mean they're a lot of product boxes that come from that and of course when we talk about probably distribution the the white here that you see they forget about it there they're even within the sub-exponential there's a lot of different different distributions okay now okay now let's go through fragility and convexity and actually everything I'm talking about here is available in two books that are three really available on the web one collection of papers in a fat build project the statistical consequences of fat tails and the second one on fragility alright so what I wrote the book of the Black Swan and then you're right we write an equation form and it's very strange people understand stuff in equation form they can't understand in English it's a complete opposite so those rewriting my books in math okay with theorems then they get it so let me explain that how you change in the risk class simply from okay for example when someone says Oh GMOs we've done it all the time we do something called breathing all right you know you cross dogs and you get a different dog no all right so that doesn't work that way it's sort of like saying oh yeah you know going down from the building's 50s floor you know whether you jump or you walk down it's the same risk class it's not the same risk last and the the idea of convexity is as follows all right the problem is that well it's very simple if I jump a meter actually I get better you know wake me up it's X called exercise jumping from here but if I jump 10 meters I die okay so they are so people don't realize that the the size or the speed makes a huge difference like if you hit the wall at 1 mile per hour a hundred times you're not gonna get the same damage of once at how many miles per hour okay so the the idea that everything that survived has to have that calm parity to risk in other words everything is dose dependent so you change this class simply with the dose very simply okay that's the idea of anti-fashion so why are we going to concave large events hits you a lot more when you're concave to risk so that's the idea of fragility so when you speed up any process okay you enter a different risk class okay know for now I'm sure someone may ask questions for unless you get scared too much to ask questions after that after what you're gonna you know see now now comes the problem of dimensionality okay i dimension is there something well-known called the curse of dimensionality for some classes of processes curse of dimensionality tells you the following that when you go from you remember the great Stephen Wolfram was here and he talked about things that are computationally irreducible so in other words don't know what happened at step n plus one from n of one you're a so from from one you have to go through all the steps to figure out what happened competition to reduce you can't see that's because when you go into higher dimensions all right the things become computationally really that's the NP problem that we have discussed that yeah near you know what note about as you go into higher dimension you have more and more computational demands that vastly exceed your computational abilities so we have all these discussions of NP P vs. and peas that's things you can check non-starter so but it's worse statistically and let me explain why when we do law of large numbers what are we doing alright the errors wash out because mathematically your everything so you have pluses or minuses and things wash out you're doing all right now if I have a vector in high dimension and I want to figure out its tail risks okay you're not using the theorem the the theorems that we know for conversions for sums and we're no longer using also something called the central limit which is also an averaging think and it's a summation what are we using the maximum domain of attraction theorems alright and how is it computed the same thing that was a casino by multiplication alright so errors multiplied you already have a problem figuring out what you're gonna do with the genes right when you have an higher stock monogenic all right combination and combination with the environment so there's a lot of problems already but the problem is if you want to track your error you got a problem tracking your error and statistically because if you probabilities have an error the thing explodes so as I'll go and higher and higher dimension okay the relative error swells big time look I mean it speeds up so you go from 20 and of 22 n of 21 I'm trying to figure out the probability because you've got to realize that God didn't give you the probabilities except maybe some here but typically you have to discover in itself so you have estimation error and the estimation error and to blow up when you compound them ok mathematically is trivial but this big problem now finally [Music] you can see the same problem when we talk about fat tails in higher dimension in terms of dimensional reduction of BCAA how for example the the you may have a spurious pc8 here look how big it is compared to spirits pca in gaussian world under summation that the point is technical but just to I'm trying to read ascribe the problem of fat tails in higher dimension when you have a lot of that tailed variables and try to figure out how they work together okay so your information the statistical data you might may get the pic to be very spurious because of that problem and and and that problem is encountered big data as you go into higher dimension you have a very large D and not enough large and you need the larger larger end experiment whereas your DB becomes larger this is one dimension and then of course people tell us hey what are you doing writing this paper in outside your domain which is probability and we answer that and something we're discovering more and more in editing anything that start having a lot of tails becomes a problem for probabilities provillus not a problem for the man let me explain when people say oh I understand the risk because I'm a biologist although we have ballast on you know on a paper people kept attacking us your guys are not biologists we're answering with the idea yes if I want to go to the casino and understand the odds of roulette do I hire and and particularly for tail events don't hire a carpenter or do I hire a probability person ok that's what we call the carpenter fallacy and pretty much any scientific problem becomes a probability problem once you go into either higher dimensions I want to go for large deviations not the medium at the center is the specialist problem for the tale becomes a probability problem so with this I'll stop and I guess we can we can take you I don't have much more time there is chairman [Music] sorry everyone what a mental problem no okay we have lived historically in a world where everybody wants everybody avoided the fat tail and those who didn't are no longer with us alright and their genes are no longer with us and because so the the point of hate collectively what what happens if there's something even more interesting what happens for fat tails is actually easier to analyze okay for large deviations so I like my story of Ebola you don't need a lot of information to realize that it's not our attacks that killed 180 million people but Ebola if you don't know what the Jesus all right so this is why it's easier sometimes to answer the tails it's one but I mean a couple more points I want to make here about about the idea that in life in science you need to get information in life what you need to do survive okay first survive because it's a condition okay so realize sometimes if people have crazy policies to make things more scientific whereas in real life you've got to make it simpler okay to survive because fewer side effects so when I take problem like when it took the GMO problems I quickly said okay what's the dream or what is their aim their aim is to produce cheaper food okay okay cheaper food okay I want to attract the price of the tomato the tomato at price Club is at 100 say is one dollar to cents go to price Club Costco how much will how much money goes to the production sorry very little okay it's a tiny toilet between say five and fifteen cents all right and we're talking so if you want to reduce the cost of the tomato and as Ian ear has observes all the submission problem you got to work on what on a simple distribution box okay who figured it out left missus this is businesses businesses to make product because you know you focus on the 80 cents not on ten cents you may say and hearing me say 40 cents so we have one older magnitude easier solve a problem with fewer side effects so this is how people think in business okay so it's fine science needs to study phenomena but creating or policies out of science is highly insufficient okay you know what you got to put it on the scrutiny of probabilistic risk management that's my whole point that was my point on Black Swan nobody got it another question so when you talked about the relation to the decision science literature and the path dependence I was wondering whether you had looked at the sort of stochastic optimal control version of decision-making problems where you can use a barrier function and actually sort of intentionally prioritize not dying stochastic optimal control of the branch that are sort of like what with because I worked on stochastic processes when I was a trader when I was you know I'm stochastic optimal control locally makes sense all right but but but you had to put it in a certain framework the point is it's a branch of mathematic and students eyes that doesn't make claims on how the world should be or what probability should be you get your probability from somewhere okay the decision size literature since memory okay supposedly got the probability from horizontally from state space which is the summation rather than apply that these other things that's the point and actually stochastic control is the the approach to take if you want to never blow up yes okay and it will give you the Kelly criterion okay naturally I'm referring to the point that you just made about like making tomatoes cheapest and so on the issue with GMOs I mean I come from a farming family right and I come from a farming family and the problem that we have is not that we get the fruit cheaper it's just that one bad year and it sort of wipes out half of your income so it is intrinsically a fat-tailed problem it just on another side of the distribution the point is that of course you're going to reduce risks and stuff like that but if you don't use the wrong tools of analysis okay you're gonna get the wrong answer but of course you're gonna have some problem Assol by people by GMOs but you but but they're not looking at problems are not solved and have not never seen a tail risk study from GMOs all I saw there's some yield improvement which actually is quite funny right so that was my point and then also by my okay if you have cycles right I ain't got the cycle and actually we wrote a paper in a paper we wrote up at the New York Times that caused an attack by fellow and it turned out that Monsanto wrote this paper and because it was removed from Forbes okay a guy called and let me name him forms remove them and all his articles it turned out sorry Harry Miller yes he he attacked us the point is as follows the crisis we had who is who is the Greenspan who is he's trying to eliminate the cycle okay which is very nice nobody nobody likes a cycle you want to live in this sighs okay so what did we get we get by only the cycle we got a bigger collapse that happened right after in 2000 I was still not out of it so so you got to be very careful when you try to eliminate something you know a natural cycle maybe we can elevate but just don't think the first order statement was that looking at side effect like mr. wheen spended okay there's fellow with a beard seems angry so maybe could be a challenge right so if you don't produce genetically modified rice what is the probability that the lack of certain something is going to cause something that you haven't foreseen that's wrong that's exactly that's exactly how would they say well they say I'm saving you from a bigger problem all right that's that's an argument every Charlton has always uses but that's not actually my question I'm sorry but it's not that we know the other problem that we're saving you from but it's that we don't when we get to very rare events we can't we don't have the capacity to imagine these paths in a complex system so we can't then quantify the two sides of the ledger either way the way we have answered that is that if you let nature is you have some quiet you let you follow how nature has been functioning or has functioned for a long time okay and then you say that if your status tician worthy of his over-salt then you would say an N of three billion is vastly better than n of 500 okay so therefore you say okay if Nature has done it and what is the closest thing we can do to the risk management system of nature and that's what we addressed in the paper at length is that there is a existent naturalistic risk management system because conditional on being here after all these years means something has worked out you know well so let's not say this is an unknown but and the Newberry you may tell me we live in the model world we have a change I doubt I doubt that is so modern okay that it that things have changed so much in nature so global warming I think most of us would agree is an extinction level event but then you hear the the environmentalists going and saying ninety seven out of a hundred scientists agree and that kind of okay we should be talking about risk she's saying global warming okay there's a consensus and what you go with it we don't we have we use a precautionary principle by saying the following okay well if you're agnostic about global warming say I'm agnostic about global more all right what do you do agnostic is actually a theological term fully used here what do you do if ice if you receive a package and you're agnostic of whether there's a bomb in it what do you do all right okay if you get on a plane if you're agnostic whether the plane will crash what do you do so this is called an asymmetry and let's address in my facility thing this comes from a symmetry if there's an asymmetry all right don't go with the route that has the negative payoff okay so my point is I don't know anything about fossil fuel okay but my point is if I if there's a and we wrote a paper actually was drawn Orman and yeah around or they say in the following you don't make decision on global warming based on what we know scientifically but but based on what we don't know okay and if I have doubt if I hear and if there's uncertainty about whether it's possible even more okay let's forget about it move elsewhere okay they got other things to do that's what a trader would do if I don't know whether the investment is gonna bankrupt me well if I don't know well guess what I'm not gonna invest in it you see so it's the same mechanism and the same answer it gives you know to GMO and no to fossil fuel okay I wonder if you've thought about fat tails at the positive end as well as at the negative end yeah I'm thinking of funding of science about things that are high risk but very very high return like an energy source that would just put fossil fuels out of business okay the way you figure out something very interesting about far tails that the right side doesn't offset the left side okay it was sort of like maybe playing Russian roulette and I'm but well but winning a billion of you know is that that doesn't offset okay so like they're separable and in classification of fragility classification how was fragile and how the fragility works you say you don't look at the the how complex the payoff is a positive payoff you only look at the negative one okay to determine if it's fragile okay so you want bounded risk and then a big payoff and if you start seeing that the hey maybe we're not bounded in other words we have a we don't have the floor when I invest in the stock market if I invest $1,000 that's bounded but in real life it's not that clear okay so you if you see that it's not bounded that you then you you know you read your exercise principle but the one thing again the precautionary principle doesn't allow to every single risk it should apply to risks that flow up on that scale and we discovered that although everyone is scared of nuclear nuclear is doesn't fall under precautionary principle it falls on the regular risk management people take care of it because you cannot blow up the planet was no clear you need a lot of nuclear to blow up things you say whereas it was Ebola Ebola is spread to the whole planet you say of course of course - nexi except mixi of course because they have protection and stuff so but but but literally the whole planet so this is how you gotta analyze things based on starting with the extreme deviation and and then flown back in first we're out of time let's thank our speaker once more [Applause]

Channel: New England Complex Systems Institute

Views: 25,651

Rating: 4.9363394 out of 5

Keywords:

Id: tD3JqtLEF1A

Length: 44min 3sec (2643 seconds)

Published: Fri Feb 01 2019