Espresso: Rory Sutherland Why Reading "Anti-Fragile' Will Change Your Life? - Nassim Taleb

0:00One thing I want to ask you about, Rory, is Nassim Taleb's book Anti-fragile. I think last time we talked about thinking fast and slow by Daniel Kahneman. i'd love to know about anti-fragile like what are the lessons the big lessons from there the first principles you take from that you think you can apply to it's really really marketing specifically marketing specifically statistical assumptions that we overlay on real world observed behavior okay yeah now full by randomness was an earlier book which is perhaps more of a behavioural science book less mathematical. Why, how so? It's the extent that we actually, that we will often extract patterns out of random events that aren't really there, or we will extrapolate.

0:47We underestimate, we underestimate. I mean, there's a great book by Brian Clast called Fluke. We underestimate the role of randomness, okay, in everyday life. And probably we have a kind of, I think the book was partly inspired by a meeting between Taleb and Kahneman, actually. They were... All by randomness was. Yeah, I think I'm right. I think I've got the order right there. No, yes. I mean, before Kahneman had written his book, but I think Kahneman and Nassim knew each other quite some time before either of them became a published author, I think. Okay, because Taleb obviously is very interested in the sort of inherent biases that happen in trading, for example.

1:30Yeah. And, you know, so one of the points he makes is that we tend to conflate the likelihood of something happening with what we should necessarily do. OK, so a poker player will understand that you occasionally place a bet which has a very low chance of success. But in the event of success, you make a fortune. OK, so Nassim would occasionally take a counterintuitive bet on what the market was going to do. And he'd say, we're almost certainly going to lose money today. But if we make money, we'll make so much money that actually these are the kind of bets which we should be making. The contrary bet, which is known in financial circles as picking up pennies in front of a steamroller, is where you take a whole series of small gains, congratulate yourself for your financial acuity, only to find out that you're making those small and continued gains at the price of a monumental risk, which is the steamroller.

2:29okay so picking up a load of gains from something which in other words if you climb an example would be if you climb very tall trees to go and get extra cherries but one time in 150 you're going to fall off the tree and die okay you can congratulate yourself on days 100 to 149 that you're the best cherry picker in the forest but on day 150 you exit the gene pool okay so So what Taleb is looking for is a much more nuanced appreciation of statistics and our interpretation of statistics and our interpretation of probability and likelihood. So, for example, we often assume things are Gaussian. Gaussian.

3:08OK, this is getting nerdy. OK, we often assume the normal distribution to events. OK, when in fact they're fat tailed. OK, and the fat tails of the graph are that. Okay, so height, human height, I'm not the best person to do this, but I'll do my best. Human height is basically Gaussian in that you will, well, actually not, it isn't really, actually, because you get midgets and dwarves. But, okay, but at the extent that you're never going to get a human, a working human, adult human, by the way. You're never going to get a working adult human who's one foot high, almost certainly. and you certainly aren't going to get an adult human who's 15 feet high.

3:51So it's distributed on a kind of normal distribution, where tail events are unbelievably rare and most people are somewhere around the middle. Is that like a bell curve? It's exactly that. It's a bell curve. You got it, OK. A lot of things in life actually... He's looking at the tail ends to the side. He's talking about mediocristan, which is human height, And he's contrasting that with something called extremistan, which is human wealth. So to give an example, there is nobody so tall that they could walk into a packed Wembley Stadium and meaningfully change the average height of everybody on the stadium simply by their presence.

4:32But if Bill Gates walked into Wembley Stadium or if Elon walked into Wembley Stadium, the average person in that stadium would now be a millionaire. OK. And so not all things are distributed by a bell curve distribution. A lot of things actually there is far more extremity. And by the way, the poorest person in the world is also likely to be extremely, you know, someone who's literally a billion dollars in debt, for example. OK. The poorest person in the world would meaningfully affect the average wealth of people in the stadium, potentially the other direction. And so his idea is that lots and lots of risk models in banking assumed bell curve probability where, you know, people referred to, I think in the run up to the financial crisis, they'd say things like, we've had five, six sigma events on five consecutive days, to which you go, literally, okay, what you're suggesting there is that something that's a one in a billion probability as a standalone event has happened on five consecutive days.

5:34days it's much more likely that your maths is wrong than you've just experienced literally an event that shouldn't happen in the entire lifetime of the universe okay so so what's happening there is that and in full by randomness he also makes the point that organic or living systems are not optimized around an average they're often optimized around a desired level of variance yeah so in other words and i would argue this also applies in things like creativity okay so i think that marketing is fat-tailed in the to the extent that um the most successful things you do will be so successful okay in other words 20 of everything you do in marketing will probably deliver 80 of the value i wouldn't be surprised if one percent of what you do in marketing delivers 50 % of the value.

6:25And therefore, asking for marketing to always wash its face, same would be true of R &D. Demanding that every piece of marketing you do to be matched to a corresponding quantum of profit is a fool's errand. Okay. And yet that's how we measure marketing effectiveness.