← Intelligence Squared
Intelligence Squared
How Should We Prepare for the Unknown? With Statistician David Spiegelhalter
00:44:49
Loading summary
Transcript81 lines
- [00:01]
Commercial Narrator
Insurance isn't one size fits all, and shopping for it shouldn't feel like squeezing into something that just doesn't fit. That's why drivers have enjoyed Progressive's name your price tool for years. With the name your price tool, you tell them what you want to pay and they show you options that fit your budget enough. Hunting for discounts, trying to calculate rates, and tinkering with coverages. Maybe you're picking out your very first policy, or maybe you're just looking for something that works better for you and your family. Either way, they make it simple to see your options. No guesswork, no surprises. Ready to see how easy and fun shopping for car insurance can be? Visit progressive.com and give the name your price tool a try. Take the stress out of shopping and find coverage that fits your life on your terms. Progressive Casualty Insurance Company and affiliates Price and coverage match limited by state law.
- [00:54]
David Spiegelhalter
You're in your apartment alone. Then you hear something. You think, was it just the storm? You realize you're not alone. Your living room is not safe. With Unhinged, the new immersive game experience brought to you by Netflix. To make it out alive, you must answer the phone. The question is, when your phone rings, will you answer, tap the banner to play Unhinged now only on Netflix.
- [01:24]
Mia Sorrenti
Welcome to Intelligence Squared, where great minds meet. I'm producer Mia Sorrenti. In a world of economic shocks, geopolitical crises, and constant predictions about the future, how should we make decisions when certainty is impossible? And what can statistics teach us about living with risk, chance, and the unknown? In this episode, physicist and science broadcaster Helen Czersky speaks with statistician David Spiegelhalter about his book the Art of Uncertainty. Drawing on probability data and real world examples, Spiegelhalter explores how we assess real risk and sheds light on what roles chance, luck and coincidence play in our lives. Let's join our host, Helen Chersky, now with more.
- [02:05]
Helen Czersky
Welcome to Intelligence Squared. David. We are here to discuss your book, which I'm holding here, the Art of Uncertainty, how to navigate chance, ignorance, risk and luck, which sounds like everything. And it's got a lovely picture of a double yolked egg on the COVID I don't know how I have never got double yolks in an egg without it, without, you know, having. I guess you can buy them. Can you now you can reduce. You can increase your chances these days.
- [02:31]
David Spiegelhalter
That's my joke in it. Of course you can go and buy them in shops.
- [02:34]
Helen Czersky
Yes, you can buy. Your choice. Okay, so there is, there is There's a lot in here. And of course, this is all particularly. It feels particularly relevant at the moment. I'm sure humans have always lived with risk and uncertainty, but I think a lot of us feel that the level of uncertainty in the world is particularly high at the moment. So this is a good time for this. So. So I wanted to start with something, actually, which is towards the. Almost right at the end of your book, which is that you have a section titled Manifesto for Uncertainty. And I just wondered why. I mean, the whole book, in a way, is a manifesto for uncertainty. But why do we need a manifesto for uncertainty?
- [03:11]
David Spiegelhalter
Oh, I think because there's a popular impression that people don't like uncertainty and that somehow uncertainty is a really bad thing. And it's not. Well, first of all, it's unavoidable. We can't be certain. We don't know what's going to happen. We don't know what's happened in the past. We don't know what's going on at the moment. We live in a life of uncertainty, and it's a huge part of being human. And without it. Can you think of anything worse than living in a completely certain world, knowing exactly what was going to happen? It'd be like a worse sort of prison sentence. And with audiences, I ask people, who likes uncertainty? One or two people put their hand up. I said, okay, all right, so the rest of you, you want to know what you're going to have for Christmas? You. You always watch the last episode of a. Of a. Of a series on Netflix first to find out what happens. You want to know the result of a football match, you know, that, you know, before you watch it, if it's recording. And so, and then I ask, well, okay, who. If I could tell you, who would want to know when they're going to die? And that. That one you usually get. Yeah, maybe about 5, 10% of people do put their hands up. The planets, the ones who. The. The planners. But most people don't. So I say, sorry, you love uncertainty. You. We live. We're un. We live in a world of uncertainty. It's fine. So. So actually, that is a complete, I think, misunderstanding and that, you know, I suppose in the book, I kind of finish off by saying, well, you know, we should. If we can actively enjoy it, because we can't avoid it. And. And actually it's what makes life, you know, interesting and varied and means there's a little bit of, you know, little free saw of risk perhaps. And that risk, meaning some things might Work out well and they may not, may not work out so well and how awful it would be to be without that. So my manifesto, uncertainty. And you know, I do talk about how one might help to deal with it a bit, but the point is that this is very personal, people's attitudes. As I said, people vary a lot in their appetite for uncertainty. Your risk appetite is, you're supposed to be assessed on that when you buy a financial product. And actually, of course for some people there's intolerance of uncertainty scales that you can do because if you're really intolerant of uncertainty, it becomes a clinical condition, in fact, because you can't bear not knowing everything, not having everything down to absolutely obsessively precise plan about absolutely everything. And this is debilitating to people. Yeah. And so, you know, it can you. For some people it can be. Some people relish it more than others, shall we say.
- [05:48]
Helen Czersky
But we have to deal with it one way or another. But the other thing, I mean, so you know, you are, you are. No, you're famous for being a statistician, for knowing your way around the, the mathematics and the explanations and the graphs and all of these complicated things that perhaps we don't see so much of in, in everyday life. So I think it would perhaps come as a surprise to a lot of people to say, to hear you say at the start of this book, as you do, which is that that uncertainty is a sort of relationship that it's not something that you can pin down with numbers and facts and graphs and it's more slippery than that. So why is it that you, the great statistician is saying that actually uncertainty is a slightly subjective human relationship based
- [06:32]
David Spiegelhalter
thing, possibly a totally subjective point of view. It depends whether you actually think the whole universe is deterministic or everything that happens is God given and it's all. You live in a fatalistic world and you know, that is an argument that people can make. I don't care. I, I don't, I avoid that discussion completely about the actual degree of randomness in the world, in other words, I mean, I think at the quantum level there is true randomness. True, utter and you know, irresolvable unpredictability. But the impact on that on our daily lives, I think is, is really minimal. Quantum effects, they don't impact what, what happens to us, I don't think. And so actually, you know, possibly everything is basically predetermined, you know, an immensely complex. It doesn't mean it's predictable because chaos theory has shown that Just if something, something may be deterministic, it does not make it predictable in any sense at all.
- [07:23]
Helen Czersky
So there's a difference. So just, just to emphasize this, the difference you're making here is that in a deterministic universe you can follow. It is possible theoretically to follow a sequence of events that if this happens, then that's going to happen, then that's going to happen. And what you're saying is that those chains of connection can be so complicated that even though in practice it might all be predetermined because you've clicked this button and it's going to do that thing, actually in practice it's unpredictable because we can't determine.
- [07:51]
David Spiegelhalter
And we know that. We know that from Chaos theory. Even the tiniest 18th decimal place change in one of these inputs changes the whole future. So determinism isn't the same as predictability. Those things completely separate.
- [08:07]
Helen Czersky
But I guess where you're going with the book really is this thing about it being relationship is that actually there are pragmatic. There's a pragmatic side to this. Whether the universe is deterministic or not, we have to think we're taking a
- [08:17]
David Spiegelhalter
decision other people might know. There are facts that other people might know, but I don't. And similarly for everybody, there are things that they're uncertain about that other people may know about and may in fact predetermine it, or may just be a fact of the world they don't know. So we have to live with this. That's why I say it's a relationship. It's a personal expression of our own natural ignorance of what's going on, which can vary from person to person.
- [08:43]
Helen Czersky
And then we have to decide how we feel about it and decide how we feel about behaving in certain ways as a consequence of it. So it is useful to know about. I mean, so the ways of quantifying it will come to whether or not. Or not they're any good in a bit. But it's possible, it's useful to know a little bit how uncertain you are. So you can make decisions.
- [09:02]
David Spiegelhalter
Yeah, yeah, I think so. Because if you just say I don't know. Well, hang on, you know, we don't know about hardly anything. So there are grades of don't know, essentially. We're basically living a whole world of ignorance. But you know, some things we know more about than, than others. And I guess my work pretty well, the whole work of statistics and, and everything I do is to do with trying to, trying to work out what those grades of uncertainty might be and how to communicate them in an adequate way that's helpful to people. That's pretty well what the whole book's about.
- [09:33]
Helen Czersky
Okay, so it this all this often, you know, in people's heads when they think about statistics and probability and chance and uncertainty. Everybody I'm sure will think back to some maths class at some point. And I think it's really interesting that maths is not like almost by definition, maths is not intuitive because the whole point is if you follow logic, it takes you somewhere different to where your intuition might take you. And I wanted to. So many years ago you organised a coincidence party that I happened to be at because you're that sort of person to do that and I'm the sort of person who would turn up at such a thing. So we're both in the geek bucket here. But, but what was interesting about that, even you had all kinds of games and we sort of, you know, I remember walking in and you got everyone to write on some kind of thing that we hung around our neck some three random things and it was amazing how many, you know, I heard people in that room going, oh, you've been in prison in India. I was also in prison in India for a day. You know, just talk to me a little bit about why it is that those things still seem surprising to us. There were so many games you could play on us that seemed surprising, whereas actually follow the logic, they're not.
- [10:44]
David Spiegelhalter
Yeah, exactly. Our intuition about matches and coincidences is very poor indeed. And the standard one is birthdays. We know mathematically and practically it happens all the time that if you've got more than 23 people in a room, there's more than a 50% chance that two of them will share a birthday. So those sorts of. That's why the party. I can't remember how many people we had. You had to put your birthday on it. And, and the chance of having an exact match is certainly more than half. Once it's 20, 23 and, and then you could be one day away, two days away. And so, so the point is, and so the chance of having something in common. And in fact we know that from our experience. When you meet somebody and you go through this business of, you know, exchanging stories about this thing, you, you're going to find very often find some connection. We know about the degree, idea, degrees of separation we'll find, especially if we share, you know, maybe an, maybe a background, maybe a class or something like that. We'll find somebody might even know someone in common. But we might have, you know, might like the same programs might do this. We get. We have an enormous amount of commonalities and they're going to show up all the time.
- [11:50]
Helen Czersky
But isn't it interesting? I mean, I find it interesting that even though we. We all live in this world, right? We all live in this world of coincidences, and yet we continue to be surprised by them.
- [12:00]
David Spiegelhalter
Well, some of them are staggering. But, you know, for example, you know, but it depends on the perspective. You know, when someone wins a lottery, 1 in 45 million or something like that, it is a massive coincidence. It's pure luck. You know, it's a massive coincidence. It's not surprising to hear about it if they sell so many tickets. And so it really depends on your perspective. And so I, we had a whole coincidence website, thousands we had submitted to us. And you. I, you know, used to grade them roughly into, you know, like, you know, a once a year sort of coincidence, which they happen for. That's not very interesting. Once every 10 years, once in a lifetime coincidence for somebody. And because people remember these things because they're telling about them about years after they happen once, and then you. Once in a hundred lifetime coincidences. And when people tell those, you think, wow, that is amazing. But really, in the end, there's a lot of people, there's a lot of opportunities for things to happen. And so I tell you, the one that's in my book about Ron Biederman's trousers, which is still my favorite anyway, either because somebody wrote it and said, well, it was a real coincidence. I was, I was robbed when, when at one point I was staying in a hostel and. And Ron Biederman gave me a pair of trousers, too, which is a sort
- [13:13]
Helen Czersky
of special, which is not usual by itself. I mean, people don't generally go around giving out their trousers.
- [13:20]
David Spiegelhalter
Ron. No, no, I got to get this in the right order. Gave him. Gave him a shirt. That's it. And then years later, he was at a hostel and he was talking to someone over the. Over the other side of the counter and he said, and they got talking about friendly people in hostels. Oh, well, I met this man called Ron Biederman, and he gave me this shirt. And the guy on the other side of the Ben stood up and said, yeah, Ron Biederman gave me these trousers. And they were a matching set of shirt, the same cloth. And I thought, oh, this is nonsense. I don't believe this. And then we got an email from Ron Biederman with a photograph of him Saying, yeah, I remember. I remember giving these things. He's obviously incredibly generous man, did a lot of traveling and he gives his stuff away to people.
- [14:03]
Helen Czersky
Right? So. But there is something that. Which is interesting, which is that actually we'll come to this in a bit. That the, the, the. It sounds incredible, but when you actually look at how it's set up. So these people are both in hostels. There is this one generous guy called Ron who seems to have a thing for giving his clothes away. So actually, so what looks like a staggering coincidence at the start, actually, when you hear it initially, it's actually very hard to judge how uncertain that actually is.
- [14:32]
David Spiegelhalter
I know, and I wouldn't want to put a probability on it. That's why I kind of think, well, this is the kind of thing that somebody. It happens once in their lifetime and they'll remember it forever. That's how I kind of categorize.
- [14:42]
Helen Czersky
So this then relates directly to something which is at the start of the book, which. So, you know, I'm a physical scientist. I spend my time measuring the environment. And that means I draw graphs with error bars on them. And you basically make the point that you think all of these error bars are, if not exactly nonsense. They are not nearly as. They're not saying what it looks like, they think, they say, yes. So just unpack that a little bit because I think people feel, people see an error bar and they go, oh, well, okay, now I've got an eye. And they are useful in science, but tell us what your caveats are.
- [15:20]
David Spiegelhalter
Yeah, and I'm someone who's worked as a, you know, my whole lifetime as a statistician, frantically always calculating margins of error and putting error bars and confidence intervals around things. And I don't believe any of them, basically because they assume the model is true. And all statistical analysis is based on assumptions of, you know, independence and normality and linearity and all this sort of stuff. And no missing, you know, maybe no missing confounders. And, you know, there's assumptions throughout every statistical analysis. Every time you use a package that isn't just for descriptive statistics, anything that's sort of doing any sort of inference and all those assumptions are wrong. Now, they may. It's the. There's no such thing as normality. There's no such thing as linearity in real life. So. But it's. Whether they're importantly wrong is the issue. So, I mean, it's back to this basic. Well, it's almost, almost kind of cliche of George Box, a wonderful statistician who said that all models are wrong, but some are useful. So that means that every conclusion based on the model is technically wrong. It's not correct, it's just that some are useful. But in particular, competence intervals or error bars are always too narrow because they assume that it's leaving out major sources of uncertainty, the systematic biases, the errors in the model and things like that. So they're always too narrow. They're not wrong. Doing wrong doesn't make them too wide or makes them too narrow.
- [16:37]
Helen Czersky
So. Yeah, well, I mean, but it's. I think it's important because, I mean, certainly in, in science we, we make a lot of judgments based on those. And I guess there's, you know, as an experimentalist, obviously I'm delighted to hear that all models are wrong because that's what I spend my time. I spend my time saying, that's not what I actually measured. Are you sure about that model? But I think, but I think it's very interesting how we're very drawn into, you know, there's something as we've become a more scientific society where we like knowledge. Right. We sort of feel we have the right to know that we can quantify everything, that we can. We have enormous confidence in our own abilities, I guess, is what I'm saying. And it's very seductive, isn't it?
- [17:17]
David Spiegelhalter
Yeah, and it's a delusion. I mean, the main thing is that I don't want to, you know, say all the science and stuff, it's all wrong, it's all useless. No, it's incredibly valuable. But it's not precisely correct. Correct. And we need to have some humility about our analyses at all times. The example I give in the book, which I still think is a fantastic example, is during COVID you know, with the R number was we're all obsessed with it. How many people someone with COVID on average will go on to infect. And in the UK, with eight teams with 12 different models trying to estimate R. And once we all got together and they put. They showed their estimates and their error bars and when most of the error intervals didn't even overlap. So they're trying to estimate the same quantity from largely from the same data, but the error don't even overlap. They cannot both be right. And it's because they very different ways of using the data. There are very different assumptions in their model, very different types of model. It's a really good way to do science of all these independent teams using different approaches. So they got together once a week and constructed A sort of, you know, interval that sort of vaguely covered most of the opinions that were being given, but a composite interval. And apparently the meetings I met people afterwards who told me about them were very good people didn't barge and say, oh yeah, you've got to use my model minds best. And they were very. Had a lot of humility. They said, well, you know, this week I don't think we've really captured what's going on and we're a bit cautious about what we've said. But the crucial thing that all just absolutely demonstrated that all these intervals were far too narrow because they were assuming their models were correct. But they did exactly the right thing. Fantastic science, brilliant science. And then published all the results, the individual results. It's very similar to what people are doing in climate modeling as well, with independent models and so producing a composite out of the model outputs. Yeah.
- [19:06]
Helen Czersky
So I was just about to mention that because we do. That's kind of routine now in predicting the future is hard. And obviously with the climate, it's not just, just, you know, this is a very big bet effectively on the future one way or another. It's important that. And actually this, this method, these ensemble models where you run it not, not even just eight times. I mean, sometimes thousands of different experiments working on different assumptions and you kind of see, you assume that, that there is an assumption built into that, that we know most of what might be useful and we express it in different ways. So we. Yeah, but I think it's, it's very. These questions of what we really know can be very unsettling if you're, if you're trying to find out. It feels very hard.
- [19:45]
David Spiegelhalter
I think the vital thing is the independence of the teams. I mean, this is more expensive because you don't want. You. You really want people who are almost actively competing, who don't necessarily think the same way, have got different perspectives. And again, the example I give in the book of this, which I found so impressive is in 2011, when Osama bin Laden. It was thought that Osama bin Laden was in this compound in Abbottabad. And Obama had to. President Obama had to decide whether to send in the seals in this massive. In this raid, which would have been. Which is high, very high risk. And so he got multiple independent intelligence teams to assess the probability that Bin Laden was actually in the compound. And they came up with a variety of opinions. So some were quite pessimistic. Only 30 to 40% probably chance he's there. And there may have been a Red team, Red Teams are deliberately set up to sort of question groupthink to think of alternative things, but to be rather pessimistic possibly and things like others were really gung ho. 80 to 90% chance you said. So that gave. Meant it's a really difficult decision. And then Obama said with all, I think it's about 50, 50. But he could see the variation in opinion among his advisors, which seems to me that every really decision, every decision maker should know about that variation. So I really think it's important almost that modelers don't talk to each other too much, that there is diversity of viewpoints for that any actual decision maker. I mean obviously you want all of them to be credible. You know, these are reputable people who know what they're doing, not just anybody's opinion, but that the diversity of viewpoints is really important.
- [21:30]
Helen Czersky
So you have, you, you basically are gathering as wide a wider collection of possible viewpoints and then you can evaluate them one by one. I guess so. Well that leads us to one of the things that you point out, which is, which is actually very worrying when I had never seen it in a table before, I don't think, which is that you talk about language, the interface between the numbers of some probabilities that have been calculated and the language we use to describe them and the idea that likely, this word likely means completely different things to different people.
- [22:04]
David Spiegelhalter
The table I think has got eight different definitions of likely and they're not that different. You know, 55 to 70%. I have climate change, it's between 66 and 90% and sort of, they're not so different. They're sort of all, you know, above 50 and, but not too high.
- [22:21]
Helen Czersky
So, but that's not a very big, I mean it's a very, it's a very broad category, but it's still, it's better than.
- [22:27]
David Spiegelhalter
Better than someone thinking likely means almost certain or something. Even a rough calibration of words is incredibly useful. And actually I got it right here. I don't know if it's. This is the. I gave a talk at MI5 and all I got was this lousy mug. And this is from JTAC, the Joint Terrorism Analysis center in MI5, the ones who decide the terrorism threat levels in the country. And it tells me that if they think a terrorist attack is likely over the next six months, that must mean between 55% and 75% probability.
- [23:00]
Helen Czersky
Right.
- [23:01]
David Spiegelhalter
And those words are used so that, you know, when we get to the threat level. Oh God, I can't remember what it's at. At the moment, but it does say it's likely over the next six months. And that means 55 to 75% chance of a terrorist attack over the next six months. So it's quite high. So that calibration of language to numbers I think is incredibly important to avoid the sort of misunderstandings we know have occurred.
- [23:25]
Helen Czersky
I can't work out whether I'm reassured or not that they put it on a mug so that, you know, if the red light goes off in a corner, somebody can go, oh, right. What does that mean?
- [23:34]
David Spiegelhalter
No, it just says that if my mug is 55 to 75% empty, I am likely to need a refill. So that's the reason. I think it's brilliant. They ought to sell this as merch outside MI5. I think it's a brilliant bit of stuff that sound.
- [23:54]
Helen Czersky
And it's a public health. A public sort of information broadcast as well. I like that. Okay. So some of the risks and probabilities that we deal with in life that we hear most about and I think are most misinterpreted are, are health risks. And you know, this has a chance of a 50% chance of causing cancer or the risk of a cancer increases by 50%. So we're talking about relative and absolute risk here. And there are many headline writers, I feel, in the country who do not understand the difference between relative and absolute risk. Just set out what that is.
- [24:28]
David Spiegelhalter
Yeah, I mean, so it is confusing because they both tend to use the percentage, but the percentage is using it in different ways. And the example I use in all of everything I've been talking about for years is I know bacon sandwiches and you know, bacon sandwich a day is supposed to. And I think, I actually think this is quite plausible on a causal basis that increase your risk of getting bowel cancer by about 20%. And so, but then that's a relative risk. Whatever your risk is before it's increased by a fifth. So it's moved up a little bit whatever your risk before. But to interpret whether that's important or not, we have to know the, the absolute risk, which is what is anyway my percentage chance of getting bowel cancer if I don't eat bacon sandwiches. And that's best expressed. I like the language of percentage points and it's about 6 percentage points. So 6 out of 100 people who don't eat bacon sandwiches will get bowel cancer anyway for other reasons. So that's 6 percentage points. And so compared then if you've got 100 people who don't eat bacon sandwiches, 6 will get bowel cancer. And if you've got 100 people who do eat bacon sandwiches that 6 is raised by 20% which takes it roughly to 7. So 7 people will get would get bowel cancer and those calculations to see how you apply relative risk to an absolute risk in order to get the sort of absolute size of the impact of eating bacon sandwiches like essentially means a 1 in 100 chance a 100 extra probability that you'll get bowel cancer. And is it's a tricky I don't know any journalist who can do that in fact and it's not that true. We've written software to do it. People have got problems. They teach it in schools really quite well is is emphasized a lot. It's so important because it puts things in perspective because you know that's 100 people stuffing some greasy bacon sandwich in their gob every day and from that you have one extra case of bowel cancer. No maybe that's important for the individuals. It might not be that important from a population point of view it can be there's a lot of there's thousands of cases of bowel cancer every year. So it can be relative risk might be actually not very important for an individual and might consider oh that's reasonable my pleasure the trade off and my pleasure counteracts it. But from a public health point of view it is important. So again it depends. You can't just say whether something's important or not. You have to look at the perspective of it.
- [27:01]
Commercial Narrator
Starting or growing your own business can be intimidating and lonely at times. Your to do list may feel endless with new tasks and lists can easily begin to overrun your life. So finding the right tool that not only helps you out, but simplifies everything as a built in business partner can be a game changer for millions of businesses. That tool is Shopify. Shopify is the commerce platform behind millions of businesses around the world and 10% of all e commerce in the US from household names like Gymshark, Rare Beauty and Heinz to brands just getting started. Shopify has hundreds of ready to use templates that can help you build a beautiful online store that matches your brand's style. And you can tackle all the important tasks in one place. From inventory to payments to analytics and more. No need to save multiple websites or try to figure out what platform is hosting the tool that you need. And if people haven't heard about your brand, you can get the word out like you have a marketing team behind you with easy to run email and social media campaigns to reach customers wherever they're scrolling or strolling. Start your business today with the industry's best business partner, Shopify, and start hearing. Sign up for your $1 per month trial today at shopify.com realm. Go to shopify.com realm that's shopify.com realm
- [28:24]
Advertisement Voice
right now get up to 15% off select storage solutions put heavy duty HDX totes to good use, protecting what's important to you. The solid impact resistant design prevents cracking and the clear base insides make items easy to find even when the totes are stacked. Find Select Shelving and Tote storage up to 15% off at the Home Depot. To organize every room in your home from your garage to your attic, visit homedepot.com how doers get more done
- [28:56]
Commercial Narrator
when you need to build up your team to handle the growing chaos at work, use Indeed Sponsored Jobs. It gives your job post the boost it needs to be seen and helps reach people with the right skills, certifications and more. Spend less time searching and more time actually interviewing candidates who check all your boxes. Listeners of this show will get a $75 sponsored job credit at indeed.com podcast that's indeed.com podcast terms and conditions apply. Need a hiring hero? This is a job for Indeed. Sponsored Jobs so good, so good, so good.
- [29:29]
David Spiegelhalter
New summer arrivals are at Nordstrom Rack stores now. Get ready to save big with up to 60% off brands. And like Rag and Bone, Levi's, Adidas and free people.
- [29:39]
Helen Czersky
Join the Norty Club to unlock exclusive discounts. Shop new arrivals first and more. Plus, buy online and pick up at
- [29:45]
Commercial Narrator
your favorite rack store for free.
- [29:47]
David Spiegelhalter
Great brands, great prices. That's why you Rack.
- [29:53]
Commercial Narrator
This episode is brought to you by Palmolive. Family time isn't just the big moments. It's weeknight Dinners, sitting around the table, everyone talking all at once. So when the plates are empty and the sink is full, use Palmolive Ultra. Palmolive's most powerful formula removes up to 99.9% of grease, leaving your dishes sparkling clean. And the new convenient pump makes cleaning even easier so you can spend less time tackling dishes and more time together. Shop now@palmolive.com.
- [30:29]
Helen Czersky
This particular time in history, it feels like, is a very difficult time to have this discussion because we are living in a world where everything seems to be getting faster and to be condensed down to memes and little videos. And you know, it. It feels. I certainly feel as though there are. There are certainly. I mean, hopefully there are people listening to this podcast or watching it who are keen and interested to go deeper into ideas. But it feels that the world is trapped almost trying to take that away from us. Everything has to be a snap judgment. Is it this or is it this? Yes. No, move on. Do you think we're getting in that context? Are we getting better or worse at assessing risk and uncertainty and nuance? I guess all the nuance in all of these questions.
- [31:11]
David Spiegelhalter
Well, it depends where you look. I see a lot of really good writing and nuance material on social media and in mainstream media. I think actually the mainstream media reporting of this kind of stuff has got a lot better. You don't see as many cats cause cancer sort of stories as there used to be. There are still in some newspapers, but you don't see that many anymore. The journalists have got a lot better. The press offices have got a lot better. They don't just write, you know, the headlines promote really shabby studies. It's still. They're still there. It's still there. But I don't. I think it's actually improved quite a lot. Covid definitely. I think improved show demonstrated the appetite of people for well reasoned information and statistics. There was huge appetite and it showed that people can actually understand things quite a lot. No, there's always is the, you know, impulse to do quick and dirty thing. Oh, there is a big number. This sounds impressive and always a big claim. It's almost certainly wrong if it's that impressive. It's almost certainly wrong. So it means you don't believe any number that sounds too impressive and put
- [32:13]
Helen Czersky
that on a T shirt can't you don't believe any number that sounds too impressive.
- [32:16]
David Spiegelhalter
Yeah, exactly. Exactly. Yeah, yeah, yeah, yeah, yeah. It's too good to be true. So. So I, I think it's a balance of stuff but I, I do. I. I'm not so pessimistic actually thinking think education is getting better in this regard. I think kids are seeing everyone's still bombarded by nonsense all the time. But I think actually the, the maybe because people especially with AI coming along where people just have to. Everybody is going to have to become more discriminating because there's going to be so much crap kind of being put out there in terms of videos and everything like that people are going to be. Otherwise they'll be embarrassed. I. I think just fear of embarrassment might make people a little bit more cautious about clicking and forward claim.
- [33:00]
Helen Czersky
I hope that is the sort of emotional argument that might work slightly better than some of them.
- [33:05]
David Spiegelhalter
Yeah, yeah. Fear of embarrassment.
- [33:07]
Helen Czersky
Yeah.
- [33:07]
David Spiegelhalter
And that sort of argument has been used by people countering misinformation, using this idea of pre bunking which I really like is. And the pre bunking misinformation is that, that actually the, in a sense experts, whatever should tell people the misinformation. They should be the first ones to give the news. Do you know, you know, some people might think that this means X or might interpret these numbers to mean X and this is why it doesn't mean this. And this is this idea of getting in there first because then people for a start might think oh cool, I know something other people don't which can sounds quite good and also that they don't. People don't want to be embarrassed and they'd rather know things really. I've got this knowledge that actually this doesn't mean this and it's shown to be effective and it's really taking off in government official statistics bulletins increasingly they pretty well almost lead with well, what this data, how you can interpret this data and also right next to it how you should not interpret this data, what you cannot conclude from this data. Some is even a table of clicks and crosses which I think is really lovely way. So. And that's part of the code of the new version, the code of practice of statistics. And I've got a conflict of interest there because I'm on the committee that oversees the body that drew this up. So I'm completely obsessed about the code of practice for statistics, which might not sound that riveting to be honest, I mean, but it's got some great bits in it about, you know, there's a duty now for statistics producers and communicators to preempt misunderstandings. I mean, this is really good to try to head the misinformation off as early as possible. Don't be on the back foot responding when people get completely the wrong end of the stick. There'll always be some people who will do it malevolently. But you want to stop people doing it by accident.
- [34:57]
Helen Czersky
I think it's a, it's really, it's a really interesting task I think for, for statisticians because I think a lot of the time these, the, the outcomes of statistics are so obvious to people who deal with them all the time. It's actually very hard for them to think of how people might misinterpret them. But it's a very good exercise.
- [35:14]
David Spiegelhalter
Yeah, exactly. So that's my thing. First rule of communication is listen, you have to understand your audiences, you have to know how they might misinterpret something because, and I can I tell you a story that I don't think it's, I'm not sure if it's in the book or not. You know, we've done a lot of risk communication about health things and the standard way to do it is to show 100 little people, 100 little dots and you color in the ones who are going to have a heart attack and you, and you leave blank the ones who aren't. You say, well, you know, this is, this represents your risk. And we had people coming back saying, oh, I don't believe that much. I don't, I don't believe that it's only based on the experiences of 100 people.
- [35:49]
Mia Sorrenti
Call
- [35:53]
David Spiegelhalter
it's based on 10,000 K cases. But careful statistic, these are 100, you know, imaginary people in the future. These are purely fictional expert, you know, so it makes you really. Oh, we didn't expect.
- [36:05]
Helen Czersky
But what's brilliant about that is they do understand the problem of sample size. I mean,
- [36:12]
David Spiegelhalter
and it shows our bad communication because we never ever thought that anyone would think that was the data.
- [36:18]
Helen Czersky
Right.
- [36:19]
David Spiegelhalter
And so again, it's only by listening that you find out you've got to make this clear, you've got to explain things because obviously that shows real sophisticated listening. A lot of it, a lot of the stuff is that people, because people don't know how to take numbers apart, they take them at absolute face value without thinking about them, without thinking this number is just ridiculous. They can't assess whether this is a reasonable number, especially once it's got a few zeros on the end. Very difficult. I find it difficult to assess. Is this a big number?
- [36:46]
Helen Czersky
So I, yeah, I think that's one of the biggest. So as a, as a physicist to talk, you know, we, we are physicists are taught that basically on day one is how to make an estimate using to get to a reasonable answer. Like it's not, it's not, you know, to get an order of magnitude answer. And, and that sort of habit I find. And the, the reason I find it's, it's getting harder now is because if I ask my students where I teach, you know, what they think their initial reaction is to ask, ask ChatGPT or something like, you know, it's so easy for them to get an answer and then they just believe it. And like that is exactly why you need to be able to do order of magnitude calculations.
- [37:24]
David Spiegelhalter
Yeah, it's probably not bad, but you're exactly right. And that's the most important thing I think in numeracy education is not, you know, be able to do you want to be able to get the sums right. But it's getting an idea of magnitude, you know, is, is absolutely vital. And at the instrument you've got more than two zeros. Frankly, it gets pretty tricky.
- [37:46]
Helen Czersky
Yeah. Okay, so there's something else that towards the end of your book, I mean we talk, we talked at the beginning about sort of uncertainty and probability as a basis for making future decisions, but the other big place it comes in is about assigning blame for the past. And so for climate. Now we're seeing, you know, big floods, rainstorms and, and climate scientists are fairly sure that this is exactly what you'd expect as climate change increases. So there's these odd statistics which we're starting, which are, you know, that this flood was made 50% more likely. Talk to us about that.
- [38:19]
David Spiegelhalter
Oh, it's fascinating. I'm so glad you asked me about that because I got a chapter on attribution and I think I've never seen anybody who's written and brought the stuff together about that because it started off in legal cases still being used in court. When someone's been exposed to a chemical at work and then they get a cancer later and you want to work out on the balance of probabilities, was their cancer caused by the exposure? And people do interpret that as being more than a 50% chance. And you can't tell that specific individual. But by looking at the epidemiology, what is the relative risk associated with this exposure? You can assess whether on the balance of probabilities it was caused. And you do that by looking at how much was the risk increased by the exposure. And those ideas of attribution have been carried, carried straight over into the climate community. The Met Office has got an attribution center, quite a rapid attribution center. That will do. The example I use in the book I think is a very hot summer in September 23 or something like 24 or something 23 it must have been, and really hot summer peak temperatures. So what they do is run a model which assumes there was not man made impact on the climate and talk about how likely such a high temperature was. And then run another model taking into account man made impact on the climate. See how likely the model was and the ratio likely it was to observe something as extreme as that. And the ratio of that essentially is how much more likely this was to have occurred given with climate, with man made input into the climate rather than without that. And that you can convert that into a probability that it was caused by climate change. Met Office don't do that. Some people do around the world because that's quite a strong thing to say. But that relative risk, how much more likely, is the basis for legal decisions and could very well become the basis for future claims against fossil fuel companies. So this idea of attribution is already part of the civil legal context on individual health effects. This is likely to be a big issue in the future and will bring even more scrutiny to these climate models. So it's actually a really simple idea and that's been around for years in medicine and so it's just fascinating to see it being applied in this new area.
- [40:47]
Helen Czersky
Yeah, well, I think, and, and a lot of, I think the problem, the reason climate science is, is like it is that a lot of, a lot of us spend our time saying, trying to convince people about distant, vague, far off effects that might happen in 10 years time. Whereas if it's, if it's just happened, it's very much on people's minds and it's a very useful tool.
- [41:07]
David Spiegelhalter
The example I use is from the uk. I remember the hot summer, autumn is lovely hot September. So you can use things that actually, as you said, are directly part of people's experience. They're recent, they're local and their past, rather than saying some claim about how things are going to be bad in 50 years time, which frankly is quite difficult for people to grasp, especially if they're going to be dead like I will be. And so I think it's really an important part. You've got to be careful. There's been some big mistakes where people have said that, oh, this climate change is making this hurricane 50% stronger, things like that, and got a lot of coverage and they've had to retract it later. So, so you've got to be quite cautious about this modeling. You've got to be really careful.
- [41:54]
Helen Czersky
It's a very difficult thing to do. I mean, yeah, you have to do it. And of course predicting things that didn't happen is always by definition either predicting things that haven't happened or modeling things that didn't happen. Like you can't actually test it. So there's a whole thing.
- [42:08]
David Spiegelhalter
Well, you just hope these models are reasonably calibrated in terms of, at the moment they are vaguely predicting what's happened. So you know how extreme this event actually is because you're talking about tail areas of a distribution.
- [42:22]
Helen Czersky
So yeah. And of course being able to think about uncertainty in a quantitative way is Important. We are coming to the end of the time we have got here. So I just want to. As a sort of final question, I'm curious about, are there uncertainties you worry about? Are you now reconciled with uncertainty and you sort of welcome it? Like how, you know, you. You've spent your career thinking about all this. What's your. What's your wisdom? Hold on.
- [42:50]
David Spiegelhalter
As I say, from the right, from the beginning, this is deeply personal, deeply personal, because it's a relationship about us, us with the outside world. And so, you know, from my personal point of view, yeah, I'm 72. I've actually got locally advanced prostate cancer. So there's a lot of uncertainty about my personal future. But there would be. Anyway, I'm not going to go on forever, so I know that if I live another 10 years, I think I'll be pretty damn good. In fact, I'll be really lucky. I don't think I'm going to live 20 years. So I've got this sort of. There's quite a finite horizon sitting there, which when I think about it, I get sad because I won't be seeing family and things like that. But actually, I don't mind too much about dying, but there's a lot of uncertainty and I prefer that it's uncertain. In fact, I love this uncertainty, uncertainty, not knowing. It's great in terms of bigger picture stuff. God, the uncertainty associated with the impact of AI, which is. And the impact of AI and climate and actually geopolitics is just so massive, what's going on? It's not quite. I don't feel we're at quite the level of risk as we are in the 1930s, which was. If you wanted an age of uncertainty in the 1930s, or risk and danger, the 1930s was there. And of course, it ended up in a complete catastrophe, essentially. So I don't think we're quite at that stage yet, but we are living in an age of uncertainty, I think. But it's very difficult to put numbers on these uncertainties. And in these situations, what people call deep uncertainty, where you can't even list the possibilities of what might happen. You've got this imagination. I mean, I often say I use it in my talks. I show a picture of a magician and then a picture of Donald Trump. Trump. And say, what have these two got in common? And someone says, misdirection, which I thought is really clever. That's not what I was thinking of. For each, you've got to expect surprises. You go and see a magician, you know, you're going to be surprised that whatever he does isn't what you thought because of misdirection. Trump's the same. You've got to expect surprises. So in these situations, we can't optimize, we can't put numbers on the probability on the events unless they're very finely defined events, as of course you get in the betting exchanges like Polymarket. But you so you need resilience and deep uncertainty, meaning that you have to act in a way that's going to be not too bad across a wide range of circumstances, including ones you've never even thought of. So that's why. And so you can't optimize. You're always going to have redundancy, you need insurance, you need all these things because you don't know what's going to happen. And so I think by trying to protect yourself against the downside, that's how you live with uncertainty. And once you've done that, I think one should take risks. Don't be reckless. Take risks. Be bold. For goodness sakes, we're only here once.
- [45:39]
Helen Czersky
Well, this is your book is the Art of Uncertainty for the Age of Uncertainty. David Spiegelhalter, It's a pleasure to speak to you. Thank you for joining us on Intelligence Squared.
- [45:49]
David Spiegelhalter
Oh, it's been great. Thank you so much. Lovely.
- [45:53]
Mia Sorrenti
Thanks for listening to Intelligence Squared. This episode was produced by me, Mia Sorrenti and it was edited by Mark Roberts. For ad free episodes and full length recordings, you can become a member at intelligencesquared.com forward/membership and if you'd like to join us at any future live events, you can find our full program or buy tickets@intelligencesquared.com forward/attend. You've been listening to Intelligent Squared. Thanks for joining us.
- [46:36]
Advertisement Voice
Uncovered windows can make your home feel up to 20 degrees hotter. Stay cool and save up to 50% off custom window treatments. During the 4th of July mega mega sale@blinds.com from outdoor shades to room darkening blinds, finding the perfect fit is easy. Get free samples, expert design help and professional measure and install services or DIY with confidence and support every step of the way. Shop up to 50% off site wide plus huge savings on door busters right now. During the 4th of July mega sale at blinds.com Athletic Brewing Company Crafts Award winning non alcoholic beers for those who want to be part of everyone with over 185 flavor awards. They're exceptional NA beers that fit your lifestyle and any social occasion. Summer's full of good times and athletic fits right in. Go to athleticbrewing.com to have brews delivered to your door or find them at a bar, restaurant or store near you near Beer Athletic Brewing Co. Fit for all times. Hey, it's Ryan Reynolds here from Mint Mobile.
- [47:36]
Mia Sorrenti
Now.
- [47:37]
Advertisement Voice
I was looking for fun ways to tell you that Mint's offer of unlimited premium wireless for $15 a month is back. So I thought it would be fun if we made $15 bills, but it turns out that's very illegal. So there goes my big idea for the commercial. Give it a try@mintmobile.com switch upfront payment
- [47:57]
Helen Czersky
of $45 for three months, $90 for six months or $180 for a 12 month plan required $15 per month equivalent taxes and fees Extra initial plan term only greater than 50 gigabytes. Me slow when network is busy.
- [48:04]
David Spiegelhalter
See terms all new drinks are now at McDonald's with refreshers like the Strawberry Watermelon Refresher and the Mango Pineapple Refresher with Popping Boba to crafted sodas like the Sprite Berry Blast with berry flavors and cold foam. Who knew ice cold drinks could be so fire six? All new drinks are here now at McDonald's. Refreshers contain caffeine.