AI transcript
0:00:00 Hey everybody, I’m Kai Rizdal, the host of Marketplace, your daily download on the economy.
0:00:16 Money influences so much of what we do and how we live.
0:00:19 That’s why it’s essential to understand how this economy works.
0:00:24 At Marketplace, we break down everything from inflation and student loans to the future
0:00:27 of AI so that you can understand what it all means for you.
0:00:32 Marketplace is your secret weapon for understanding this economy.
0:00:35 Listen wherever you get your podcasts.
0:00:37 Hey, it’s Jacob.
0:00:42 I recently was a guest co-host on another Pushkin podcast.
0:00:46 The show is called Cautionary Tales.
0:00:48 The host is Tim Harford.
0:00:50 And it was a lot of fun.
0:00:51 And I thought you might enjoy hearing it while what’s your problem is on a break.
0:00:55 So we’ll be back with what’s your problem in a few weeks.
0:00:57 In the meantime, I hope you like this episode of Cautionary Tales.
0:01:01 Let’s play around.
0:01:05 Warder Studios, it sounds very elven.
0:01:09 I always think of David Bowie when I think of Wardog in one of his early songs talks
0:01:13 about bright lights, Soho, Wardog Street.
0:01:16 Bowie himself has an elven aspect.
0:01:20 He certainly does.
0:01:21 His breakthrough gig at Aylesbury, my hometown, which basically is about the only thing
0:01:26 remarkable about Aylesbury.
0:01:27 Well, and also home of Tim Harford as the plaque says.
0:01:31 Yeah, they’ve got a statue for Bowie, not yet for me.
0:01:34 I don’t know why.
0:01:36 Should we go? OK, let’s go.
0:01:38 OK, I’m ready.
0:01:39 Hello, and welcome back to another episode of Cautionary Questions, our first of 2024.
0:02:07 I am, of course, Tim Harford.
0:02:09 You are our loyal listeners.
0:02:11 You’ve been sending in your burning questions on money, technology, economics and problem
0:02:16 solving. And thank you so much to everyone who’s done so.
0:02:20 And today is the day I do my best to answer them.
0:02:24 And thankfully, I won’t be alone in this endeavor here to help me out both with the questions.
0:02:29 And I think with some of the answers is the brilliant, brilliant Jacob Goldstein, the
0:02:35 host of Pushkin Podcast What’s Your Problem, author of the book Money, the True Story of
0:02:41 a Made-Up Thing.
0:02:42 Jacob was my inaugural Cautionary Questions co-host.
0:02:45 Jacob’s wonderful to have you back.
0:02:47 Tim, it’s an honor.
0:02:48 Hi.
0:02:49 So we should just get on with the questions because we always have so many and so much
0:02:52 to say.
0:02:53 So what have you got in your big bag of listener questions for me?
0:02:57 Let’s start with a question from Alex in Melbourne, Australia.
0:03:01 Alex writes, “I work with artificial intelligence, image generation software, almost daily now.
0:03:08 And I’m quickly seeing how so much of my workforce and processes can either be sped
0:03:13 up or entirely replaced by AI.
0:03:15 And this makes me nervous.”
0:03:17 And then he asks about universal basic income, right?
0:03:21 This idea of a government giving all its citizens money.
0:03:24 And he says, “It seems for the first time that computers and software will actually
0:03:28 replace jobs in a deeply concerning way, which is both exciting and terrifying.
0:03:33 What are your thoughts on UBI, universal basic income, as a solution to an AI crisis and
0:03:38 the widespread philosophical and economic implications of this?”
0:03:43 I love this question.
0:03:45 I mean, it’s so big.
0:03:47 And I think the first thing to say is we don’t really have any idea if what Alex is thinking
0:03:53 about comes true, and if most people just have no economic value.
0:03:58 They have value as human beings, have value as members of society, but there’s nothing
0:04:01 that they could actually sell their labor to do, then that’s completely uncharted territory.
0:04:07 We’ve never been anywhere like that before.
0:04:10 So everything we do is kind of speculative.
0:04:12 We’ve feared it for a long time, right?
0:04:15 We’ve had 200 years of being afraid of technological unemployment.
0:04:20 And my prior on this is to be somewhat skeptical, right?
0:04:25 Like clearly there can be some large number of people who lose their jobs, and we should
0:04:30 be concerned about that, and we should think about how to mitigate that.
0:04:33 But the idea of more or less everybody losing their jobs, I’m skeptical of for the simple
0:04:38 reason that it hasn’t happened in 200 years of incredible technological progress.
0:04:42 And right now, after decades of extreme technological progress in the U.S., unemployment is below
0:04:48 a percent.
0:04:49 The share of working-age people who are working is near all-time highs.
0:04:54 And so somehow we keep coming up with new things to do for money, no matter how many
0:05:00 things computers can do.
0:05:01 And so my first thought is, I don’t think we’re going to have everybody losing their
0:05:06 jobs to AI.
0:05:07 I definitely could be wrong, but that’s what I think.
0:05:09 No, I think that’s a good working assumption.
0:05:12 If you think back a few centuries, basically almost all the labor that people did, they
0:05:17 might wash their clothes occasionally, or that’s been outsourced to the washing machines.
0:05:21 They’d spend a lot of time moving water around, just drinking water, cooking water, throwing
0:05:26 out human excrement.
0:05:28 That’s all now handled by automated systems.
0:05:30 Digging, people did a lot of digging, right, pulling a plow.
0:05:34 Almost everything we used to do is now done by machines, but somehow we still all have
0:05:39 jobs.
0:05:40 Let’s at least accept the premise that maybe this time it might be different.
0:05:44 Because the robots are doing everything, there’s still material prosperity, there’s food out
0:05:48 there, in all the services we could possibly want.
0:05:51 We just have to find some system whereby the humans who have no economic value get to enjoy
0:05:57 all this cool stuff that’s being produced by the machines.
0:06:00 Yeah.
0:06:01 And you know, technological prosperity has given us in the developed world, UBI for old
0:06:08 people, right, in the US, as in every developed country as far as I know.
0:06:13 Once you get to some age, if you are a citizen and you have worked, the government gives
0:06:18 you money every month until you die, right.
0:06:21 And so you could imagine a kind of creeping extension of that.
0:06:26 Certainly right now, they’re not talking about lowering the retirement age in the US, they’re
0:06:29 talking about raising it, as in many countries.
0:06:32 Yeah, raising it in the UK as well, but it’s still a long way away from life expectancy.
0:06:36 My recollection is that when Bismarck introduced the first pension, which was in Germany in
0:06:41 the late 19th century, I think the pension started to be paid at the age of 67 and the
0:06:49 life expectancy was 63.
0:06:51 You get negative four years in expectation.
0:06:54 People who survive long enough to claim any pension at all are already exceptional.
0:06:57 So that would be like having a pension today that started at age 90.
0:07:02 Like some people will get it, but most people won’t.
0:07:04 But actually a lot of people could easily collect 30 years of pension, certainly 20 years.
0:07:09 So they’re living a large proportion of their adult life receiving money from the state
0:07:14 and they’re also receiving money from their own savings, their own investments, which might
0:07:18 be a replacement for UBI.
0:07:21 Maybe we all just have shares in the robots instead.
0:07:23 We have shares in Google or whatever, and that’s how we get paid.
0:07:26 Yes, whether that is mediated by the government or not looks somewhat different, right.
0:07:30 Either the government is taxing the owners of Nvidia and Google stock and distributing
0:07:35 the money or everybody owns Google and Nvidia stock.
0:07:39 There is the piece of this, which is about the non-financial parts of work, right.
0:07:45 Yes.
0:07:46 I mean, there’s a political economy piece, the sort of bridge, how do we get from here
0:07:49 to there if that happens?
0:07:50 Yeah.
0:07:51 And that is complicated and maybe ugly.
0:07:52 If the robots can do all of the stuff, the computers can do it, there’s no economic
0:07:57 reason why humans couldn’t just receive whatever an allowance, they’re given 10 robots or they’re
0:08:03 given $10,000 a month to spend or whatever they like, there’s no economic reason why
0:08:08 that couldn’t happen.
0:08:09 But I think what you’re getting at is what does it do to us if we’re in that situation?
0:08:13 Yeah.
0:08:14 Yeah.
0:08:15 I don’t want to not have a job.
0:08:16 I recognize that I am fortunate to have a job that I enjoy, that I derive a part of my
0:08:22 identity from.
0:08:23 I recognize that for many, in fact, probably most people, a job is not that.
0:08:27 It’s some unpleasant thing they do because they need the money.
0:08:31 If they got more money, they would quit their job.
0:08:34 Well you probably know the kind of empirical evidence on this better than I do.
0:08:37 People have looked at lottery winners and my sense is it’s not great for you to win
0:08:42 the lottery and quit your job.
0:08:43 It doesn’t actually make you happier.
0:08:45 Actually the evidence on lottery winners is a bit mixed and I think slightly here all
0:08:49 the disastrous stories of lottery winners.
0:08:51 Actually it’s great to win the lottery, that’s amazing.
0:08:53 Is that true?
0:08:54 Tell me.
0:08:55 I think it’s fine.
0:08:56 Yes.
0:08:57 It’s just fine to win the lottery.
0:08:58 It’s not really a problem to win the lottery.
0:08:59 Okay.
0:09:00 I wrote about this a couple of years ago.
0:09:02 But we may not all want a job, but we all want something to do.
0:09:08 We all want to feel useful.
0:09:11 We all want a sense of some kind of purpose.
0:09:14 We all want, I think, that experience of mastery, that experience of knowing that you can do
0:09:17 something that not everybody can do.
0:09:20 Those things don’t have to come from a job, but for a lot of people they do come from
0:09:23 a job.
0:09:24 Yeah.
0:09:25 It’s like narrow provincial experience of life.
0:09:28 It’s frankly hard to imagine getting those things without a job.
0:09:32 Yeah.
0:09:33 It comes back to my initial reaction to Alex’s terrific question, which is this is such unknown
0:09:37 territory that we can only really speculate.
0:09:40 But I think you and I, Jacob, are in agreement that the fundamental issue here is not economic.
0:09:45 It’s really to do with our souls.
0:09:47 How would we react if our desire for mastery, our desire for meaning, our desire to feel
0:09:52 useful, if that had to be satisfied without having a job?
0:09:57 And what would we do?
0:09:58 And could we cope?
0:09:59 And I don’t know.
0:10:00 I mean, Harford, if the robots come and take our jobs, let’s just you and me make a podcast
0:10:04 for free.
0:10:05 I’m in.
0:10:06 I’ll commit right now in the Robot Utopia Apocalypse to making a weekly podcast with
0:10:12 you for free.
0:10:13 Deal.
0:10:14 Sorry.
0:10:15 We have no more time.
0:10:16 You must give in to your robot overlord.
0:10:19 Give us another question.
0:10:21 Hello, Rice Abinav.
0:10:24 I would like to ask with the onset of AI, what is the next cautionary tale you anticipate
0:10:30 talking about in the years to come?
0:10:32 Hello, Abinav.
0:10:33 I have just finished the first draft of a script about the coming of AI.
0:10:38 And without giving too much away, the cautionary note is about what happens when automation
0:10:46 gets so good that we lose our own skills.
0:10:49 We hand over control to the machine.
0:10:52 And then how do we respond when the machine says, actually, this one’s too hard.
0:10:57 Could you take over again?
0:10:58 Suddenly, we’re back at the wheel and we’re out of practice.
0:11:01 Harford, it’s not another plane crash.
0:11:04 Is it?
0:11:05 Don’t you have a moratorium on doing plane crashes at this point because it sounds like
0:11:07 that.
0:11:08 I’m going to say nothing more.
0:11:09 Fair enough.
0:11:10 Let’s go for a break.
0:11:11 Right.
0:11:12 Right.
0:11:13 When it comes to ensuring your company has top notch security practices, things can get
0:11:23 complicated fast.
0:11:25 Vanta automates compliance for SOC2, ISO 27001 and more, saving you time and money.
0:11:31 With Vanta, you can streamline security reviews by automating questionnaires and demonstrating
0:11:36 your security posture with a customer facing trust center.
0:11:40 Over 7,000 global companies like Atlassian, FlowHealth and Quora use Vanta to build trust
0:11:46 and prove security in real time.
0:11:48 Listeners can claim a special offer of $1,000 off Vanta at Vanta.com/Special.
0:11:54 That’s V-A-N-T-A.com/Special for $1,000 off Vanta.
0:12:02 Hey everybody, I’m Kai Rizdal, the host of Marketplace, your daily download on the economy.
0:12:07 The economy influences so much of what we do and how we live.
0:12:11 That’s why it’s essential to understand how this economy works.
0:12:15 At Marketplace, we break down everything from inflation and student loans to the future
0:12:19 of AI so that you can understand what it all means for you.
0:12:24 Marketplace is your secret weapon for understanding this economy.
0:12:26 Listen, wherever you get your podcasts.
0:12:33 We’re back.
0:12:34 Welcome to Jacob Goldstein, the host of What’s Your Problem.
0:12:38 We are doing a cautionary questions Q&A episode.
0:12:43 Jacob has the questions.
0:12:44 Jacob’s also helping with the answers.
0:12:45 I’m really surplus to requirements here, but I’m doing my best.
0:12:48 Jacob, what have you got for me?
0:12:50 Tim, our next question comes from Adam, who writes, “Hi, I have a question about investing
0:12:56 in cryptocurrency.”
0:12:57 Oh dear.
0:12:58 Oh dear.
0:12:59 Come on.
0:13:00 Give it to him.
0:13:01 Okay.
0:13:02 It’s an interesting one.
0:13:03 When buying shares in a company, you would be supporting that company to grow, create
0:13:08 a new product, or enter a new retail space.
0:13:11 This would hopefully create jobs and new products.
0:13:14 When people invest money into cryptocurrency, or any currency for that matter, isn’t that
0:13:19 money just sitting around not doing anything until you withdraw it?
0:13:23 Would investing in crypto be bad for the economy compared with investing in businesses?
0:13:29 This is a deep question.
0:13:31 I really like it.
0:13:32 It is.
0:13:33 As the author of the wonderful book, Money, the True Story of a Made-Up Thing, I am sure
0:13:38 you have thoughts, Jacob.
0:13:39 Let me have her first crack and you can tell me everything I’ve missed.
0:13:42 Yeah, yeah.
0:13:43 Let’s say you buy Bitcoin or whatever, and then the money’s just sitting there because
0:13:47 Bitcoin is not building anything.
0:13:49 You’re not buying investment in, say, a road building company, or you’re not buying an
0:13:53 investment in Google, which is developing new technology.
0:13:56 There’s one of two possible things that happens.
0:13:59 Most likely is whoever you bought the Bitcoin from now has your money, well, now it’s their
0:14:06 money, and then they’re going to do something with that money.
0:14:08 Maybe they then buy shares in Google, or they then set up a business, or they will then
0:14:13 go on to do something with that money, or they’ll lend it to somebody else and that
0:14:17 person will do something with it.
0:14:19 Eventually, the money will find its way into some productive investment.
0:14:23 The fact that you bought Bitcoin off somebody does not mean that they won’t then do something
0:14:27 useful with the money.
0:14:28 But then, let’s say for some reason, they just go, “You know what?
0:14:31 I’m just going to sit on this money.
0:14:32 It’s just dollars.
0:14:33 I’m going to stick them under the mattress.”
0:14:34 I guess a Bitcoin guy might do that.
0:14:38 That would reduce inflationary pressures on the economy, which is something we want anyway.
0:14:43 If inflationary pressures are reduced too much, the Federal Reserve could then go, “You know
0:14:49 what, guys?
0:14:50 We should probably print some money or otherwise stimulate the economy.”
0:14:53 Actually, the worst thing that could possibly happen is that the money gets invested in
0:14:58 buying more computers that could be put to better use, but in fact, just end up doing
0:15:02 Bitcoin mining.
0:15:03 I guess what I’m saying is, as long as they don’t spend it on Bitcoin mining computers,
0:15:07 it’s totally fine.
0:15:08 What am I missing?
0:15:09 I went around the same track as you went around, and different cryptocurrencies have different
0:15:15 methods of regulation, so Bitcoin by design is this very energy-intensive computing process
0:15:22 in order to mine Bitcoin.
0:15:24 That is, in my opinion, a bad outcome socially.
0:15:27 I agree.
0:15:28 Because you could imagine a world where maintaining the Bitcoin network was socially
0:15:32 desirable in the world as it has evolved.
0:15:34 I don’t find it particularly socially desirable, but at this point, I don’t think most of the
0:15:38 money going into Bitcoin is going to miners.
0:15:41 There’s a large stock of Bitcoin that exists in the world.
0:15:44 An interesting thing about money is it almost never sleeps.
0:15:49 It keeps going, and so you can always ask, “What happens next?”
0:15:54 The stock market, sometimes people talk about money on the sidelines when they’re like,
0:15:58 “Well, the stock market can go up some more because there’s a lot of cash on the sidelines.”
0:16:02 There’s this famous billionaire investor, Cliff Asnes.
0:16:05 He’s an interesting guy.
0:16:06 He got a PhD from the University of Chicago, studied with Gene Fama, a famous economist,
0:16:11 and Cliff Asnes gets driven up the wall when people say cash on the sidelines for the same
0:16:16 reason you just said.
0:16:18 Every time somebody buys a share of stock, someone else is selling it.
0:16:23 The money is changing hands.
0:16:25 The stock is changing hands, but there are no sidelines with money.
0:16:30 Even if someone puts just cash in a checking account, well, then now the bank has money
0:16:36 and you can bet the bank is going to want to do something with that money, something
0:16:39 productive.
0:16:40 Even if they don’t put it in the bank, even if, as I say, it’s under a mattress, well,
0:16:45 even then, the Federal Reserve could always print money.
0:16:48 The stock of money is not fixed.
0:16:50 I think there’s an assumption in the question.
0:16:53 I think goes back to the pre-modern era, the era before the 1930s, when money was based
0:17:00 on gold or silver and it was finite.
0:17:03 There was a fixed amount of gold and silver in the world.
0:17:05 In those days, indeed, if you sat on money, you were effectively reducing the supply of
0:17:11 money in the world, but to your point, the supply of money in the world is as big as
0:17:15 the central bank wants it to be.
0:17:18 It’s not a meaningful constraint on economic growth.
0:17:21 The ultimate answer to the question is, if you are just sitting on money, it’s not really
0:17:25 going to have an effect on the broader economy.
0:17:28 I think we’ve come to agreement.
0:17:29 Hopefully, that makes sense to Adam.
0:17:32 Next question.
0:17:33 Our next question comes from Dr. Yvonne Couch, who is Associate Professor of Neuroimmunology,
0:17:41 Alzheimer’s Research UK Fellow, Associate Research Fellow at St. Hildes College, St.
0:17:47 Penderi Lecturer at Somerville College, University of Oxford.
0:17:51 Yeah, University of Oxford, I know it well.
0:17:53 One of my sisters went to St. Hildes College and one of my sisters went to Somerville College,
0:17:57 so this feels very much in the family.
0:17:59 What does Dr. Couch got to say?
0:18:02 She writes, “My question was vaguely economics-based, although probably not very.
0:18:08 Is the way we currently fund science feasible going forward, do we just have too many scientists
0:18:15 and not enough resources?”
0:18:17 All the best, Yvonne.
0:18:19 Oh, love it.
0:18:21 Let me throw out a few thoughts then.
0:18:22 I think that probably there is a problem with the way we fund science, but there’s no one
0:18:28 way that we fund science.
0:18:29 You have university research, you have various grants, various sources of funding, philanthropy
0:18:33 and so on.
0:18:35 You also have private sector research, which is often incentivized by the patent system,
0:18:42 and then you have big block grants that are handed out by agencies such as the National
0:18:46 Institutes for Health.
0:18:49 So there are lots of different ways that science gets funded.
0:18:53 A couple of things that worry me is that, first of all, there’s a big incentive to make really
0:19:00 incremental improvements rather than to take risks.
0:19:05 There’s a great economics paper studying scientists who are funded by the National Institutes
0:19:11 for Health, which is a wonderful institution, very important.
0:19:14 And scientists who on paper seem to be on the same career track, they’ve got very similar
0:19:20 publication records.
0:19:22 And they were instead funded by a private foundation called the Howard Hughes Medical
0:19:27 Institute.
0:19:28 Listeners who want to hear more about Howard Hughes and the Howard Hughes Biography can
0:19:32 go to the back catalog of cautionary tales.
0:19:33 So the Howard Hughes Medical Institute basically takes risks.
0:19:37 It wants people to do something new.
0:19:39 It’s happy with a high risk of failure as long as there’s some chance of a real breakthrough
0:19:45 success.
0:19:46 And this particular paper studying the results that come from these two funding systems basically
0:19:52 finds that the grant funders get what they pay for.
0:19:56 So the National Institutes for Health get a high success rate, but it’s often quite
0:20:00 incremental progress.
0:20:02 And the Howard Hughes Medical Institute has lots and lots of failures, but when it succeeds,
0:20:06 it really succeeds and these are hugely important papers.
0:20:09 And I just feel that we’re not deliberate enough about saying, well, how much of our
0:20:16 funding ecosystem should be aiming a kind of venture capital style for really big wins
0:20:22 and how much should be incremental?
0:20:24 I think those are not the sorts of questions that get asked.
0:20:27 So that’s one of the things that worries me, but what’s your take, Jacob?
0:20:31 One of the things that I have read is that over time, the average age of grant recipients
0:20:36 from the NIH has gone up and up and up.
0:20:39 So kind of just think of this whole universe as getting older and more risk averse and
0:20:44 more kind of bureaucratic.
0:20:46 There is an interesting set of counter pressures, I think, rising up, partly out of Silicon
0:20:52 Valley.
0:20:53 I don’t know if you’ve come across the work of Patrick Collison and his brother.
0:20:58 They’re from Ireland, they found its stripe, their very rich stripe is a big company that
0:21:02 does basically payment stuff online.
0:21:05 They have a really interesting set of kind of philanthropic endeavors around the idea
0:21:11 of progress.
0:21:12 They’re trying to create a kind of field of progress studies that is, it’s very meta.
0:21:18 What are the conditions that best foster technological and scientific progress?
0:21:24 Yeah, but this question from Dr. Couch is very meta.
0:21:27 Yes.
0:21:28 The progress studies, isn’t it?
0:21:29 Yes.
0:21:30 There is this institute in the Bay Area called the ARC Institute that hires leading scientists.
0:21:36 The basic idea is give talented people freedom and money, and encourage them to take big
0:21:44 swings and they are interested, not just in outcomes, but in new tools, again, continuing
0:21:50 on the nerdy thing.
0:21:52 If you take a tool like CRISPR, CRISPR is an intermediate tool that allows people to
0:21:56 cut up a genome, basically.
0:21:57 We just had the first transplant from an animal into a human a few weeks ago because of CRISPR.
0:22:02 We have the first treatments for sickle cell disease because of CRISPR.
0:22:07 I do think there is a wave of people trying to rethink scientific funding.
0:22:15 There is a bottom line aspect to this question from Dr. Couch that I don’t feel like I know
0:22:20 enough to answer.
0:22:21 I’m curious if you do.
0:22:22 I mean, the question is the way we currently fund science feasible going forward.
0:22:26 There is a yes/no version of the answer.
0:22:29 Do you think you know?
0:22:30 No.
0:22:31 Is it?
0:22:32 I don’t know the answer, but I guess I don’t know either.
0:22:35 Another thing that concerns me on this, which is potentially really existential, is the
0:22:41 question of how we fund new antibiotics.
0:22:45 I think this points to a real weakness in the ecosystem of research funding.
0:22:50 If you think about the basic way we develop drugs, the fundamental incentive is the patent.
0:22:56 A drug company spends a lot of money, tries to develop drugs.
0:22:59 Some of them work, some of them fail, a lot of them fail, but in the end you have a drug
0:23:04 and the deal is we give you this artificial monopoly called a patent and it will only
0:23:09 last for a certain amount of time, which is kind of a problem because it takes so long
0:23:12 to develop the drugs.
0:23:13 Maybe the patent’s nearly expired.
0:23:16 You can charge an incredible amount for these drugs for a while and then your patents will
0:23:20 run out and then loads and loads of people will make the same drug and it’ll come down
0:23:24 in value, for example Viagra.
0:23:26 You could sell this for a huge amount of money and now Viagra is off-patent and anyone can
0:23:30 make a generic Viagra.
0:23:32 That’s the incentive that we’ve given to private companies, that you will have this
0:23:36 temporary monopoly.
0:23:38 Now think about antibiotics.
0:23:40 The problem with antibiotics, we have lots of antibiotics that work really well except
0:23:44 the bacteria have figured them out.
0:23:46 Why have they figured them out?
0:23:48 Because we’ve used them, the bacteria develop resistance.
0:23:51 What we really need is new antibiotics that we don’t use.
0:23:57 And now think about how the patent system deals with that.
0:24:00 So you’re basically saying, if you develop a new antibiotic, we’ll give you a temporary
0:24:05 monopoly, but we really need you to just not sell any, don’t sell any of this drug.
0:24:10 Except in cases of dire emergency, it’s like a break glass in case of emergency.
0:24:16 I feel like a bounty, like every econ nerd storyteller loves a good bounty, right?
0:24:21 Think this done with malaria vaccine, which is actually coming along quite well.
0:24:25 You have some universe of people say, we will pay a billion dollars to anyone who comes
0:24:30 up with a new antibiotic that meets this set of criteria, that treats this set of bugs
0:24:36 that are resistant to these existing antibiotics.
0:24:39 And we’ll give you the money and you give it to us and we’ll put it on the break glass
0:24:42 in case of emergency shelf.
0:24:44 You’ve used the word bounty, but the term that is normally used is an advanced market
0:24:48 commitment.
0:24:49 These were proposed most famously by Michael Kramer, who’s a Nobel Prize winner in economics.
0:24:54 And basically the way this prize, this bounty tends to get paid is as a kind of extra payment
0:25:00 on top of each dose you sell.
0:25:02 So we’ll give you extra for every kid that gets vaccinated.
0:25:06 So we’re back to the same problem in that universe.
0:25:08 Back to the same problem.
0:25:09 But the reason we go, well, why does it have to be like that?
0:25:11 It doesn’t have to be like that.
0:25:12 But the reason that they tend to be structured like that is because you need to demonstrate
0:25:16 some kind of market demand.
0:25:18 You need to be willing to buy your product.
0:25:20 If they are, we’ll give you an extra payment for every product you sell.
0:25:25 But that wouldn’t work for antibiotics.
0:25:26 There are so many different ways in which science funding could be said to be broken.
0:25:30 But I think Dr. Couch is right to raise the issue.
0:25:34 We need to do a lot more of this kind of meta thinking about progress studies.
0:25:39 Tim, that was a lot of answer.
0:25:41 Let’s take a break.
0:25:42 Cushionary tales will be back in a moment.
0:25:52 Hey, everybody.
0:25:53 I’m Kai Rizdal, the host of Marketplace, your daily download on the economy.
0:25:57 Money influences so much of what we do and how we live.
0:26:00 That’s why it’s essential to understand how this economy works.
0:26:04 At Marketplace, we break down everything from inflation and student loans to the future
0:26:08 of AI so that you can understand what it all means for you.
0:26:13 Marketplace is your secret weapon for understanding this economy.
0:26:16 Listen, wherever you get your podcasts.
0:26:22 And we’re back.
0:26:23 I’m Jacob Goldstein here with Tim Harford on Cushionary Tales.
0:26:26 Hello, Jacob.
0:26:27 More questions.
0:26:28 What have you got for me?
0:26:29 All right, Tim, I got another one for you.
0:26:32 Comes from Graham in Florida.
0:26:36 Graham writes.
0:26:37 I’m fascinated by errors made by falsely identifying correlation as causation.
0:26:44 What examples of this error stand out to you?
0:26:47 I love the question, but I’m going to slightly sidestep it because I’m worried by this question,
0:26:51 but I think that it’s generally more complicated than simply some fool saw a correlation and
0:27:00 thought it was causation.
0:27:01 You don’t want to talk about sunspots and crop yields?
0:27:04 Well, let’s talk about stalks and babies briefly.
0:27:07 Okay.
0:27:08 A classic, a classic of the genre.
0:27:10 Yeah, the most successful book ever published about statistics, How to Lie with Statistics
0:27:15 by Darrell Hough, includes this example of stalks and babies and shows that there’s a
0:27:20 correlation between the number of stalks and the number of babies.
0:27:23 And there are various ways to demonstrate this correlation.
0:27:26 One way to do it is you just look at national populations.
0:27:29 You go, “Hey, countries with lots of stalks also have lots of babies.”
0:27:32 And there’s a very, very strong correlation.
0:27:34 But of course, the reason is big places like the United States have a lot of room for
0:27:38 stalks and a lot of room for babies.
0:27:40 And small places like Luxembourg or the Vatican City don’t have a lot of room for babies and
0:27:45 don’t have a lot of room for stalks.
0:27:46 So then you go, “Oh, that’s a great example of this mistake.”
0:27:49 The sting in the tale of that story, as I describe in my book, The Data Detective, is
0:27:55 that Darrell Hough, the guy who created that story, then went on to tell the same story
0:28:01 in congressional testimony, saying that there was no compelling evidence that smoking cigarettes
0:28:08 would give you lung cancer.
0:28:10 And it was just like stalks and babies.
0:28:12 And he was actually hired by the tobacco lobby.
0:28:14 It was just correlation.
0:28:15 It was just correlation.
0:28:16 They seemed about the same.
0:28:18 They were like, “Sure, yeah, there are people who smoke and there are people who get cancer,
0:28:21 but no, there’s no causal evidence.”
0:28:23 One theory was cigarettes are so soothing.
0:28:25 And if you have some early symptoms of lung cancer, maybe you soothe that with the soothing
0:28:30 vapors of cigarettes.
0:28:31 I mean, it’s completely ridiculous, but this sort of rhetoric was deployed.
0:28:35 And so one of the things I’m very concerned about in my work on statistics is that it’s
0:28:40 great to be skeptical about statistics and to point out lots of examples of statistics
0:28:44 being misused.
0:28:45 But if that’s all you do, you just get to a kind of nihilistic situation where you’re
0:28:50 just constantly rejecting statistical evidence because, oh, it’s just another of those downlives
0:28:55 and statistics.
0:28:56 But when you look at the real world, I think this gets to be incredibly interesting.
0:29:01 So a real hot topic at the moment is our smartphone’s destroying a generation, basically.
0:29:08 Are our kids having their mental health wrecked?
0:29:10 There’s a new book that basically makes that a new book by a prominent academic.
0:29:14 Yeah, by Jonathan Hyde.
0:29:16 And there have been others by Gene Twenge.
0:29:17 And lots of people have said this.
0:29:19 And the evidence for it is mostly correlational.
0:29:24 Not completely.
0:29:25 There are some experiments, but they’re not that convincing.
0:29:27 None of them are perfect, but there are lots of different ways of measuring it.
0:29:31 And the really interesting evidence basically says, look, there appears to be a mental health
0:29:37 crisis.
0:29:38 The kids seem to get really distressed, particularly the girls, when children have access to social
0:29:44 media on their phones.
0:29:46 And that happens around sometime between 2010 and 2014.
0:29:49 And at the same time, suddenly you’ve got this outbreak of suicidal ideation, self-harm,
0:29:56 poor mental health, and so on.
0:29:57 And that’s correlational evidence.
0:29:59 I don’t completely believe it, but I don’t completely not believe it either.
0:30:02 I think that’s what makes it interesting and makes it important to engage with.
0:30:06 You have to start going, well, what’s the alternative explanation?
0:30:09 Is there something else that happened sometime about 10, 12, 14 years ago that might explain
0:30:15 this mental health distress?
0:30:16 So the timing of the great financial crisis is probably not quite right.
0:30:19 Donald Trump, maybe Donald Trump is upsetting the kids.
0:30:22 Timing doesn’t work.
0:30:24 So partly it does the pattern of the correlation make enough sense to explain this causal story,
0:30:30 which, funnily enough, is basically exactly what the scientists’ finding connection between
0:30:34 cigarettes and lung cancer were doing.
0:30:36 They only had correlational evidence for a long time.
0:30:38 They weren’t running randomized trials saying, you know, half of you smoke and half of you
0:30:42 don’t smoke.
0:30:43 I mean, they couldn’t do that.
0:30:44 They had to look at correlational evidence, and sometimes that’s what we’ve got.
0:30:48 This question got me thinking about the rise in social science of what they call natural
0:30:54 experiments, right, trying to find instances in the real world where you have something
0:31:00 that obviously is not as good as a randomized trial, because you just can’t get that with
0:31:04 large numbers of people in the world, but that gives you some element of randomization
0:31:10 or pseudo-randomization.
0:31:12 Something that allows you to try and make the leap from correlation to causation.
0:31:16 And you know, as far as I know, it goes back to the Vietnam War, right, where in the U.S.
0:31:21 there was a draft lottery, and there were social scientists after the war who looked
0:31:27 and said, oh, look, here are people who are at an aggregate level, very similar on many
0:31:32 dimensions, socioeconomic dimensions.
0:31:34 We can look at people who randomly got drafted versus those who randomly didn’t and who
0:31:40 appear quite similar in the aggregate and see how their outcomes differ.
0:31:44 And that’s a really good example of a natural experiment, because it’s actually very close
0:31:48 to a real experiment.
0:31:49 Yeah, when you get a lottery of the real world, now social scientists flock to it, right?
0:31:54 Similarly, there was one in Oregon some years ago with Medicaid, which is the healthcare
0:32:00 program for low-income people in the U.S.
0:32:03 And Oregon got some new Medicaid funds, and they randomly allocated them to a group of
0:32:08 people over time, right, so that social scientists could say, oh, look, here are people who are
0:32:13 basically identical, some of them got this health insurance and some of them didn’t.
0:32:16 And in that case, the findings were quite interesting.
0:32:19 Medicaid didn’t appear to be as helpful as I would have thought, as the researchers
0:32:24 themselves would have thought, according to the one I interviewed.
0:32:27 Lowered mental health problems, people worried less about money, but like their basic health
0:32:31 outcomes didn’t improve, which nobody would have guessed, right?
0:32:35 And the evidence is quite robust.
0:32:38 When you don’t have those lotteries, as you say, the world is just hard to understand.
0:32:42 I mean, even in some instances where you do have randomized trials, people are complicated,
0:32:47 the body is complicated, society is complicated.
0:32:50 And so, you know, correlation in a certain way, I think, is underrated, like, yes, obviously
0:32:55 it doesn’t equal causation, but it’s a place to start looking, right?
0:32:59 It’s a place to start asking questions.
0:33:00 I think that’s absolutely right.
0:33:02 And we need more randomized experiments.
0:33:04 And there are more opportunities to run them than people seem to think.
0:33:07 For example, one of the things that John Hyde in his book is arguing for is, you know,
0:33:11 shouldn’t have smartphones in schools.
0:33:13 Schools should be phone-free zones.
0:33:15 I could completely imagine a state saying, we’re going to introduce a rule whereby in
0:33:20 all of the schools in the state, no smartphones, strictly forbidden.
0:33:25 You have to put them in a locker when you show up and then unlock them at the end of
0:33:28 the day.
0:33:29 You could introduce that rule and just go, well, we’re going to introduce it for a semester
0:33:33 at random in 50% of the schools, and then we’ll introduce it in the other 50% of the
0:33:38 schools in the next semester.
0:33:40 And we’ll just randomize that because we want to know whether there’s any point in
0:33:43 this experiment or not.
0:33:45 That’s not very difficult to do.
0:33:47 It would create so much information about children’s performance in the classroom, their mental
0:33:52 health, that could inform policy across the world.
0:33:56 That would be the dream.
0:33:57 You know, there is now a robust set of methods, essentially, where social scientists could
0:34:03 look at that state and compare it to neighboring states and see the difference in differences,
0:34:08 as they say, see the change over time.
0:34:11 And that’s an instance where it wouldn’t be as elegant as randomizing within a state.
0:34:15 But I feel like you could start to get pretty good data, even if you did one state compared
0:34:20 to other states.
0:34:21 Yeah.
0:34:22 If you’ve got no experiments, we should run them.
0:34:24 If we don’t run them, there are still ways of making correlation talk.
0:34:28 Should we have another question?
0:34:31 OK, Tim.
0:34:33 We got one more.
0:34:34 We had to have one game-related question for you.
0:34:37 Oh, great.
0:34:38 Yes, you’re welcome.
0:34:39 This one comes from Joost, who writes, “Hi, Tim.
0:34:43 Diving into the board game-shaped flank you’ve now exposed for discussion,” that is a very
0:34:48 gamerish way in, “what board games or board game mechanisms that you enjoy are particularly
0:34:54 insightful on some aspect of real-world economics?
0:34:58 For a unique angle to look at a maybe-nitch problem?
0:35:02 And what non-D&D games are you particularly enjoying right now?
0:35:08 Love the show however many episodes you produce.”
0:35:11 That is a listener we all want.
0:35:13 Thank you, Joost.
0:35:14 Yeah, so kind, Joost.
0:35:15 There are a couple of mechanisms that I do see used quite often that shed important
0:35:20 light on economics.
0:35:21 So one is trading.
0:35:22 A lot of games involve trading.
0:35:24 Now, Monopoly, that classic of sadly not a very good game, there’s a cautionary tale
0:35:30 about the history of Monopoly if people want to listen, Monopoly in theory involves trading,
0:35:35 but in practice not a lot of trading happens.
0:35:37 It turns out that you need to give people a reason to trade.
0:35:43 The great modern game Settlers, now about 30 years old.
0:35:46 By the way, Settlers, I assume that Settlers of Catan as an outsider, not on a first-name
0:35:50 basis with the game.
0:35:52 Settlers of Catan is the game.
0:35:54 If the game that Monopoly wishes it was, it’s like if Monopoly had been designed with a
0:35:59 modern eye to be super exciting, people need different resources and the supply of resources
0:36:06 fluctuates.
0:36:07 So sometimes you’ve got loads of wood, sometimes there’s no wood, so there’s an active incentive
0:36:11 to trade all the time.
0:36:12 Oh, and by the way, if you don’t trade every now and then, the robber baron comes and
0:36:14 takes stuff.
0:36:15 So there’s an interesting insight that trading doesn’t just happen because it’s allowed.
0:36:20 It needs to be some difference in value and perhaps some incentive to get on with it.
0:36:24 So my personal favourite game is Agricola, a wonderful game about developing a farm,
0:36:30 which doesn’t sound promising, but it’s really, really good.
0:36:33 But one of the clever things about Agricola is it uses an auction in an interesting way.
0:36:38 It’s sometimes called a descending clock auction or a Dutch auction.
0:36:42 The prize gets more and more tempting.
0:36:46 So in Agricola, there’s just more and more good stuff on the board.
0:36:49 In a traditional descending clock auction, basically the price is going down and down
0:36:53 and down and down.
0:36:54 And everyone is just sitting there, and then it’s a question of who grabs it first.
0:36:57 The longer you leave it, the better it gets, but then only one person can have it.
0:37:02 It’s very, very elegant.
0:37:03 They sell flowers at Aalsmere in the Netherlands, and it’s just incredibly quick, much, much
0:37:09 quicker than selling, say art, you know, you’re selling a van Gogh.
0:37:12 Huh.
0:37:13 And the price rises and rises and rises.
0:37:14 So auction design is the whole thing, right?
0:37:16 When would you choose a traditional prices going up eBay art style auction in the real
0:37:22 world?
0:37:23 And when would you choose a Dutch prices going down auction?
0:37:25 What do you get in a relative sense out of each one of those?
0:37:28 There’s a couple of differences.
0:37:30 One is that in certain types of ascending auction, you get to see people dropping out
0:37:36 as the price rises.
0:37:37 It depends on the way the auction is designed, but you could imagine an ascending auction
0:37:41 where you just see people going, no, I’m out.
0:37:43 I’m out.
0:37:44 I’m out.
0:37:45 That’s sort of the classic kind of art auction in a movie.
0:37:47 Yeah.
0:37:48 And so you’re learning information as the price rises.
0:37:50 How many people are still interested?
0:37:51 Are there still 10 people interested?
0:37:52 Is it just two people?
0:37:53 Right.
0:37:54 So that generates information.
0:37:55 And by generating information, it might raise the final price overall.
0:38:00 The advantage of the descending auction is it’s just so quick.
0:38:04 You can run an auction in 10 seconds.
0:38:07 It’s a 100,000 tulips.
0:38:09 Here are the tulips.
0:38:10 We’re going to sell them for whatever, 10,000 euros, 9,900, 9,800, 9,000, and it’s literally
0:38:17 a clock.
0:38:18 The clock just runs around showing what the price is.
0:38:21 And then someone presses the button, sold, okay, now bring in the roses.
0:38:25 So it’s very fast.
0:38:26 You would want to have a good idea of the market clearing price if you were going to run that
0:38:30 kind of auction so that you didn’t sell it for too cheap, right?
0:38:33 It’s fast, but you run the risk of selling it for too cheap.
0:38:36 But if it’s just commodity tulips, then you basically know the market clearing price anyway.
0:38:39 Yeah.
0:38:40 It’s almost like a stock market for tulips.
0:38:42 That I think is why it works so well, whereas if it’s a unique work of art, you need to
0:38:46 give maximum information, maximum comfort to people, maybe a little bit of theater as
0:38:50 well.
0:38:51 Yeah.
0:38:52 We’re not in a hurry to run this auction.
0:38:53 The cost of the auction itself is trivial compared to the value of what’s being sold.
0:38:58 Now we go over to Agricola and have a descending auction.
0:39:01 It feels exciting and it also feels like you’re trying to guess what is in your opponent’s
0:39:06 head because they’re not revealing it as they would in a descending price auction.
0:39:10 You have to think, “Oh, what are they willing to pay for it?”
0:39:13 Absolutely.
0:39:14 And it works slightly differently, but functionally what’s going on is that every turn you take,
0:39:19 you get to grab some resource on the board.
0:39:21 And then once you’ve grabbed the resource on the board, no one else is allowed to grab
0:39:24 it until the next round.
0:39:26 And there are several moves you can make in each turn, so you might be able to get two
0:39:30 or three things each turn.
0:39:32 And so, again, you go backwards and forwards, it’s your move and it’s my move, and each
0:39:35 time we want to grab something.
0:39:37 And you might go, “Well, do I want to grab it now or do I want to wait until next turn
0:39:40 where it might be better?”
0:39:41 And you’re trying to figure out, like, what is the thing that the other person is desperate
0:39:45 to have?
0:39:47 Can I just let this pile get bigger because they can’t afford to take it because they’ve
0:39:51 got to use their move on something else?
0:39:53 It’s a very, very clever game.
0:39:55 I do enjoy it.
0:39:56 You’re also asked what non-D&D games I am enjoying.
0:40:01 I am enjoying a modified version of Blades in the Dark, which is a classic role-playing
0:40:06 game, very fast-moving, you get to do Ocean’s Eleven or other kind of heists.
0:40:11 And it’s very modifiable.
0:40:14 We’re speaking on Monday, just yesterday on Sunday, I had a whole bunch of old friends
0:40:18 around to my house, and we just played this game all day, and we had an absolutely terrific
0:40:24 time.
0:40:25 I like your live-in-the-dream, Harford.
0:40:26 Makes me happy.
0:40:27 Living my best life.
0:40:28 Jacob, this has been such fun.
0:40:31 Thank you so much for doing this.
0:40:32 Ah, it was a delight.
0:40:33 I’ll come back any time.
0:40:35 We would love to have you back.
0:40:37 Thank you, everybody who sent in a question.
0:40:39 Sorry we weren’t able to answer all the questions, but we are going to be back with another cautionary
0:40:43 questions episode later this year.
0:40:46 Please do keep the queries coming.
0:40:47 Send them into tales@pushkin.fm, that’s T-A-L-E-S @pushkin.fm, and I will be back with another
0:40:56 cautionary tale in two weeks’ time.
0:40:58 Jacob Goldstein has a wonderful podcast called What’s Your Problem?
0:41:02 Thank you, Jacob.
0:41:05 Thanks, Jim.
0:41:12 Cautionary Tales is written by me, Tim Harford, with Andrew Wright.
0:41:16 It’s produced by Alice Fiennes with support from Marilyn Rust.
0:41:20 The sound design and original music is the work of Pascal Wise.
0:41:24 Sarah Nix edited the scripts.
0:41:27 It features the voice talents of Ben Crow, Melanie Gushridge, Stella Harford, Gemma Saunders
0:41:32 and Rufus Wright.
0:41:34 The show also wouldn’t have been possible without the work of Jacob Weisberg, Ryan Dilly,
0:41:39 Greta Kohn, Litao Moulard, John Schnars, Erics Handler, Carrie Brody and Christina Sullivan.
0:41:47 Cautionary Tales is a production of Pushkin Industries.
0:41:50 It’s recorded at Wardall Studios in London by Tom Berry.
0:41:54 If you like the show, please remember to share, rate and review.
0:42:00 Tell your friends.
0:42:01 And if you want to hear the show ad-free, sign up for Pushkin+ on the show page in Apple
0:42:06 Podcasts or at pushkin.fm/plus.
0:42:17 Hey everybody, I’m Kai Rizdal, the host of Marketplace, your daily download on the economy.
0:42:22 Money influences so much of what we do and how we live.
0:42:25 That’s why it’s essential to understand how this economy works.
0:42:29 At Marketplace, we break down everything from inflation and student loans to the future
0:42:33 of AI so that you can understand what it all means for you.
0:42:38 Marketplace is your secret weapon for understanding this economy.
0:42:41 Listen wherever you get your podcasts.
0:42:43 (upbeat music)
0:00:16 Money influences so much of what we do and how we live.
0:00:19 That’s why it’s essential to understand how this economy works.
0:00:24 At Marketplace, we break down everything from inflation and student loans to the future
0:00:27 of AI so that you can understand what it all means for you.
0:00:32 Marketplace is your secret weapon for understanding this economy.
0:00:35 Listen wherever you get your podcasts.
0:00:37 Hey, it’s Jacob.
0:00:42 I recently was a guest co-host on another Pushkin podcast.
0:00:46 The show is called Cautionary Tales.
0:00:48 The host is Tim Harford.
0:00:50 And it was a lot of fun.
0:00:51 And I thought you might enjoy hearing it while what’s your problem is on a break.
0:00:55 So we’ll be back with what’s your problem in a few weeks.
0:00:57 In the meantime, I hope you like this episode of Cautionary Tales.
0:01:01 Let’s play around.
0:01:05 Warder Studios, it sounds very elven.
0:01:09 I always think of David Bowie when I think of Wardog in one of his early songs talks
0:01:13 about bright lights, Soho, Wardog Street.
0:01:16 Bowie himself has an elven aspect.
0:01:20 He certainly does.
0:01:21 His breakthrough gig at Aylesbury, my hometown, which basically is about the only thing
0:01:26 remarkable about Aylesbury.
0:01:27 Well, and also home of Tim Harford as the plaque says.
0:01:31 Yeah, they’ve got a statue for Bowie, not yet for me.
0:01:34 I don’t know why.
0:01:36 Should we go? OK, let’s go.
0:01:38 OK, I’m ready.
0:01:39 Hello, and welcome back to another episode of Cautionary Questions, our first of 2024.
0:02:07 I am, of course, Tim Harford.
0:02:09 You are our loyal listeners.
0:02:11 You’ve been sending in your burning questions on money, technology, economics and problem
0:02:16 solving. And thank you so much to everyone who’s done so.
0:02:20 And today is the day I do my best to answer them.
0:02:24 And thankfully, I won’t be alone in this endeavor here to help me out both with the questions.
0:02:29 And I think with some of the answers is the brilliant, brilliant Jacob Goldstein, the
0:02:35 host of Pushkin Podcast What’s Your Problem, author of the book Money, the True Story of
0:02:41 a Made-Up Thing.
0:02:42 Jacob was my inaugural Cautionary Questions co-host.
0:02:45 Jacob’s wonderful to have you back.
0:02:47 Tim, it’s an honor.
0:02:48 Hi.
0:02:49 So we should just get on with the questions because we always have so many and so much
0:02:52 to say.
0:02:53 So what have you got in your big bag of listener questions for me?
0:02:57 Let’s start with a question from Alex in Melbourne, Australia.
0:03:01 Alex writes, “I work with artificial intelligence, image generation software, almost daily now.
0:03:08 And I’m quickly seeing how so much of my workforce and processes can either be sped
0:03:13 up or entirely replaced by AI.
0:03:15 And this makes me nervous.”
0:03:17 And then he asks about universal basic income, right?
0:03:21 This idea of a government giving all its citizens money.
0:03:24 And he says, “It seems for the first time that computers and software will actually
0:03:28 replace jobs in a deeply concerning way, which is both exciting and terrifying.
0:03:33 What are your thoughts on UBI, universal basic income, as a solution to an AI crisis and
0:03:38 the widespread philosophical and economic implications of this?”
0:03:43 I love this question.
0:03:45 I mean, it’s so big.
0:03:47 And I think the first thing to say is we don’t really have any idea if what Alex is thinking
0:03:53 about comes true, and if most people just have no economic value.
0:03:58 They have value as human beings, have value as members of society, but there’s nothing
0:04:01 that they could actually sell their labor to do, then that’s completely uncharted territory.
0:04:07 We’ve never been anywhere like that before.
0:04:10 So everything we do is kind of speculative.
0:04:12 We’ve feared it for a long time, right?
0:04:15 We’ve had 200 years of being afraid of technological unemployment.
0:04:20 And my prior on this is to be somewhat skeptical, right?
0:04:25 Like clearly there can be some large number of people who lose their jobs, and we should
0:04:30 be concerned about that, and we should think about how to mitigate that.
0:04:33 But the idea of more or less everybody losing their jobs, I’m skeptical of for the simple
0:04:38 reason that it hasn’t happened in 200 years of incredible technological progress.
0:04:42 And right now, after decades of extreme technological progress in the U.S., unemployment is below
0:04:48 a percent.
0:04:49 The share of working-age people who are working is near all-time highs.
0:04:54 And so somehow we keep coming up with new things to do for money, no matter how many
0:05:00 things computers can do.
0:05:01 And so my first thought is, I don’t think we’re going to have everybody losing their
0:05:06 jobs to AI.
0:05:07 I definitely could be wrong, but that’s what I think.
0:05:09 No, I think that’s a good working assumption.
0:05:12 If you think back a few centuries, basically almost all the labor that people did, they
0:05:17 might wash their clothes occasionally, or that’s been outsourced to the washing machines.
0:05:21 They’d spend a lot of time moving water around, just drinking water, cooking water, throwing
0:05:26 out human excrement.
0:05:28 That’s all now handled by automated systems.
0:05:30 Digging, people did a lot of digging, right, pulling a plow.
0:05:34 Almost everything we used to do is now done by machines, but somehow we still all have
0:05:39 jobs.
0:05:40 Let’s at least accept the premise that maybe this time it might be different.
0:05:44 Because the robots are doing everything, there’s still material prosperity, there’s food out
0:05:48 there, in all the services we could possibly want.
0:05:51 We just have to find some system whereby the humans who have no economic value get to enjoy
0:05:57 all this cool stuff that’s being produced by the machines.
0:06:00 Yeah.
0:06:01 And you know, technological prosperity has given us in the developed world, UBI for old
0:06:08 people, right, in the US, as in every developed country as far as I know.
0:06:13 Once you get to some age, if you are a citizen and you have worked, the government gives
0:06:18 you money every month until you die, right.
0:06:21 And so you could imagine a kind of creeping extension of that.
0:06:26 Certainly right now, they’re not talking about lowering the retirement age in the US, they’re
0:06:29 talking about raising it, as in many countries.
0:06:32 Yeah, raising it in the UK as well, but it’s still a long way away from life expectancy.
0:06:36 My recollection is that when Bismarck introduced the first pension, which was in Germany in
0:06:41 the late 19th century, I think the pension started to be paid at the age of 67 and the
0:06:49 life expectancy was 63.
0:06:51 You get negative four years in expectation.
0:06:54 People who survive long enough to claim any pension at all are already exceptional.
0:06:57 So that would be like having a pension today that started at age 90.
0:07:02 Like some people will get it, but most people won’t.
0:07:04 But actually a lot of people could easily collect 30 years of pension, certainly 20 years.
0:07:09 So they’re living a large proportion of their adult life receiving money from the state
0:07:14 and they’re also receiving money from their own savings, their own investments, which might
0:07:18 be a replacement for UBI.
0:07:21 Maybe we all just have shares in the robots instead.
0:07:23 We have shares in Google or whatever, and that’s how we get paid.
0:07:26 Yes, whether that is mediated by the government or not looks somewhat different, right.
0:07:30 Either the government is taxing the owners of Nvidia and Google stock and distributing
0:07:35 the money or everybody owns Google and Nvidia stock.
0:07:39 There is the piece of this, which is about the non-financial parts of work, right.
0:07:45 Yes.
0:07:46 I mean, there’s a political economy piece, the sort of bridge, how do we get from here
0:07:49 to there if that happens?
0:07:50 Yeah.
0:07:51 And that is complicated and maybe ugly.
0:07:52 If the robots can do all of the stuff, the computers can do it, there’s no economic
0:07:57 reason why humans couldn’t just receive whatever an allowance, they’re given 10 robots or they’re
0:08:03 given $10,000 a month to spend or whatever they like, there’s no economic reason why
0:08:08 that couldn’t happen.
0:08:09 But I think what you’re getting at is what does it do to us if we’re in that situation?
0:08:13 Yeah.
0:08:14 Yeah.
0:08:15 I don’t want to not have a job.
0:08:16 I recognize that I am fortunate to have a job that I enjoy, that I derive a part of my
0:08:22 identity from.
0:08:23 I recognize that for many, in fact, probably most people, a job is not that.
0:08:27 It’s some unpleasant thing they do because they need the money.
0:08:31 If they got more money, they would quit their job.
0:08:34 Well you probably know the kind of empirical evidence on this better than I do.
0:08:37 People have looked at lottery winners and my sense is it’s not great for you to win
0:08:42 the lottery and quit your job.
0:08:43 It doesn’t actually make you happier.
0:08:45 Actually the evidence on lottery winners is a bit mixed and I think slightly here all
0:08:49 the disastrous stories of lottery winners.
0:08:51 Actually it’s great to win the lottery, that’s amazing.
0:08:53 Is that true?
0:08:54 Tell me.
0:08:55 I think it’s fine.
0:08:56 Yes.
0:08:57 It’s just fine to win the lottery.
0:08:58 It’s not really a problem to win the lottery.
0:08:59 Okay.
0:09:00 I wrote about this a couple of years ago.
0:09:02 But we may not all want a job, but we all want something to do.
0:09:08 We all want to feel useful.
0:09:11 We all want a sense of some kind of purpose.
0:09:14 We all want, I think, that experience of mastery, that experience of knowing that you can do
0:09:17 something that not everybody can do.
0:09:20 Those things don’t have to come from a job, but for a lot of people they do come from
0:09:23 a job.
0:09:24 Yeah.
0:09:25 It’s like narrow provincial experience of life.
0:09:28 It’s frankly hard to imagine getting those things without a job.
0:09:32 Yeah.
0:09:33 It comes back to my initial reaction to Alex’s terrific question, which is this is such unknown
0:09:37 territory that we can only really speculate.
0:09:40 But I think you and I, Jacob, are in agreement that the fundamental issue here is not economic.
0:09:45 It’s really to do with our souls.
0:09:47 How would we react if our desire for mastery, our desire for meaning, our desire to feel
0:09:52 useful, if that had to be satisfied without having a job?
0:09:57 And what would we do?
0:09:58 And could we cope?
0:09:59 And I don’t know.
0:10:00 I mean, Harford, if the robots come and take our jobs, let’s just you and me make a podcast
0:10:04 for free.
0:10:05 I’m in.
0:10:06 I’ll commit right now in the Robot Utopia Apocalypse to making a weekly podcast with
0:10:12 you for free.
0:10:13 Deal.
0:10:14 Sorry.
0:10:15 We have no more time.
0:10:16 You must give in to your robot overlord.
0:10:19 Give us another question.
0:10:21 Hello, Rice Abinav.
0:10:24 I would like to ask with the onset of AI, what is the next cautionary tale you anticipate
0:10:30 talking about in the years to come?
0:10:32 Hello, Abinav.
0:10:33 I have just finished the first draft of a script about the coming of AI.
0:10:38 And without giving too much away, the cautionary note is about what happens when automation
0:10:46 gets so good that we lose our own skills.
0:10:49 We hand over control to the machine.
0:10:52 And then how do we respond when the machine says, actually, this one’s too hard.
0:10:57 Could you take over again?
0:10:58 Suddenly, we’re back at the wheel and we’re out of practice.
0:11:01 Harford, it’s not another plane crash.
0:11:04 Is it?
0:11:05 Don’t you have a moratorium on doing plane crashes at this point because it sounds like
0:11:07 that.
0:11:08 I’m going to say nothing more.
0:11:09 Fair enough.
0:11:10 Let’s go for a break.
0:11:11 Right.
0:11:12 Right.
0:11:13 When it comes to ensuring your company has top notch security practices, things can get
0:11:23 complicated fast.
0:11:25 Vanta automates compliance for SOC2, ISO 27001 and more, saving you time and money.
0:11:31 With Vanta, you can streamline security reviews by automating questionnaires and demonstrating
0:11:36 your security posture with a customer facing trust center.
0:11:40 Over 7,000 global companies like Atlassian, FlowHealth and Quora use Vanta to build trust
0:11:46 and prove security in real time.
0:11:48 Listeners can claim a special offer of $1,000 off Vanta at Vanta.com/Special.
0:11:54 That’s V-A-N-T-A.com/Special for $1,000 off Vanta.
0:12:02 Hey everybody, I’m Kai Rizdal, the host of Marketplace, your daily download on the economy.
0:12:07 The economy influences so much of what we do and how we live.
0:12:11 That’s why it’s essential to understand how this economy works.
0:12:15 At Marketplace, we break down everything from inflation and student loans to the future
0:12:19 of AI so that you can understand what it all means for you.
0:12:24 Marketplace is your secret weapon for understanding this economy.
0:12:26 Listen, wherever you get your podcasts.
0:12:33 We’re back.
0:12:34 Welcome to Jacob Goldstein, the host of What’s Your Problem.
0:12:38 We are doing a cautionary questions Q&A episode.
0:12:43 Jacob has the questions.
0:12:44 Jacob’s also helping with the answers.
0:12:45 I’m really surplus to requirements here, but I’m doing my best.
0:12:48 Jacob, what have you got for me?
0:12:50 Tim, our next question comes from Adam, who writes, “Hi, I have a question about investing
0:12:56 in cryptocurrency.”
0:12:57 Oh dear.
0:12:58 Oh dear.
0:12:59 Come on.
0:13:00 Give it to him.
0:13:01 Okay.
0:13:02 It’s an interesting one.
0:13:03 When buying shares in a company, you would be supporting that company to grow, create
0:13:08 a new product, or enter a new retail space.
0:13:11 This would hopefully create jobs and new products.
0:13:14 When people invest money into cryptocurrency, or any currency for that matter, isn’t that
0:13:19 money just sitting around not doing anything until you withdraw it?
0:13:23 Would investing in crypto be bad for the economy compared with investing in businesses?
0:13:29 This is a deep question.
0:13:31 I really like it.
0:13:32 It is.
0:13:33 As the author of the wonderful book, Money, the True Story of a Made-Up Thing, I am sure
0:13:38 you have thoughts, Jacob.
0:13:39 Let me have her first crack and you can tell me everything I’ve missed.
0:13:42 Yeah, yeah.
0:13:43 Let’s say you buy Bitcoin or whatever, and then the money’s just sitting there because
0:13:47 Bitcoin is not building anything.
0:13:49 You’re not buying investment in, say, a road building company, or you’re not buying an
0:13:53 investment in Google, which is developing new technology.
0:13:56 There’s one of two possible things that happens.
0:13:59 Most likely is whoever you bought the Bitcoin from now has your money, well, now it’s their
0:14:06 money, and then they’re going to do something with that money.
0:14:08 Maybe they then buy shares in Google, or they then set up a business, or they will then
0:14:13 go on to do something with that money, or they’ll lend it to somebody else and that
0:14:17 person will do something with it.
0:14:19 Eventually, the money will find its way into some productive investment.
0:14:23 The fact that you bought Bitcoin off somebody does not mean that they won’t then do something
0:14:27 useful with the money.
0:14:28 But then, let’s say for some reason, they just go, “You know what?
0:14:31 I’m just going to sit on this money.
0:14:32 It’s just dollars.
0:14:33 I’m going to stick them under the mattress.”
0:14:34 I guess a Bitcoin guy might do that.
0:14:38 That would reduce inflationary pressures on the economy, which is something we want anyway.
0:14:43 If inflationary pressures are reduced too much, the Federal Reserve could then go, “You know
0:14:49 what, guys?
0:14:50 We should probably print some money or otherwise stimulate the economy.”
0:14:53 Actually, the worst thing that could possibly happen is that the money gets invested in
0:14:58 buying more computers that could be put to better use, but in fact, just end up doing
0:15:02 Bitcoin mining.
0:15:03 I guess what I’m saying is, as long as they don’t spend it on Bitcoin mining computers,
0:15:07 it’s totally fine.
0:15:08 What am I missing?
0:15:09 I went around the same track as you went around, and different cryptocurrencies have different
0:15:15 methods of regulation, so Bitcoin by design is this very energy-intensive computing process
0:15:22 in order to mine Bitcoin.
0:15:24 That is, in my opinion, a bad outcome socially.
0:15:27 I agree.
0:15:28 Because you could imagine a world where maintaining the Bitcoin network was socially
0:15:32 desirable in the world as it has evolved.
0:15:34 I don’t find it particularly socially desirable, but at this point, I don’t think most of the
0:15:38 money going into Bitcoin is going to miners.
0:15:41 There’s a large stock of Bitcoin that exists in the world.
0:15:44 An interesting thing about money is it almost never sleeps.
0:15:49 It keeps going, and so you can always ask, “What happens next?”
0:15:54 The stock market, sometimes people talk about money on the sidelines when they’re like,
0:15:58 “Well, the stock market can go up some more because there’s a lot of cash on the sidelines.”
0:16:02 There’s this famous billionaire investor, Cliff Asnes.
0:16:05 He’s an interesting guy.
0:16:06 He got a PhD from the University of Chicago, studied with Gene Fama, a famous economist,
0:16:11 and Cliff Asnes gets driven up the wall when people say cash on the sidelines for the same
0:16:16 reason you just said.
0:16:18 Every time somebody buys a share of stock, someone else is selling it.
0:16:23 The money is changing hands.
0:16:25 The stock is changing hands, but there are no sidelines with money.
0:16:30 Even if someone puts just cash in a checking account, well, then now the bank has money
0:16:36 and you can bet the bank is going to want to do something with that money, something
0:16:39 productive.
0:16:40 Even if they don’t put it in the bank, even if, as I say, it’s under a mattress, well,
0:16:45 even then, the Federal Reserve could always print money.
0:16:48 The stock of money is not fixed.
0:16:50 I think there’s an assumption in the question.
0:16:53 I think goes back to the pre-modern era, the era before the 1930s, when money was based
0:17:00 on gold or silver and it was finite.
0:17:03 There was a fixed amount of gold and silver in the world.
0:17:05 In those days, indeed, if you sat on money, you were effectively reducing the supply of
0:17:11 money in the world, but to your point, the supply of money in the world is as big as
0:17:15 the central bank wants it to be.
0:17:18 It’s not a meaningful constraint on economic growth.
0:17:21 The ultimate answer to the question is, if you are just sitting on money, it’s not really
0:17:25 going to have an effect on the broader economy.
0:17:28 I think we’ve come to agreement.
0:17:29 Hopefully, that makes sense to Adam.
0:17:32 Next question.
0:17:33 Our next question comes from Dr. Yvonne Couch, who is Associate Professor of Neuroimmunology,
0:17:41 Alzheimer’s Research UK Fellow, Associate Research Fellow at St. Hildes College, St.
0:17:47 Penderi Lecturer at Somerville College, University of Oxford.
0:17:51 Yeah, University of Oxford, I know it well.
0:17:53 One of my sisters went to St. Hildes College and one of my sisters went to Somerville College,
0:17:57 so this feels very much in the family.
0:17:59 What does Dr. Couch got to say?
0:18:02 She writes, “My question was vaguely economics-based, although probably not very.
0:18:08 Is the way we currently fund science feasible going forward, do we just have too many scientists
0:18:15 and not enough resources?”
0:18:17 All the best, Yvonne.
0:18:19 Oh, love it.
0:18:21 Let me throw out a few thoughts then.
0:18:22 I think that probably there is a problem with the way we fund science, but there’s no one
0:18:28 way that we fund science.
0:18:29 You have university research, you have various grants, various sources of funding, philanthropy
0:18:33 and so on.
0:18:35 You also have private sector research, which is often incentivized by the patent system,
0:18:42 and then you have big block grants that are handed out by agencies such as the National
0:18:46 Institutes for Health.
0:18:49 So there are lots of different ways that science gets funded.
0:18:53 A couple of things that worry me is that, first of all, there’s a big incentive to make really
0:19:00 incremental improvements rather than to take risks.
0:19:05 There’s a great economics paper studying scientists who are funded by the National Institutes
0:19:11 for Health, which is a wonderful institution, very important.
0:19:14 And scientists who on paper seem to be on the same career track, they’ve got very similar
0:19:20 publication records.
0:19:22 And they were instead funded by a private foundation called the Howard Hughes Medical
0:19:27 Institute.
0:19:28 Listeners who want to hear more about Howard Hughes and the Howard Hughes Biography can
0:19:32 go to the back catalog of cautionary tales.
0:19:33 So the Howard Hughes Medical Institute basically takes risks.
0:19:37 It wants people to do something new.
0:19:39 It’s happy with a high risk of failure as long as there’s some chance of a real breakthrough
0:19:45 success.
0:19:46 And this particular paper studying the results that come from these two funding systems basically
0:19:52 finds that the grant funders get what they pay for.
0:19:56 So the National Institutes for Health get a high success rate, but it’s often quite
0:20:00 incremental progress.
0:20:02 And the Howard Hughes Medical Institute has lots and lots of failures, but when it succeeds,
0:20:06 it really succeeds and these are hugely important papers.
0:20:09 And I just feel that we’re not deliberate enough about saying, well, how much of our
0:20:16 funding ecosystem should be aiming a kind of venture capital style for really big wins
0:20:22 and how much should be incremental?
0:20:24 I think those are not the sorts of questions that get asked.
0:20:27 So that’s one of the things that worries me, but what’s your take, Jacob?
0:20:31 One of the things that I have read is that over time, the average age of grant recipients
0:20:36 from the NIH has gone up and up and up.
0:20:39 So kind of just think of this whole universe as getting older and more risk averse and
0:20:44 more kind of bureaucratic.
0:20:46 There is an interesting set of counter pressures, I think, rising up, partly out of Silicon
0:20:52 Valley.
0:20:53 I don’t know if you’ve come across the work of Patrick Collison and his brother.
0:20:58 They’re from Ireland, they found its stripe, their very rich stripe is a big company that
0:21:02 does basically payment stuff online.
0:21:05 They have a really interesting set of kind of philanthropic endeavors around the idea
0:21:11 of progress.
0:21:12 They’re trying to create a kind of field of progress studies that is, it’s very meta.
0:21:18 What are the conditions that best foster technological and scientific progress?
0:21:24 Yeah, but this question from Dr. Couch is very meta.
0:21:27 Yes.
0:21:28 The progress studies, isn’t it?
0:21:29 Yes.
0:21:30 There is this institute in the Bay Area called the ARC Institute that hires leading scientists.
0:21:36 The basic idea is give talented people freedom and money, and encourage them to take big
0:21:44 swings and they are interested, not just in outcomes, but in new tools, again, continuing
0:21:50 on the nerdy thing.
0:21:52 If you take a tool like CRISPR, CRISPR is an intermediate tool that allows people to
0:21:56 cut up a genome, basically.
0:21:57 We just had the first transplant from an animal into a human a few weeks ago because of CRISPR.
0:22:02 We have the first treatments for sickle cell disease because of CRISPR.
0:22:07 I do think there is a wave of people trying to rethink scientific funding.
0:22:15 There is a bottom line aspect to this question from Dr. Couch that I don’t feel like I know
0:22:20 enough to answer.
0:22:21 I’m curious if you do.
0:22:22 I mean, the question is the way we currently fund science feasible going forward.
0:22:26 There is a yes/no version of the answer.
0:22:29 Do you think you know?
0:22:30 No.
0:22:31 Is it?
0:22:32 I don’t know the answer, but I guess I don’t know either.
0:22:35 Another thing that concerns me on this, which is potentially really existential, is the
0:22:41 question of how we fund new antibiotics.
0:22:45 I think this points to a real weakness in the ecosystem of research funding.
0:22:50 If you think about the basic way we develop drugs, the fundamental incentive is the patent.
0:22:56 A drug company spends a lot of money, tries to develop drugs.
0:22:59 Some of them work, some of them fail, a lot of them fail, but in the end you have a drug
0:23:04 and the deal is we give you this artificial monopoly called a patent and it will only
0:23:09 last for a certain amount of time, which is kind of a problem because it takes so long
0:23:12 to develop the drugs.
0:23:13 Maybe the patent’s nearly expired.
0:23:16 You can charge an incredible amount for these drugs for a while and then your patents will
0:23:20 run out and then loads and loads of people will make the same drug and it’ll come down
0:23:24 in value, for example Viagra.
0:23:26 You could sell this for a huge amount of money and now Viagra is off-patent and anyone can
0:23:30 make a generic Viagra.
0:23:32 That’s the incentive that we’ve given to private companies, that you will have this
0:23:36 temporary monopoly.
0:23:38 Now think about antibiotics.
0:23:40 The problem with antibiotics, we have lots of antibiotics that work really well except
0:23:44 the bacteria have figured them out.
0:23:46 Why have they figured them out?
0:23:48 Because we’ve used them, the bacteria develop resistance.
0:23:51 What we really need is new antibiotics that we don’t use.
0:23:57 And now think about how the patent system deals with that.
0:24:00 So you’re basically saying, if you develop a new antibiotic, we’ll give you a temporary
0:24:05 monopoly, but we really need you to just not sell any, don’t sell any of this drug.
0:24:10 Except in cases of dire emergency, it’s like a break glass in case of emergency.
0:24:16 I feel like a bounty, like every econ nerd storyteller loves a good bounty, right?
0:24:21 Think this done with malaria vaccine, which is actually coming along quite well.
0:24:25 You have some universe of people say, we will pay a billion dollars to anyone who comes
0:24:30 up with a new antibiotic that meets this set of criteria, that treats this set of bugs
0:24:36 that are resistant to these existing antibiotics.
0:24:39 And we’ll give you the money and you give it to us and we’ll put it on the break glass
0:24:42 in case of emergency shelf.
0:24:44 You’ve used the word bounty, but the term that is normally used is an advanced market
0:24:48 commitment.
0:24:49 These were proposed most famously by Michael Kramer, who’s a Nobel Prize winner in economics.
0:24:54 And basically the way this prize, this bounty tends to get paid is as a kind of extra payment
0:25:00 on top of each dose you sell.
0:25:02 So we’ll give you extra for every kid that gets vaccinated.
0:25:06 So we’re back to the same problem in that universe.
0:25:08 Back to the same problem.
0:25:09 But the reason we go, well, why does it have to be like that?
0:25:11 It doesn’t have to be like that.
0:25:12 But the reason that they tend to be structured like that is because you need to demonstrate
0:25:16 some kind of market demand.
0:25:18 You need to be willing to buy your product.
0:25:20 If they are, we’ll give you an extra payment for every product you sell.
0:25:25 But that wouldn’t work for antibiotics.
0:25:26 There are so many different ways in which science funding could be said to be broken.
0:25:30 But I think Dr. Couch is right to raise the issue.
0:25:34 We need to do a lot more of this kind of meta thinking about progress studies.
0:25:39 Tim, that was a lot of answer.
0:25:41 Let’s take a break.
0:25:42 Cushionary tales will be back in a moment.
0:25:52 Hey, everybody.
0:25:53 I’m Kai Rizdal, the host of Marketplace, your daily download on the economy.
0:25:57 Money influences so much of what we do and how we live.
0:26:00 That’s why it’s essential to understand how this economy works.
0:26:04 At Marketplace, we break down everything from inflation and student loans to the future
0:26:08 of AI so that you can understand what it all means for you.
0:26:13 Marketplace is your secret weapon for understanding this economy.
0:26:16 Listen, wherever you get your podcasts.
0:26:22 And we’re back.
0:26:23 I’m Jacob Goldstein here with Tim Harford on Cushionary Tales.
0:26:26 Hello, Jacob.
0:26:27 More questions.
0:26:28 What have you got for me?
0:26:29 All right, Tim, I got another one for you.
0:26:32 Comes from Graham in Florida.
0:26:36 Graham writes.
0:26:37 I’m fascinated by errors made by falsely identifying correlation as causation.
0:26:44 What examples of this error stand out to you?
0:26:47 I love the question, but I’m going to slightly sidestep it because I’m worried by this question,
0:26:51 but I think that it’s generally more complicated than simply some fool saw a correlation and
0:27:00 thought it was causation.
0:27:01 You don’t want to talk about sunspots and crop yields?
0:27:04 Well, let’s talk about stalks and babies briefly.
0:27:07 Okay.
0:27:08 A classic, a classic of the genre.
0:27:10 Yeah, the most successful book ever published about statistics, How to Lie with Statistics
0:27:15 by Darrell Hough, includes this example of stalks and babies and shows that there’s a
0:27:20 correlation between the number of stalks and the number of babies.
0:27:23 And there are various ways to demonstrate this correlation.
0:27:26 One way to do it is you just look at national populations.
0:27:29 You go, “Hey, countries with lots of stalks also have lots of babies.”
0:27:32 And there’s a very, very strong correlation.
0:27:34 But of course, the reason is big places like the United States have a lot of room for
0:27:38 stalks and a lot of room for babies.
0:27:40 And small places like Luxembourg or the Vatican City don’t have a lot of room for babies and
0:27:45 don’t have a lot of room for stalks.
0:27:46 So then you go, “Oh, that’s a great example of this mistake.”
0:27:49 The sting in the tale of that story, as I describe in my book, The Data Detective, is
0:27:55 that Darrell Hough, the guy who created that story, then went on to tell the same story
0:28:01 in congressional testimony, saying that there was no compelling evidence that smoking cigarettes
0:28:08 would give you lung cancer.
0:28:10 And it was just like stalks and babies.
0:28:12 And he was actually hired by the tobacco lobby.
0:28:14 It was just correlation.
0:28:15 It was just correlation.
0:28:16 They seemed about the same.
0:28:18 They were like, “Sure, yeah, there are people who smoke and there are people who get cancer,
0:28:21 but no, there’s no causal evidence.”
0:28:23 One theory was cigarettes are so soothing.
0:28:25 And if you have some early symptoms of lung cancer, maybe you soothe that with the soothing
0:28:30 vapors of cigarettes.
0:28:31 I mean, it’s completely ridiculous, but this sort of rhetoric was deployed.
0:28:35 And so one of the things I’m very concerned about in my work on statistics is that it’s
0:28:40 great to be skeptical about statistics and to point out lots of examples of statistics
0:28:44 being misused.
0:28:45 But if that’s all you do, you just get to a kind of nihilistic situation where you’re
0:28:50 just constantly rejecting statistical evidence because, oh, it’s just another of those downlives
0:28:55 and statistics.
0:28:56 But when you look at the real world, I think this gets to be incredibly interesting.
0:29:01 So a real hot topic at the moment is our smartphone’s destroying a generation, basically.
0:29:08 Are our kids having their mental health wrecked?
0:29:10 There’s a new book that basically makes that a new book by a prominent academic.
0:29:14 Yeah, by Jonathan Hyde.
0:29:16 And there have been others by Gene Twenge.
0:29:17 And lots of people have said this.
0:29:19 And the evidence for it is mostly correlational.
0:29:24 Not completely.
0:29:25 There are some experiments, but they’re not that convincing.
0:29:27 None of them are perfect, but there are lots of different ways of measuring it.
0:29:31 And the really interesting evidence basically says, look, there appears to be a mental health
0:29:37 crisis.
0:29:38 The kids seem to get really distressed, particularly the girls, when children have access to social
0:29:44 media on their phones.
0:29:46 And that happens around sometime between 2010 and 2014.
0:29:49 And at the same time, suddenly you’ve got this outbreak of suicidal ideation, self-harm,
0:29:56 poor mental health, and so on.
0:29:57 And that’s correlational evidence.
0:29:59 I don’t completely believe it, but I don’t completely not believe it either.
0:30:02 I think that’s what makes it interesting and makes it important to engage with.
0:30:06 You have to start going, well, what’s the alternative explanation?
0:30:09 Is there something else that happened sometime about 10, 12, 14 years ago that might explain
0:30:15 this mental health distress?
0:30:16 So the timing of the great financial crisis is probably not quite right.
0:30:19 Donald Trump, maybe Donald Trump is upsetting the kids.
0:30:22 Timing doesn’t work.
0:30:24 So partly it does the pattern of the correlation make enough sense to explain this causal story,
0:30:30 which, funnily enough, is basically exactly what the scientists’ finding connection between
0:30:34 cigarettes and lung cancer were doing.
0:30:36 They only had correlational evidence for a long time.
0:30:38 They weren’t running randomized trials saying, you know, half of you smoke and half of you
0:30:42 don’t smoke.
0:30:43 I mean, they couldn’t do that.
0:30:44 They had to look at correlational evidence, and sometimes that’s what we’ve got.
0:30:48 This question got me thinking about the rise in social science of what they call natural
0:30:54 experiments, right, trying to find instances in the real world where you have something
0:31:00 that obviously is not as good as a randomized trial, because you just can’t get that with
0:31:04 large numbers of people in the world, but that gives you some element of randomization
0:31:10 or pseudo-randomization.
0:31:12 Something that allows you to try and make the leap from correlation to causation.
0:31:16 And you know, as far as I know, it goes back to the Vietnam War, right, where in the U.S.
0:31:21 there was a draft lottery, and there were social scientists after the war who looked
0:31:27 and said, oh, look, here are people who are at an aggregate level, very similar on many
0:31:32 dimensions, socioeconomic dimensions.
0:31:34 We can look at people who randomly got drafted versus those who randomly didn’t and who
0:31:40 appear quite similar in the aggregate and see how their outcomes differ.
0:31:44 And that’s a really good example of a natural experiment, because it’s actually very close
0:31:48 to a real experiment.
0:31:49 Yeah, when you get a lottery of the real world, now social scientists flock to it, right?
0:31:54 Similarly, there was one in Oregon some years ago with Medicaid, which is the healthcare
0:32:00 program for low-income people in the U.S.
0:32:03 And Oregon got some new Medicaid funds, and they randomly allocated them to a group of
0:32:08 people over time, right, so that social scientists could say, oh, look, here are people who are
0:32:13 basically identical, some of them got this health insurance and some of them didn’t.
0:32:16 And in that case, the findings were quite interesting.
0:32:19 Medicaid didn’t appear to be as helpful as I would have thought, as the researchers
0:32:24 themselves would have thought, according to the one I interviewed.
0:32:27 Lowered mental health problems, people worried less about money, but like their basic health
0:32:31 outcomes didn’t improve, which nobody would have guessed, right?
0:32:35 And the evidence is quite robust.
0:32:38 When you don’t have those lotteries, as you say, the world is just hard to understand.
0:32:42 I mean, even in some instances where you do have randomized trials, people are complicated,
0:32:47 the body is complicated, society is complicated.
0:32:50 And so, you know, correlation in a certain way, I think, is underrated, like, yes, obviously
0:32:55 it doesn’t equal causation, but it’s a place to start looking, right?
0:32:59 It’s a place to start asking questions.
0:33:00 I think that’s absolutely right.
0:33:02 And we need more randomized experiments.
0:33:04 And there are more opportunities to run them than people seem to think.
0:33:07 For example, one of the things that John Hyde in his book is arguing for is, you know,
0:33:11 shouldn’t have smartphones in schools.
0:33:13 Schools should be phone-free zones.
0:33:15 I could completely imagine a state saying, we’re going to introduce a rule whereby in
0:33:20 all of the schools in the state, no smartphones, strictly forbidden.
0:33:25 You have to put them in a locker when you show up and then unlock them at the end of
0:33:28 the day.
0:33:29 You could introduce that rule and just go, well, we’re going to introduce it for a semester
0:33:33 at random in 50% of the schools, and then we’ll introduce it in the other 50% of the
0:33:38 schools in the next semester.
0:33:40 And we’ll just randomize that because we want to know whether there’s any point in
0:33:43 this experiment or not.
0:33:45 That’s not very difficult to do.
0:33:47 It would create so much information about children’s performance in the classroom, their mental
0:33:52 health, that could inform policy across the world.
0:33:56 That would be the dream.
0:33:57 You know, there is now a robust set of methods, essentially, where social scientists could
0:34:03 look at that state and compare it to neighboring states and see the difference in differences,
0:34:08 as they say, see the change over time.
0:34:11 And that’s an instance where it wouldn’t be as elegant as randomizing within a state.
0:34:15 But I feel like you could start to get pretty good data, even if you did one state compared
0:34:20 to other states.
0:34:21 Yeah.
0:34:22 If you’ve got no experiments, we should run them.
0:34:24 If we don’t run them, there are still ways of making correlation talk.
0:34:28 Should we have another question?
0:34:31 OK, Tim.
0:34:33 We got one more.
0:34:34 We had to have one game-related question for you.
0:34:37 Oh, great.
0:34:38 Yes, you’re welcome.
0:34:39 This one comes from Joost, who writes, “Hi, Tim.
0:34:43 Diving into the board game-shaped flank you’ve now exposed for discussion,” that is a very
0:34:48 gamerish way in, “what board games or board game mechanisms that you enjoy are particularly
0:34:54 insightful on some aspect of real-world economics?
0:34:58 For a unique angle to look at a maybe-nitch problem?
0:35:02 And what non-D&D games are you particularly enjoying right now?
0:35:08 Love the show however many episodes you produce.”
0:35:11 That is a listener we all want.
0:35:13 Thank you, Joost.
0:35:14 Yeah, so kind, Joost.
0:35:15 There are a couple of mechanisms that I do see used quite often that shed important
0:35:20 light on economics.
0:35:21 So one is trading.
0:35:22 A lot of games involve trading.
0:35:24 Now, Monopoly, that classic of sadly not a very good game, there’s a cautionary tale
0:35:30 about the history of Monopoly if people want to listen, Monopoly in theory involves trading,
0:35:35 but in practice not a lot of trading happens.
0:35:37 It turns out that you need to give people a reason to trade.
0:35:43 The great modern game Settlers, now about 30 years old.
0:35:46 By the way, Settlers, I assume that Settlers of Catan as an outsider, not on a first-name
0:35:50 basis with the game.
0:35:52 Settlers of Catan is the game.
0:35:54 If the game that Monopoly wishes it was, it’s like if Monopoly had been designed with a
0:35:59 modern eye to be super exciting, people need different resources and the supply of resources
0:36:06 fluctuates.
0:36:07 So sometimes you’ve got loads of wood, sometimes there’s no wood, so there’s an active incentive
0:36:11 to trade all the time.
0:36:12 Oh, and by the way, if you don’t trade every now and then, the robber baron comes and
0:36:14 takes stuff.
0:36:15 So there’s an interesting insight that trading doesn’t just happen because it’s allowed.
0:36:20 It needs to be some difference in value and perhaps some incentive to get on with it.
0:36:24 So my personal favourite game is Agricola, a wonderful game about developing a farm,
0:36:30 which doesn’t sound promising, but it’s really, really good.
0:36:33 But one of the clever things about Agricola is it uses an auction in an interesting way.
0:36:38 It’s sometimes called a descending clock auction or a Dutch auction.
0:36:42 The prize gets more and more tempting.
0:36:46 So in Agricola, there’s just more and more good stuff on the board.
0:36:49 In a traditional descending clock auction, basically the price is going down and down
0:36:53 and down and down.
0:36:54 And everyone is just sitting there, and then it’s a question of who grabs it first.
0:36:57 The longer you leave it, the better it gets, but then only one person can have it.
0:37:02 It’s very, very elegant.
0:37:03 They sell flowers at Aalsmere in the Netherlands, and it’s just incredibly quick, much, much
0:37:09 quicker than selling, say art, you know, you’re selling a van Gogh.
0:37:12 Huh.
0:37:13 And the price rises and rises and rises.
0:37:14 So auction design is the whole thing, right?
0:37:16 When would you choose a traditional prices going up eBay art style auction in the real
0:37:22 world?
0:37:23 And when would you choose a Dutch prices going down auction?
0:37:25 What do you get in a relative sense out of each one of those?
0:37:28 There’s a couple of differences.
0:37:30 One is that in certain types of ascending auction, you get to see people dropping out
0:37:36 as the price rises.
0:37:37 It depends on the way the auction is designed, but you could imagine an ascending auction
0:37:41 where you just see people going, no, I’m out.
0:37:43 I’m out.
0:37:44 I’m out.
0:37:45 That’s sort of the classic kind of art auction in a movie.
0:37:47 Yeah.
0:37:48 And so you’re learning information as the price rises.
0:37:50 How many people are still interested?
0:37:51 Are there still 10 people interested?
0:37:52 Is it just two people?
0:37:53 Right.
0:37:54 So that generates information.
0:37:55 And by generating information, it might raise the final price overall.
0:38:00 The advantage of the descending auction is it’s just so quick.
0:38:04 You can run an auction in 10 seconds.
0:38:07 It’s a 100,000 tulips.
0:38:09 Here are the tulips.
0:38:10 We’re going to sell them for whatever, 10,000 euros, 9,900, 9,800, 9,000, and it’s literally
0:38:17 a clock.
0:38:18 The clock just runs around showing what the price is.
0:38:21 And then someone presses the button, sold, okay, now bring in the roses.
0:38:25 So it’s very fast.
0:38:26 You would want to have a good idea of the market clearing price if you were going to run that
0:38:30 kind of auction so that you didn’t sell it for too cheap, right?
0:38:33 It’s fast, but you run the risk of selling it for too cheap.
0:38:36 But if it’s just commodity tulips, then you basically know the market clearing price anyway.
0:38:39 Yeah.
0:38:40 It’s almost like a stock market for tulips.
0:38:42 That I think is why it works so well, whereas if it’s a unique work of art, you need to
0:38:46 give maximum information, maximum comfort to people, maybe a little bit of theater as
0:38:50 well.
0:38:51 Yeah.
0:38:52 We’re not in a hurry to run this auction.
0:38:53 The cost of the auction itself is trivial compared to the value of what’s being sold.
0:38:58 Now we go over to Agricola and have a descending auction.
0:39:01 It feels exciting and it also feels like you’re trying to guess what is in your opponent’s
0:39:06 head because they’re not revealing it as they would in a descending price auction.
0:39:10 You have to think, “Oh, what are they willing to pay for it?”
0:39:13 Absolutely.
0:39:14 And it works slightly differently, but functionally what’s going on is that every turn you take,
0:39:19 you get to grab some resource on the board.
0:39:21 And then once you’ve grabbed the resource on the board, no one else is allowed to grab
0:39:24 it until the next round.
0:39:26 And there are several moves you can make in each turn, so you might be able to get two
0:39:30 or three things each turn.
0:39:32 And so, again, you go backwards and forwards, it’s your move and it’s my move, and each
0:39:35 time we want to grab something.
0:39:37 And you might go, “Well, do I want to grab it now or do I want to wait until next turn
0:39:40 where it might be better?”
0:39:41 And you’re trying to figure out, like, what is the thing that the other person is desperate
0:39:45 to have?
0:39:47 Can I just let this pile get bigger because they can’t afford to take it because they’ve
0:39:51 got to use their move on something else?
0:39:53 It’s a very, very clever game.
0:39:55 I do enjoy it.
0:39:56 You’re also asked what non-D&D games I am enjoying.
0:40:01 I am enjoying a modified version of Blades in the Dark, which is a classic role-playing
0:40:06 game, very fast-moving, you get to do Ocean’s Eleven or other kind of heists.
0:40:11 And it’s very modifiable.
0:40:14 We’re speaking on Monday, just yesterday on Sunday, I had a whole bunch of old friends
0:40:18 around to my house, and we just played this game all day, and we had an absolutely terrific
0:40:24 time.
0:40:25 I like your live-in-the-dream, Harford.
0:40:26 Makes me happy.
0:40:27 Living my best life.
0:40:28 Jacob, this has been such fun.
0:40:31 Thank you so much for doing this.
0:40:32 Ah, it was a delight.
0:40:33 I’ll come back any time.
0:40:35 We would love to have you back.
0:40:37 Thank you, everybody who sent in a question.
0:40:39 Sorry we weren’t able to answer all the questions, but we are going to be back with another cautionary
0:40:43 questions episode later this year.
0:40:46 Please do keep the queries coming.
0:40:47 Send them into tales@pushkin.fm, that’s T-A-L-E-S @pushkin.fm, and I will be back with another
0:40:56 cautionary tale in two weeks’ time.
0:40:58 Jacob Goldstein has a wonderful podcast called What’s Your Problem?
0:41:02 Thank you, Jacob.
0:41:05 Thanks, Jim.
0:41:12 Cautionary Tales is written by me, Tim Harford, with Andrew Wright.
0:41:16 It’s produced by Alice Fiennes with support from Marilyn Rust.
0:41:20 The sound design and original music is the work of Pascal Wise.
0:41:24 Sarah Nix edited the scripts.
0:41:27 It features the voice talents of Ben Crow, Melanie Gushridge, Stella Harford, Gemma Saunders
0:41:32 and Rufus Wright.
0:41:34 The show also wouldn’t have been possible without the work of Jacob Weisberg, Ryan Dilly,
0:41:39 Greta Kohn, Litao Moulard, John Schnars, Erics Handler, Carrie Brody and Christina Sullivan.
0:41:47 Cautionary Tales is a production of Pushkin Industries.
0:41:50 It’s recorded at Wardall Studios in London by Tom Berry.
0:41:54 If you like the show, please remember to share, rate and review.
0:42:00 Tell your friends.
0:42:01 And if you want to hear the show ad-free, sign up for Pushkin+ on the show page in Apple
0:42:06 Podcasts or at pushkin.fm/plus.
0:42:17 Hey everybody, I’m Kai Rizdal, the host of Marketplace, your daily download on the economy.
0:42:22 Money influences so much of what we do and how we live.
0:42:25 That’s why it’s essential to understand how this economy works.
0:42:29 At Marketplace, we break down everything from inflation and student loans to the future
0:42:33 of AI so that you can understand what it all means for you.
0:42:38 Marketplace is your secret weapon for understanding this economy.
0:42:41 Listen wherever you get your podcasts.
0:42:43 (upbeat music)
Tim Harford is joined by Jacob Goldstein to answer your questions. Does winning the lottery make you unhappy? Is Bitcoin bad for the economy? When does correlation imply causation? And what will Tim and Jacob do when the robot overlords come for their jobs? Enjoy this episode from Cautionary Tales, another Pushkin podcast.
See omnystudio.com/listener for privacy information.