Nate Silver explains how Harris could beat Trump

we have Harris just ahead in the polls right now because after the convention the party is usually on a little bit of a sugar high we have that more like 50/50 for election day so yeah Harris would love to the if we could have a snap election in the US Harris would call one right now right she has momentum but things could change hello and welcome to ways to change the world I'm Christian Guru Murphy this is the podcast in which we talk to extraordinary people about the big ideas in their lives and the events that have helped shape them my guest this week is Nate Silva now Nate became famous I guess as the guy whose political predictions and polling analysis we all wanted to read in the US elections around 2008 and then 2012 he was the founder of 538 um which was a brand that went through various different ownerships and versions and he has now actually left uh and Nate now uh does a substack which is a Blog in old money um called the silver bulletin in which he still does uh political analysis and uh and and sort of probability predictions of what's going to happen in the November elections but he's here because he has written a book called on the edge the art of risking everything which is really a sort of a return to who you were before you were famous politics which was which was a gambler and looking at people who take risks professionally I suppose because they are they are the Modern Masters of the Universe and they are the people who have a huge influence over modern politics for sure if you look at the skill set that um is present in Wall Street or Silicon Valley or Las Vegas um a it's becoming the dominant skill set financially where those are the SE of the economy finance and Tech that grow as a part of the economy every year um and B in a datadriven world that mindset is very important um it's not necessarily the abstract academic of this it's it's actually having skin in the game and taking risk the physical thing that does to your body the actual um you know reputational risk that you're putting on the line and so wanting to like study those people who are kind of my people so to speak was the impetus for this book so so those people might well be professional gamblers on one part or Venture capitalists but might also be those kinds of tech Giants Elon Musk Peter teal who is the guy who's pushed JD Vance more than anybody else uh so so so this is inherently political as well for sure I mean you know in the book I talk about the Divide between what I call the river which is this community we've been describing of gambling types um and the village which is the establishment you know in the US it's Harvard in the New York Times The Establishment the sort of the Poli EST people call it the cathedral for example I call it the Village um and that's a more risk averse place also when it's more the village is more about the community and the good of the whole um and politics I suppose whereas the river is more about individualism fairly unabashed capitalism in a lot of cases but that's kind of the skill sets winning out for For Better or For Worse and in America at least and to some extent here I suppose the village is broadly left liberal for sure um you know uh sort of uh think thinks with not one voice but but but you know it's sort of more of a homogeneous set of values and the river can be anything they can be crazy out there on the right there there definitely like a right Fringe of the river I mean look the average college educated person in the United States um is a Democrat probably so probably you still have more Harris voters and Trump voters but yeah people like Peter teal and Elon Musk are are part of the river and are very much on the right Fringe what why did you want to delve into this world and try to understand them it's partly a journey of self-discovery you know I like you said I got into poker and forecasting baseball statistics and building models before I ever got into politics um and I've always felt aranged in the village um you know I was taught to be be critical and be skeptical be willing to make bets on things don't worry too much about whether you offend person what matters is that is that you have the right answer you're contributing knowledge to the world um and I've seen a lot of group think by contrast in the village um you know they are have become I mean in the US everything is now divided did you get a college degree or not the more education you have the more likely you are to be a Democrat um the problem is that I think has led to a lot of partisanship where our institutions are not performing as well as they once did and so I wanted to have these two groups understand one another better I want to reign in some of the excesses of the river but have the village be more entrepreneurial in certain ways and and less risk averse and is the risk that you focus on essentially Financial for the most part there is one chapter where I talk to people who take physical risks um for example an astronaut the woman who invented um the co mRNA vaccines she actually had to smuggle money in a teddy bear out of hungry out of communist hungry didn't come over and was a real physical risk taker um an Explorer an NFL player um it's actually more similar than you would think I mean their ability to be calm and collected under pressure is very rare and that does carry over to Poker you know poker if you're playing a hand for thousands of dollars um you physically sense that your body registers that I'm in a moment right now of of risk or Peril um some people rise to thecas they love being in the zone and love feeling the pressure and some people choke what what then do you think are the the key lessons that the village if you like can learn from the river because you know the bottom line is when you're running a public institution and you hear this conversation over and over again you know people will say well we're not a business we can't just run the odds and say well it doesn't matter if we fail because we're talking about people's lives and and EST State and public services so in in the way that an entrepreneur can just say you know I'm just going to keep I'm going to keep throwing money at these different ideas years and if one of them wins Big then it doesn't matter if three or four of them uh are money down the drain you can't do that in what you call the Village can you you should I mean look so the woman who invented this vaccine Caln Caro who won the Nobel Prize for it um she was repeatedly styed in Academia because it wasn't a safe enough idea yeah if you can develop NRA vaccines could save millions of lives but you want to get your next Grant you want to like keep your job as a mid-level bureaucrat working for some University somewhere and so she had to go to the private sect sector to develop that idea and in the past 20 years in 2006 the US and and the UK had the same GDP per capita now you're at 47,000 per capita we're at 77,000 now there are downsides to the us we die earlier that's a big problem especially man in the US our life expectancy has stagnated um but you know but running more efficient governments and and trusting the private sector and looking at expected value and incentives and things like that I mean I think we should actually have you know pay people more to have good government jobs and compete more with the private sector because the collective is important so just explain expected value what do you think it role should be in decision making so this is what you get if you simulate the world a thousand times so in poker it's very straightforward where there's 52 cards imagine how the situation works out every one of those 52 cards is dealt um in real life it's more complicated for everyday things like oh should I take a taxi or take the tube to the interview it's kind of an expected value calculation you might look at the amount of traffic and whether there delays on the Northern line or whatever else um when it comes to really big decisions like you know my life choices or my life partner or my career choices then then it's more of an abstraction because you only get to simulate the world we only live once right um depending on where you want to go theologically I suppose at least um but still being able to take risks when the rewards outweigh the risks is very valuable being willing to leave a job for one that is higher upside in terms of your professional fulfillment in terms of your finances whatever else but might be more risky you know especially if you're young that's probably a good idea overall but how do you how do you calculate the probability of things that are infinitely complex you know and ascribe a number to them in terms of well if I'm trying to work out whether I should literally leave this job and go to that job how do I work out what the number is it's you well you have to here's you can't avoid making a decision at the end of the day a probability is a number between zero and 100 um so you know one of the things that poker players get good at is being skilled at estimation they know the big important things and they know the things that are maybe less important and matter around the margin um one problem I think the The Village has is that in bureaucracy you might worry about about the trivial marginal things right um in the US for example our vaccine distribution scheme had many many different categories and subcategories based on the political influence of the different groups in the UK and places they were like you know what older people are dying more readily because of Co and are at more risk let's just go in adverse order of age right like that was actually a good assumption that the UK made that the US didn't in that case because I mean I wonder whether what you're actually describing is is the real power struggle that's going on today between politics and big business where you know we we can legitimately ask ourselves who has more power over my life right now is it is it the likes of Elon Musk and Peter teal um or is it my government and my government might well be very cautious very risk averse very afraid of offending certain groups of Voters who then won't elect them you know reelect them in in four years time where whereas these Giants of business who are running the things that actually are the things that govern my life every day from my phone to my car they just do things they don't care look I'm pretty much a capitalist for the most part um but the degree to which the 10 richest people in the world get more and more and more wealthy it doubles every 10 years I mean that gets to the point now where Elon Musk has more power than probably half the countries in the United Nations or maybe half of them combined on some level um and that does seem problematic potentially I mean I guess you see um as an American I'm very Pro free speech you see now the village fighting back with you know Brazil is Banning Twitter and you see Telegraph president arrested you see things like that I object to that by the way you know I'm happy that we in the US have a more robust culture of free speech um but you see that there really is a clash not between governments but between governments and the private sector if you want to call them ol I mean they mostly are self-made right but this oligarchical culture that I think is the increasing access of conflict so so where where do you think that goes um I think it probably gets worse before it gets better I think we need smart regulation I hope this book is a way to um understand help people in the village understand that by the way on balance technological growth has been very very good for Human Society um until the enlightenment which started here um in roughly the late um 18th century um there was no such thing as progress really people's lifespans were the same there was no economic growth there was no real expansion of Rights and the franchise to of voting for example um and then something happened that sparked growth and ever since we've seen you know poverty go radically down in countries like China and India for example and lifespans until recently in the US so that is a worry expand um so to get people in the village to understand that like yes growth is good basically and improves The Human Condition and Technology ol is good but to re in the excesses of the river because there are excesses already I mean you see Sam McMan freed in this book for example um basically stealing $10 crooked Banker crooked Banker yeah stealing 10 billion doar right and you see the issue of moral hazard and the financial crisis where people take risks that they're not responsible for you know capitalism is very efficient at responding to its incentives so if you give people an incentive to take risk that we will have to pay the price for then the system will find a way to incentivize that well I mean this brings me to sort of whether you think this is is there anything new in this you know or has it always happened you know why why does it feel different now why does it feel to be on a different scale I think it's partly um the fact that we now have data about everything you know when you kind of click the button here in the UK you have more barriers when you click the button and say okay yeah I'll go ahead and I'll share my data with you I mean that's being used to exploit you potentially right um you know when all of a sudden you have a suggestion to follow somebody on social media who your phone overheard you having a conversation with in real life I mean things get like get really eerie and so um so the kind of Moneyball notion that like data will help the underdog I think is the reverse of that is true right data helps the the hegemon data helps the people who have a lot of computing power in AI in particular you need huge amounts of computing power and so basically you know the four largest US companies plus the government of China will become like even more disproportionately powerful so we do have to think about this stuff so what did you learn about the kinds of people who succeed at this I mean there there's a little your book has little sort of sub essays in it yeah and one of them is you know is everyone in the river on the Spectrum yeah um so I I think you'll definitely meet an above average number of people who would self- diagnose or I might diagnose as having Aspergers or other other related conditions to autism um I think it's you know I prefer the term neurode Divergent because they fit some stereotypes but not others so um people with Aspergers or Autism typically are less concerned about social conventions or less well able to read social conventions and that can be an advantage if you're in a Founder a founder of a Silicon Valley company trying to do something different or if you're trying to start a new um a political movement right and you're maybe not so concerned about RMI offending people um however typically people who are autistic want routine and are not very good at dealing with like novelty and risk so I think what happens is you get people who have this unusual combination like Elon Musk who has described himself as having Aspergers where on the one hand um he doesn't care about offending people to a point where it's self-destructive at times that probably is helpful to him on balance um on the other hand he obviously has a crazy insane tolerance for for gambling and for risk and it is L gambling in his case right but I mean that there there's one thing about sort of um you know the ability to offend people or not care about offending people the the the next level to that question really is about morality isn't it and good you know right and wrong you know and does he know the difference between right and wrong wrong or does he have such an extreme view of utility and utilitarianism that he's prepared to let people die or EXT you know experience extreme poverty or you know along the way for whatever he thinks the great to good is I mean look I there's a long critique of utilitarianism in the book and how I think it's kind of an incomplete and potentially even very dangerous moral theory especially the kind of more short form versions of it um but look I do think directionally speaking it's correct to be skeptical of the conventional wisdom um if you go and look at kind of the halflife of a truth right like what people believe 50 years ago about like homosexuality or things like that and how much that has changed for instance um or how much science evolves during the pandemic in the US we were told at first oh you don't want to wear masks they're not very effective they're for healthcare workers and then that flip-flops a couple of weeks later um yeah look I think I think the village needs to have its conventional wisdom questioned more I mean you know as someone who identify as as a liberal not necessarily on the left but a liberal in the way that term is understood in Europe for example a Libertarian or I wouldn't a Libertarian who supports a generous welfare state and believes a welfare state helps people to take more risk and build Society up further um but like you know we are the ones who used to be skeptical of the establishment and I don't like the idea that now conservatives are the only ones who are allowed to get away with being skeptical right so so so how do you think this changes or will change our politics and the role of politics because the danger is that it makes politics irrelevant um well look you know it just reduces politics to guys who make regulations look in some ways I wish parties were more strategic about politics I mean politics meets the game theory definition the academic definition of a game right you have people who are competing for different resources and they're trying to strategize and trying to adapt to one another you know Donald Trump saw shortcomings in in what Democrats were offering to people and what the establishment of the GOP like the George Bush era was offering too so he kind of staked out new territory um however you know how can we have competition but also also build trust when you have trust in institutions declining people do become more selfish and self-interested you get in the prisoners dilemma as we talk about it from Game Theory um and so the fact that like in the US now it is every man and woman for themselves because Trust in Media Academia government the church um all going way downhill like that is bad for the future of the US and we have those problems also elsewhere in Europe obviously and and so are you are you saying politics should be more like the river in order to stop the erosion of politics in our I'm saying people if I were a Democrat for example I would want the Democrats to do as much as they can to try to win not be polite but try to try to win right um and that includes adapting to like the media environment you don't want to tell like outright lies but I think people in the village are too afraid of offending one another and offending parts of the Coalition and not concerned enough about about winning give me an example Democrats took an awful long time um to get rid of Joe Biden and they did they did make the smart decision in the end right um but they were like oh that's not how it's done even though clearly Joe Biden was not capable of being an effective president for another four years was losing the polls to Donald Trump who has lost a popular vote twice for example and they did make the right move in the end but it took a lot of nudging to do that um and they nearly didn't and they nearly didn't and they might have made the wrong decision in their VP pick um one reason why our model has election is a toss-up now is because Pennsylvania which is the most important state the way the math works in the US um you know KL Harris bypassed Josh Shapiro very popular governor in Pennsylvania because of some opposition on the left and like you know I think that might prove to be mistake too I I think you you you've said before that you've ended up voting Democrat in most of the elections you you You' voted in I mean so if you were advising Harris you know how can I win the election what would you say I think continue to attack to the center um I think I'd say that like you know don't worry too much about what like annoying people on Twitter are saying right like people in the in the US now are Democrats are convinced that like the liberal media is out to get them which it's not it's the liberal media it's for the most part pretty friendly it hasn't talked much about Biden's age for example since Biden stepped aside um and so you know talk to voters in the middle of the country find ways to motivate younger voters to turn out um go after voters who might be a little bit weird now Democrats are like phobic like people who are weird you know the RFK Jr voters um who he dropped out his voters mostly have gone to Trump and that's given a little bit of M momentum back to Trump in recent days so don't be don't be so proud to chase down anybody's vote right anyone has an open mind figure out some speech you can give or some policy issue concession that can have them turn out for you don't don't cut yourself off at 51% of the electorate and would you be happy to advise Trump as well what would you say to him um I would say kind of Common Sense stuff um you know don't talk about KL Harris's race and things like that um go back in time and replace JD Vance I guess it's kind of too late for that maybe um obviously tactically he's smart I think to try to shift to the center on abortion and Democrats are correct to respond that like it's not very credible necessarily given the justices that he appointed um but look I think he's um I think he got very surprised when the nominee was changed he thought Democrats were a cult of personality like Republicans were and they weren now he's maybe finding his footing a little bit more I mean you talk about your model let's just talk about that for a moment I mean what is it for a start it's a model that takes polling data from elections dating back to 1936 and sets odds based on them right so we all know you know in the UK I know in the US the polls are sometimes wrong um so essentially what it's saying is like given where the polls stand then How likely are they to be right um right now we're in a phase where we're still in a middle phase I mean our elections are very long compared to yours um we just had the Democratic Convention Harris would probably win an election if you held it today um but a it's not today and B um in the US we have a complicated system with the Electoral College so she's ahead in the popular vote but like in Pennsylvania Wisconsin Michigan Georgia the states that probably even your viewers know matter the most in the US it's likely to be very close by election day so not a cop at our model has been very close to 50/50 since since she entered the race and so you wait each polling organization according to how accurate you think they've historically been how accurate they historically been their methodological standards um and how recent the poll is obviously you want polls from yesterday and not a month ago and then you somehow you you aggregate them and you come out with a figure we agree I mean the tricky part is understanding that like different states vote in a way that we call correlated right so um you know Michigan and Wisconsin border one another they're both pretty old and pretty white right but you know socially moderate um so they'll probably tend to move in the same direction it's like you know Scotland has a big shift toward the SNP or something so like so ma ma mathematical modeling gets a little bit more complex than than you might think um in this case though it's kind of obvious that it's a close election we've had close elections in the past in the US by election day maybe the model will be a bit more confident if one C gets ahead by four or five points but that's that's far from a safe lead at the moment you you've got Harris just ahead of trump it's a little conf we have Harris just ahead in the polls right now because after the convention the party is usually on a little bit of a sugar high we have that more like 50/50 for election day so yeah Harris would love to the if we could have a snap election in the US Harris would call one right now right she has momentum but things could change there's a debate for example in a couple of weeks between now and November do does the polling data suggest then that people are actually changing their minds or is it a difference in who's prepared to go out and vote probably some of both I mean enthusiasm had been very low in the Biden Trump matchup um and now among Democrats is close to as as high as it was during the Obama years in 2008 oh right so so the Obama Coalition appears to have been activated already I I think some of it right I mean look Democrats have lost some of their dominance among for example Hispanic American voters um have become more purple again there are kind of complicated reasons for that you know one thing Trump has been effective at For Better or Worse is building a coalition of the working class um they're no longer just the white working class right Hispanic voters without a college education are plurality Trump voters for example you see that more with like Asian-American working-class voters um so Democrats have been kind of like paid a price for being the party of the village most people in the US do not have college degrees um I think people with college degrees sometimes don't know how to communicate to quote unquote average citizens you've seen um obviously a rise in popularism um in many parts of Europe there's been perhaps less of it here than elsewhere I mean one thing about the UK it has a a culture where there's relatively less partisanship than in the US for example um but yeah so you see these shifts and and you know Harris would love to have the Obama 2008 Coalition rebuilt but it's a little bit harder um you know younger Americans too there have been splits over things like the war in Gaza for example um so not quite as easy but but you know she's closer to getting that back than than Biden had been so so when you're taking those different factors whether it's the war or black voters and Hispanic voters and you ascribe them some sort of mathematical value I mean how is that not a guess it's simpler than that which is I'm taking the polls which I rely on the I'm not a pollster I rely on the pulsers to take surveys and of course you're not equally likely to reach every person in a survey you'll get a lot of if you call people on the phone you get a lot of old white women basically and old white men right who are sitting by their landline phone and still taking calls and so there are various um techniques that pollsters use to weight and massage data in different ways and my job is to say how accurate would it actually be in the end and how what are the relationships like between between different states but we should be clear um any model that you see maybe except weather models where they're so tested for many years but like any model involves some degree of assumptions or if you want to call it priors is the fancy term or guesswork is maybe the more accurate term um you can always make different decisions when you build a model that's why it's important to be transparent about what you're doing and don't just let anyone say oh I just I have an oracular model that designed with AI or whatever and like no I mean like you know most models are bad you know most academic papers the findings in those papers can't be replicated again um you know most people who are going into the casino thinking they can beat Vegas with some system wind up losing even more money than they expected so so you should be skeptical of somebody who says um oh my model is Right unless I can explain how it works well I mean everyone was sort of hanging off your every prediction in 2008 and 2012 yeah and then 2016 comes along yeah and it goes differently the popular or not you know the the sort of the the common myth explanation was well that was the one Nate Sila got wrong so so just explain why that's wrong in your terms so there are two things one is we look toward how these are calibrated over the long run right so if you have Trump as a 7030 Underdog you actually the 30% chance is supposed to happen 30% of the time if it's supposed to rain 30% of the time then it it will rain three out of 10 times um but also I came into politics by way of being a poker player and being and making Sports models and things like that and being a gambler um and in gambling or investing it's where are you relative to the consensus so in 2016 Leicester city when the Premier League and they were 5,000 to1 underdogs going in right if I had a model saying they were 50 to1 underdogs instead that's going to be wrong 98% of the time but you would have made millions of pounds betting on that outcome as a long shot outcome um and so because the betting Market said oh Trump has a 15% chance 6 to1 and we're 3 to1 or two to one then then you know we were on the right side of the conventional wisdom and again I understand that for the village that's like not how they think about politics um but like but like that's how I look at it right like as someone who actually literally would think that actually placing bets on politics is an honorable thing to do but when it came down to it on in 2016 you you you thought yourself Trump would lose no I didn't I thought that there was a 30% chance that Trump would win which means that there was a 70% chance that he wouldn't yeah I mean if I'm rolling a dice right I think there's a one he was more than double likely to lose than to win yeah you know but most people thought he had no chance of winning yeah I mean if I were to invest in Amazon stock early in the lifespan of that and say yeah it's a goofy idea it probably won't work but I think the Market's undervaluing this let's buy a few shares right um is that a bet that Amazon would fail I mean I guess technically but like it's a it's a good bit relative to the price on being offered what what effect did that have on your standing you know with the people you respect among the people I respect very little but people I don't respect a lot I suppose right um you know because we actually went out of our way at 538 to explain that this is a very close election that Hillary Clinton had gotten complacent by not campaigning in States like Michigan and Wisconsin for example um that you know polling errors of a few points happen all the time they're pretty routine they hadn't happened in recent memory Obama plls been accurate in 2008 and 2012 um but we kind of stuck our neck out and took a lot of criticism to say this election is a long way from over it's a close election could go either way by by the way brexit had happened earlier that year people you know rain had only a tiny half a point lead in the polls people treated it as being certain when it wasn't so we had internalized that lesson um but yeah there was a lot of I think shooting the messenger after the fact but but look I mean I don't control the polls I don't control how people vote I don't control things that happened in that campaign like the FBI director reopening the investigation into Clinton's emails with eight days to go in the campaign um and that's why we have to be probabilistic I mean things like you know a month ago month and a half ago there was an assassination attempt against former president Trump um if he had ducked his head slightly differently that would have radically changed American history sometimes events that you don't foresee in advance are going to affect things and that's why you have to think look look at things in terms of setting odds and not certainties and and sometimes things affect things uh in surprising ways don't they I mean you you mentioned the the the shooting of Donald Trump's ear and at the time that happened um I mean I remember the moment that happened I turned to my wife and said well that's it it's a landslide yeah um yeah yeah I mean and even then even probabilistic Nate at that point was like okay maybe he's just blessed somehow he's going to but but but within five or six days it seemed like the whole news cycle was over and the assassination attempt was irrelevant and now it doesn't also think about it had some unpredictable effects right so number one it temporarily put Trump in a big polling lead against Biden which may have caused two decisions right one was picking JD Vance who is not a very good candidate um in terms of polling at least I'm not trying to be you know normative here but like he's not polling very well and they kind of knew he'd beat a mediocre candidate but thought hey look we're going to win anyway let's pick the guy who will cement our Legacy and be kind of fun very Maga make America great again candidate um and two Democrats polls got so bad for Biden including their internal polls that finally they were able to twist his arm enough to force him out right so ironically without that Trump might have been in better shape even though it made people more sympathetic to him understandably in the short run so coming back to the question of risk then I mean you know you say effectively this this election is 50/50 right now now what what effect that normally has on political parties is to to to go very very low risk you know to try and TCT to the center so you see Trump trying to distance himself from anti-abortion campaigners within his party you see um kamla Harris uh being very cautious over Israel um are they doing the wrong thing you know should they be more risk taking look I think the evidence shows that in general playing for the center of the electorate especially in a two-party system where you don't have a lot of choice um is the positive expected value play um that you want to play for the middle um that when you finally build up a largic ho if you're you know if you're running the government in California or something where you're kind of guaranteed to have a Democratic governor then that's different right then you want to like actually pass more Progressive policy but no I mean given given this takes the election I think any time a party is able to credibly tact to the center um then that's tact to the center that's I think a a a good strategic Choice yeah positive expected value for those of us who are still getting our heads around the terminology that that's that's the right that's where you want to be yeah and look and look Trump has been I mean look he's been inconsistent on abortion and American voters know that Republican candidates for Congress are likely to try to restrict abortion further um but at least Trump has some type of instinct that um look I used to run a casino although not very well it's kind of hard to lose money the casino business and Trump managed to do it um but still he has some Instinct for strategy um that I think you know has allowed him to win one election and now he's 5050 to win a second one despite having you know lots of deficits in other ways you you have a chapter on AI in in the book I mean is AI able to do what what you do with the model and is it able to do it as well or better the short answer is no but the reason is a little bit of a quirk which is that politics is actually like a small data right um we only have like one election every four years in the US so maybe we have 20 elections where we have any type of of reliable scientific polling um 20 is not a very large sample size so you therefore have to do more guesswork make more assumptions um it's not the thing like where you can like scan all the text on the internet and predict language models because that's a much that's a big data problem I'm working in a small data area and and and so where what what do you see as the the lik of AI playing out more in you know in these sort of risky roles um so look I mean looking look looking at some of the basic facts um people inic in Silicon Valley who run these Labs like Sam mman say out loud that it could be very risky but the risk outray the reward or the reward outweighs the risk right he thinks that yeah maybe we'll like wind up destroying civilization if it's misalign with some small percentage probability but we can he says we can cure Global poverty and solve all disease if not right um you know I suppose you know I'm a little more skeptical of both of those eventualities I think it most likely be more like other technologies that Society grows and adapts to but there are some differences um one is that we don't really understand very much about the inner workings of language models like chat GPT um they do amazing things but they kind of are these like magic boxes and that I think is always very dangerous right whenever you have a model I want to know how the model Works to make sure that you know as a scientist the researcher what you're doing um two is something we talked about a little bit before about how hegemonic it is how you need to have lots of powerful computing power um to make systems that are faster and that are trained for longer on more powerful gpus and so it's not Steve Jobs in his parents garage working on a PC in his backyard right it's the already powerful you know meta and Amazon and Google open AI Etc um and thropic and the Chinese government running these models training these models and they'll you know so that's going to be a win or take all Market people who are already winners that kind of defies the traditional logic of Silicon Valley so so there is a significant danger that by the time we realize the outcome of the risk it's too late perhaps I mean look there have been some good things one is that you've seen this kind of plateau in in llm large language model development where um you know there's been no Great Leap Forward since chat GPT 3.54 there have been minor improvements but nothing nothing astounding um also the community you know there is a broad consensus in the community to treat this as a risk that we have to take seriously to treat it as analogous to pandemics or global warming or nuclear war which is we don't know that much yet this is a relatively new field the progress that's happened in the past 10 years has been astonishing to a lot of people but you know there is a broad consensus among both academics and Industry leaders that this is something we have to and government that we should think about this carefully and and and you know and have smarter regulations potentially do you think you're making the river sound a bit cooler than really is you know like the gang that we all really want to be part of or really ought to be part of um I mean if you look at the book I think it plays a little bit of a plays this card carefully so to speak um you'll meet a lot of poker players in the first half of the book who I think are are admirable characters um and interesting and fun right um you know poker's played for relatively high stakes but it's not going to dramatically impact Society right if I lose a poker hand it won't cause the whole world to be destroyed um and then but you know the kind of the ultimate character in the book an anti-hero Sam McMan freed who is a ravarian very much um and so yeah I look I don't as an author I believe in in showing not telling right I don't want to beat you over the head with oh this person's a good person this person's a bad person but the fact that you know there are nine chapters three focus on one guy SPF who is very much ravarian and is not treated kindly because he you know he stole a lot of money from people um I think that that you know I think it's more nuanced than people might assume but but you you sort of end with a with a sort of a warning about what the role of politics and Humanity should be yeah what is that well look one comparison you can make between Ai and history is in the industrial revolution when we had that happen um in the late 18th century excuse me am I getting that right yeah 1790s roughly um alongside that came political revolution um you had the American Revolution and the French Revolution in particular you had the European Enlightenment um you know the values of democracy and rule of law and free market economic um if we are in a world A Brave New World where AI is dominant or oligarchs using AI are dominant um that's going to create a huge amount of political upheaval and the book tries to articulate some values that are shared values that that we can have if we do want up in that Brave New World um so one for example is agency so I'm a big fan in Liberty I'm kind of Quasi libertarian um but agency means that you have like real choices to make um that you're not having a button pressed for you and saying oh you can opt out if you want that you have you can have good information that you have two or more actual robust options you can exercise as a person that you're not misinformed about them that you're not coerced into them like that's very important um you know I worry we might be in a very inequal world where yeah people who are are high IQ and well networked and have capital and are not and you know have good health and things like that are in the right social classes can live like giants like never before and and other people are kind of like sorted by the algorithm into into different groups are tense and like they technically have freedom but like actually don't have agency over their lives and and um and and do you think we will have to organize differently in order to get a handle on the way things are right now we we might I mean I I think we have to start thinking about this stuff I think we have to start thinking about the fact that um you know we're all depending on information that's like more easily manipulated um however I'm very wary to move away from Traditional Values of free speech and Liberty and things like that I mean you know that seems pretty dangerous again you know the fact that we've had in the US especially this long-term decline now for for 30 or 40 Years of trust in institutions um which because it's not a zero some game right um you know when you build public infrastructure for example or I think even capitalism is not a zero some game it in in theory it should stand to create new technologies that help everybody in the long run but but yeah I I don't know you know I would think about how do we reverse the decline in trust in society but H could you just explain having spoken to a lot of these people why so many of these sort of tech Giants if you like seem to think that free speech is under attack when when everything about their words and the platforms that they own it's under attack of X are all about people saying the most free often irresponsible untrue fake things I mean you without without punity I mean I think you see um you know the fact that you have the founder of telegram arrested the fact that you have Twitter turned off in Brazil I mean those are those are direct attacks on Free Speech now you can say they're worthwhile because the benefit outweighs the cost but they are absolutely a retreat from Free Speech values why are they an attack on Free Speech rather than attack on the business ethics and behavior of those individuals I mean you know I have an American understanding of free speech um which is that the controversial cases are precisely the cases that need protection right if you say okay don't say anything that gets you out of Lind and then you're fine well that's not really free speech right um and so so yeah I I think people need to be able to you know take risk with the way they speak because speech does have consequences that's the whole point um this is part of the risk aversion of the village is being unwilling to to tolerate speech that can have real consequences and and can create misinformation but like that's that's that's the value you know the value is that we believe in competition and Liberty and that we'll and that having more information in the world will sort itself out eventually positively and that when Elites get get their control speech that they're often wrong right you had censor over things like um like you know the lab leak hypothesis which is now regarded as a credible Theory I don't know what happened there um like that wasn't very good in the US you had censorship of you know Hunter Biden's laptop which I couldn't care about but it turned out that like that was correct information that was misreported as misinformation and so I I think this censorship culture gives rise to Too Much political partisanship and does that mean you you in your terms you you should just then tolerate the straightforwardly wrong indefensible and irresponsible responsible such as musk tweeting about British I'm proud to be an American who listen on the American Constitution I believe that the American Constitution protects those Liberties and I wish the UK were more like the US in that regard but I mean in that example say where where musk musk was responsible for um you know publicizing fake news that possibly inspired violence on the streets of Britain so is is you know does Free Speech have no limits um I believe that the limits imposed by countries like the UK and the U and the EU are far too on the far too on the side of restrictions yeah but should there be any restrictions in your term I think it's a straw man just explain that I I think I think we're not debating about the absolute limit you know we're talking about like in Brazil no I'm I'm talking about that example so that example about musk and the UK riots a few weeks ago that should be protected speech I believe yes that should be protected yes so he should be allowed to do that even though it had clearly bad consequences that's the whole notion of Rights I mean you were the one who was criticizing me before for utilitarianism right and I'm saying no actually believe that morality involves giving individuals Liberty and rights and protecting those rights even if they have harmful consequences and so if you could change the world in any way how would you change it I'm not sure if I'm someone who wants to change the world right I want to try to describe the world for as it is and I think the basic ideas of the Enlightenment including including you know individual liberty including democracy is a very very important part of this right um including the free market um and by the way we've come a long way from then in terms of expanding the franchise to women to people who are not just white right to lgbtq people and things like that I think all of that's been pretty good and I want to take liberalism and adopt it for for you know a potentially Brave New World of technological adaptation Nate Sila thank you very much indeed for joining us thank you of course