Hey everyone, Eric here.We've got something exciting in the works and we want you to be the first to know about it.
Turpentine, the network behind the show you're listening to right now, is launching a publication and we're offering early access to our listeners.
We'll have our biggest hosts and expert guests writing pieces and leverage our group chats for content inspiration.For an early preview, drop your email at the link in the show notes.You can also head to turpentine.co slash exclusive dash access.
Now, onto the show. Welcome to the riff, where writer and investor burn Hobart and I discuss the major inflection points caused by technological change.
Our weekly conversation covers the obvious and not so obvious ways in which markets and businesses will adapt as a result.Let's jump right in.
Hello, sir.Hey.Hey, it's good to see you at Hereticon.Yeah, absolutely.It was fun.Before we get into election stuff, any reactions or reflections on Hereticon that should go on the pot?
I think there's an interesting question about how heresy has changed over the last couple of years, even since the last Hereticon.
It does feel like one of the things that inspired the original Hereticon was this observation that there was an election in 2016, and it was pretty close to 50-50, and then there was a candidate who was just completely anathema to support in fairly large numbers.
large cities where people should be kind of sophisticated, open, tolerant of multiple viewpoints.And there's also a justification for that.Trump is an unusual candidate and you can definitely have lots of really good reason to oppose him.
But I think that observation was kind of interesting that there was maybe a monoculture in tech that didn't really need to exist or that was hard to explain and was worth resisting.But Now, you can't make that claim anymore.
Maybe there is some other thing that is close to 50-50 nationally, but completely unacceptable among tech people and other high-status professions, but it's not what it once was.
And that means that the two ways that a heretic can go, or that a self-styled heretic from a couple years ago can go, would be One, you say, well, actually, when I supported Trump, that was just my concession to moderation.
I'm actually much more extreme than that now.And then another one is you could say, well, I support these things still.They're just more mainstream now.I think you've lost a lot of the excitement and fun of saying something controversial.
In fact, there was at I think at Hereticon, the more heretical thing was to say that, yeah, on balance, Harris is a slightly better candidate.
And so I'm going for Harris or like, you know, I like some things about Trump's policies, but I think that January 6th or whatever else just completely rules him out as an acceptable person to vote for in America.
And so, yeah, you just maybe maybe in that sense, Hereticon 1.0 served its purpose very well.The Overton window has been widened, but. just relentlessly widening the overton window eventually just has night of payoffs.
Eventually it just means that you just seek out like randomly controversial things or you know that you always define yourself as part of this oppositional culture.
Like actually I think maybe the libertarians would be an interesting example of this because in many ways I think if you I'd have to double check the platforms to be sure but my suspicion is that if you look at party platforms from like half a century ago
that the libertarians have actually accomplished more of their goals than any other party where it's like taxes, you know, top marginal tax rate is down since then.
And there was that burst of deregulation in the 70s and early 80s, kind of a mixed bag since then. On immigration, border, it's much more possible for people to move to the US.
And then, of course, on social issues, legalization of more drugs and on gay marriage and things like that, it's been just total libertarian victory.And I think if you're total libertarian in, say, 1974, that
These are the viewpoints of the major parties on these particular issues.
The Libertarians might assume there had been some kind of really big transition where the main parties were the Libertarians and then whichever of the Republicans and Democrats had survived, or maybe they would have merged into the statist party.
that totally didn't happen.Actually, what happened was like each party absorbed some of these libertarian issues and also absorbed issues that were just not partisan issues at all beforehand.
And the party system turned out to be a lot stickier than the beliefs of those parties.And we see another iteration of that post 2017, where like a lot of the people who were part of the Democratic base are now Republicans and vice versa.
So the party is sort of they did this trade where a lot more blue collar workers and lower-education workers moved to the Republican side, and then a lot more high-earning elites moved to the Democratic side.
But the overall balance of power is really not that different.It just slowly converges more and more on 50-50 as each party gets a little bit better at targeting voters and as the market in political ideology gets more and more efficient.
So yeah, all that is to say, if you call yourself a heretic, you're calling yourself a heretic in reference to a particular set of beliefs at a particular time.
If you're just a universal heretic, then your viewpoints have no information content because if you get everything you want, you just have to switch all of your opinions to something else.
So but I do think like it's it's useful to to bring together a bunch of people with heterodox views and just get them to argue.
And there were there were interesting discussions about things like do vaccines actually work where I had not really gotten the the sort of cogent, informed, heretical view that no, they don't.And I still bet that, yeah, they probably do.
And I'm still you know, I still get vaccinated.My kids still get vaccinated.Like on balance, it was not not a persuasive argument.
But I think one of the things you always have to recognize about contrarians is that they almost always have better information on on the issues than you do.You know, now now there's this sort of like default soft atheism.
But back when Internet atheism was big, it was pretty common for the Internet atheists to be a lot more familiar with the Bible than the people they were arguing with.Not in every case, but in many cases.
And I think you'll you'll find that with a lot of a lot of issues that like the people with really weird views do have a lot more data often to support those views.
It's just that if you, you know, if you if you just enjoy trolling people and you enjoy this oppositional thing where because you're always opposed to the people in power, you never have to make hard choices about what to do once you're in power.
That can be fun.It can be a way to learn a lot, but it is ultimately fruitless.So like be if you're a heretic, if you're never a heretic, that's a problem.
But if you're a heretic for your entire life, while everyone else is always changing their opinions, that is that is also a problem.It is just a different kind of conformity.It's it's like the low agreeableness conformity.
But I do think hereticon is much more tilted towards like there are interesting people with weird ideas.Those ideas have shifted closer to the mainstream.We can kind of haven't figured out what is the meta Overton window of like
What views are acceptable to people who think that there should be a wider range of views and are not just totally anathema to everybody?
Yeah, it's, it's, yeah, it's interesting.It's, it's turned from sort of the, the conference of heretical ideas to basically the tech right and, and, you know, tech adjacent right and friends.
Right.It's check adjacent right and right adjacent tech.
Yeah.Yeah.Yeah.No, absolutely.Let's, let's segue into election stuff.So you have some good thoughts around how even just the nature of how we think about elections has changed because of prediction markets.
But the thing I really want to know and our listeners want to know is how do we get rich?
How do we trade on if we, if we're, if we're so sure that Trump is going to win, or if we're so sure that Kamala is going to win, you know, what, what should we be doing here?
I think that's mostly the wrong question because it's just such an incredibly competitive market.
It's also a market where you shouldn't be all that confident in your variant views unless you've identified a specific material error that everybody is making and you have high confidence that they'll correct it next time.
you're operating in this environment.And Nate Silver was complaining about this, where he said there was a recent round of polls where everyone, every poll was like up within 1.5 points of his prediction.
And he realized that they, the pollsters are just dragging their polls closer to that media.And they're, they're just shrinking the variance because they don't want to look really bad and have a crazy outlier poll.And that is
That's a reasonable thing to do if your poll is trying to be a prediction of what the outcome will be.
But if your poll is just trying to be whatever it is, where you have a methodology, you ask a set number of people a certain number of questions, and you're some degree of transparent about how you turn that into your estimate,
then it's actually a disservice to the analysts to try to make that poll adjust for what the consensus is.You want the poll to be a fresh new data point that is only influenced by the responses that people gave and not influenced by other stuff.
So I guess talking about the election, for viewers who are interested in context, it is roughly noon on election day.
So nobody has known anything for the entire duration of the campaign, but this is the point of maximal ignorance and the point of maximum potential embarrassment if we make any definitive claims.
I think in a very early episode of The Riff, I did say that I thought it was 50-50 and that I heard a good argument from Nate Silver on why it's roughly 50-50.And that argument was actually tied to the fact that it was going to be Trump and Biden.
And so you just had two people who've been public figures for my entire lifetime.And since I'm right at the median age of Americans, these people have been famous for longer than the median American has been alive.
So yeah, there was nothing to change your mind about.And then it turned out the election became more of a referendum on Trump and that Kamala was very smoothly swapped in for Biden when it was clear that Biden was just not going to work.
And then we got right back to this 50-50 dynamic.It just turned out that Trump is just such a weird aberration compared to normal politicians that it is still Trump versus whoever, no matter whoever that whoever happens to be.
You'd probably have like slightly different marketing if it were someone else running against Trump, but you'd end up with roughly the same, the same questions, the same confusions.
And so I think on the, on the, the idea of polls, like pollsters do want to adjust for the fact that people are not consistent and how they respond to polls.And they're not consistent in how likely they are to pick up the phone.
They're also not consistent in how likely they are to admit what they believe.
But I think there is this possibility that the shy Trump voter phenomenon exists more in pollsters models of how to adjust the numbers than it does in the actual hearts and minds of voters.
It's just for very unpredictable reasons, more socially acceptable in 2024 to be a Trump person than it was in 2016.I don't entirely get that.It does feel like Trump is a more polarizing figure after his first term.
I think the depolarizing thing you can say is some combination of, look at what he did over his first term.He wasn't really serious about a lot of the extreme stuff, or look at what he did over his first term.
He obviously cannot accomplish all of this extreme stuff that he generally wants to accomplish.
But either way, I think that if you look at the polls and you assume that the error is of the same magnitude and in the same direction as the last couple times, you have to ask yourself, did the pollsters not know that they were wrong and what direction they were wrong?
And if they make that adjustment and then voters get a little bit more willing to say that they're Trump supporters, then you actually have them over-indexing on Trump.And so you get a correlated loss.And then
Even further complicating things, we do have early voting numbers and those numbers are what they are, but they're also completely impossible to compare to 2020's early voting numbers.
You just have very different correlations between early voting and party identification and the sample size is small.
So anyone who has a really high confidence view, my guess is that they are underestimating the extent to which other people are factoring in whatever their special view is.
Nate Silver actually, I think in 2016, the thing that made his model give Trump much better odds than other people did was that he did take into account that there would be correlated errors and correlated swings across different states, such that if you have a bunch of swing states that are slightly Hillary favored, the conditional probability of Trump getting one of them goes up a lot if he's got another one of them.
So, it was basically, I think the main thing that Silver did right was knowing that there were these correlated swings and that therefore the odds of a big Trump victory were not just the cumulative odds of Trump winning a bunch of 40% chance things.
Those were actually not completely unrelated, completely independent events. So yeah, all that is a long-winded way to say, if you're betting, have fun.
You are making the market more efficient and you are making it much easier for me to just hit refresh and see what's actually going on.
But you're probably not, you probably don't have a big edge, but since it is 50-50 and since the prediction markets are a lot more liquid than they have been historically, you can put a lot more money to work.
Like there will be people with great bragging rights.It could be that Frenchman is just, you know, a famously successful trader after all of this, but it's also possible he got totally wiped out.
Or yeah, by the time you're listening to this episode, he may have been totally wiped out.
I heard that a couple sort of non-obvious effects of what happens here is one, that voters tend to vote for whoever's winning or something.
So the polls are somewhat or sort of market results are somewhat self-fulfilling in that they influence voter behavior.And that's a response to someone who says, why do we care about all this stuff?It's a coin flip.
We'll find out in a couple of days what happens. that there is some sort of, you know, influenced behavior that this influences.
But then also that there seems to be this almost like efficient, you know, like market thing that tends to, like, why is it always 50-50?Or the game theoretic?
Well, the why is it always 50-50 thing?Like that is, that is just the result you'd expect from a two-party system where
whoever wins the majority of the electoral votes gets the presidency and where you don't have a parliamentary system where people assemble coalitions and form a government after they know what the voting results are.
In that system, you would expect that if there's a party that is 55-45 favored in the majority of elections, there will be some interest group in that party who knows that if they switch to the other party,
they have turned that party from just a rump party that exists to lose elections to a party that actually has a viable shot at the national elections.
So whatever issue it is that is the weakest part of the winning party's coalition can move to the other party and suddenly make that party viable.And there's an extent to which Trump did that with immigration and trade where
Immigration wasn't a Republican-flavored issue.I think trade and trade protection was more of a Democrat-flavored issue in the 90s and 2000s.
It's tricky because there were intra-Democratic fissures over do we like free trade, do we love trade deals, or do we worry that this is big companies shipping American jobs overseas and getting rich off of it.
But I think it was net more of a Democratic thing.
And that did mean that Republicans just had this electoral disadvantage where a lot of union workers would just by default vote Democratic, and they were just a very reliable category for the Democratic Party.
And Trump was able to identify that or like implicitly able to identify that as an opportune set of voters to go after.And he mostly seems to have mostly gotten them.
So that's why you get 50-50 is that anything other than 50-50 is unstable as a coalition, because it means that there's a huge incentive for someone to defect from the winning coalition.
and then be the reason that the other side is now the winning coalition.
And then I think that that's probably getting stronger as an effect just as the media environment gets more liquid, as it gets easier for people to become high-profile politicians because their clips get shared rapidly on social media and they can go pretty direct to their supporters for
And then their supporters are really, really jazzed by the idea of having been an early backer of someone who turns out to be really, really important.
And I think both Trump and AOC have a lot of that where people feel just really invested in their success because back when they were an underdog, those people were still backing them and backing them when the candidate was, I think AOC was seen as, I think less of a joke and more of just like meaningless long shot thing.
And with Trump, yeah, it was people, people felt very vindicated that after election day, they could, they could ask like, who's laughing now.
Hey everybody, Eric here with a word from our sponsors.
Well, one thing I'm surprised by is basically why don't more candidates try to take issues that are like, you know, which is more popular and, you know, turn the 50-50 and like take issues that have an 80% of, you know, popular support.
It feels like they're not as responsive to voter sort of preferences as they might be incentivized to be.
I don't know.I think that's much more of a David Shore question than a Martin Hobart question.
It is a good question, but I think it's something where if I knew offhand the ideological stories of a couple dozen prominent politicians, then it would be easy for me to answer that.I don't know for sure.I think there are
One one piece of it is just that it takes a while for someone to get ramped up on an issue and there's like ramped up in the sense of you understand the policy tradeoffs.
But is that really what you need to actually persuade the American electorate like to persuade that that median voter?Probably not.But there is a ramp up period of just.
defining yourself in terms of that issue and making that issue synonymous with you.And that's something that actually does take time and does take effort.
And I think if you look at something like Bernie Sanders on health care, where he does, he is in this position of basically always telling his party they could move a little to the left on health care or they could expand Medicare a little bit more.
They can make these programs a little bit more generous.Like that is his thing, but it's taken him a long time to actually get to the point where that is a thing that he is reliably doing. and direction he's reliably talking to the party.
And there's also, for a lot of high-profile issues, there is going to be a lot of competition.
So you sort of have to do the sort of venture thing of you find an issue that's not a big deal right now, you make it your issue knowing that it will be a big deal in the future.
that again is something that i think trump did where immigration it was something where yeah the republicans did have a little bit more of a restrictionist view than democrats but actually the the trade thing was kind of tied into the immigration thing where whether you were worried about competition from foreign goods or competition from foreign workers in the u.s like either way
If the U.S.border is a little bit less porous to the movement of people or of goods, then those workers feel a little bit safer and more secure.
And also they feel like they don't have their back, which I think is really important thing for politicians to do is just to make voters feel like
there's actually someone looking out for them, actually someone who's thinking about their interests and is trying to pursue policies that favor those interests.And that's like, you know, politicians.
I'm sure some people get into politics out of just pure lust for power.Perhaps it's a lot of them.But I think a lot of them do view the political system as a way to enact positive change for people they care about.
And it's not just like immediate friends and family, but like whole categories of people where they feel sympathy for these people and want to do things that make their lives better.
So yeah, you, you do have, you have some incentive to, to do that for politicians who attach their, their name to a particular set of issues that have very high approval, but it's just hard for them to do it.And it takes time.
Yeah.Yeah.Is there, I do want to segue to sort of AI at some point.Is there anything more on the election that's, that's worth getting into before we do?
Yeah, yeah, let's let's make fun of Nate Silver's critics for a little bit.So Silver in at the manifest conference in 2023, he said it's basically a toss up election.He probably won't bother publishing a model because everyone will just yell at him.
And then he did publish a model.The model show that it was basically a toss up election with some fluctuations and everybody yelled at him.
But I think what's useful about that model is if you do think it's a toss-up election, you do still have to ask yourself, how do you update based on new information?And I don't think that it's a trivial and straightforward thing.
I don't think there's a really obvious range in which the odds should have fluctuated.
And I haven't actually seen anyone produce their own chart of what they thought the odds were, where it was always 50-50, but fluctuated between 45-55 and 50-50 versus getting as high as 60 plus or something like that.
So I think knowing knowing what the amplitude of those changes is does does tell you something like it does tell you basically how much attention you should pay.And I think that is the actual reading of Nate Silver's work over the course of this
cycle is that, yes, it's 50-50, but also there were things that swung the odds pretty dramatically in different directions, and that there were just points in time where, had the election been held, it would have been a landslide in one direction or another.
And that's very different from an election where if you'd held it any time over the summer, it would have been a coin toss.
Like there were definitely points this summer, like right after the assassination attack, for example, where there's no chance it would have been a coin toss, like it would not have been 50 50.And so I think I think that is useful information.
And I also think that the other flip side of this and unless you publish this episode really, really quickly, this will be useless information.But.
the the other side of an election is 50 50 and therefore everything all the trends are mean reverting and you should just heavily discount any information that you get over the course of the election but that also means that once it's actually election day like once the information is not here is the best guess as to what will happen in pennsylvania instead the information is
here are the votes cast in these parts of Pennsylvania.Once that happens, you do have to update.
Because you're close to 50-50, 50-50 is also the point of maximum leverage where every new piece of information either tells you this person wins or this person loses.Suddenly, those same moves do really, really matter.
That's something that I think is another takeaway from silver is that we should expect to see much bigger swings just over the course of election day.
And, you know, depending on which market you're looking at and whether it's defined based on which major publications call the race one way or another, or on who actually gets inaugurated, you'll, you'll still see some, some exciting volatility.
Let's segue to why is it reasonable to believe in AI apocalypse, but not short the market?
Okay.Yeah.Yeah.So I love apocalypse trades just like the concept.I don't really do them, but I like them.
And the thought experiment that you want to use with when you're thinking about bets on existential risk is if you get a push notification on your phone that says North Korea has launched a nuclear weapon and it's going to hit you in 10 minutes, what, what financial decisions should you make?
assume that you already, maybe assume that you have the latest version of Apple intelligence.So it has automatically said, I love you to all your loved ones.
It has summarized all the I love you texts that you've received back in your last few minutes of life.So having gotten that kind of thing out of the way, you can focus on what really counts, which is having an optimal portfolio.
And so you might think to yourself, okay, SF is getting nuked or Austin is getting nuked.And therefore I should be betting against American equities and particularly companies that are headquartered here because they all get wiped out.
But if they get wiped out, you get wiped out.And so you don't actually collect on that bet.Whereas if the push notification is a false alarm, but a lot of people receive that notification, then they have an incentive to sell stocks.
So like a Japanese investor, for example, they have an incentive to sell U.S.stocks if the U.S.gets nuked by North Korea. You have a different payoff function because you only collect in the event that it's a false alarm.
So you should actually be buying US stocks from foreign investors and you would be long.And then if you die, you die broke, but you're already dead.You don't care.If you live, then you actually got a little game.
And the market does not react all that strongly to existential risk.The market was not incredibly happy about the Cuban Missile Crisis, but didn't collapse.
And I think some of that may have been people asking themselves, if I did sell all my stocks, I would just be a corpse with a lot of cash rather than a corpse with equities that have a slightly lower dollar value.
Now with AI, you have an additional complication, which is that The apocalyptic vision of AI does involve AI interacting with the real world economy, and it involves interactions that are a positive sum until they become very negative sum.
So AI optimists, on any given day, the more pessimistic you are about AI, the more optimistic you should be about NVIDIA's next quarter, because every time more GPUs are deployed, that brings us closer to the end of the world, but it does mean more GPUs got sold every time
Meta launches a new version of Llama.It means we have more powerful models and we can't control their deployment because they are, in some sense, open source.So again, closer to apocalypse.
But every one of those is also probably positive for GDP, positive for tech stock valuations, et cetera.
So the internally consistent thing for somebody who's worried about existential risk and things that's happening really, really soon, the consistent thing for them to do is be levered long every company that has economic upside from AI
with the explicit plan of at some point flipping from I am betting these stocks go up to I'm cashing in my bet and using that money to avert the apocalypse.
So it's basically like you pyramid, you're out of the money, Nvidia calls until you've made enough money that you can personally buy the rocket that is used to blow up the data center.
So I think that's an internally consistent thing to do, but I would also not expect someone who was literally planning that to tell someone that that was their plan and that's why they're not short the market.
But I don't actually think it makes sense to short the market if you believe in existential risk from AI.If you believe that AI will be
net harmful, but maybe it's not harmful in the way that leaded gasoline was not harmful, where it actually solved one problem.
It solved the engine knocking problem, made cars run more efficiently, but then it caused a lot of brain damage for a lot of people, so not very bad. In that world, you do want to be short.
You think that there's a world where there's this benefit and there's a cost that more than outweighs it, but that would be a very, very unique view of AI pessimism, which is actually more in the AI skeptic camps.
The AI skeptic camps basically has the view that these tools are not all that impressive and
there are some trivial problems that they still get tripped up on and the hallucinations are bad and we're overusing them etc and you know we're we're replacing good workers who can do their jobs with ai tools that are cheaper but can't actually do the job and so everyone ends up worse off if you have that view you should be short the market and shorten vidya but that's a totally different view from existential risk it's actually a view that ai is just not good enough to kill us all because it's only just now gotten good enough to have the right number of fingers on a hand or to
Be able to count the number of hours in the word strawberry or whatever.No, I don't.I don't think that view actually makes any sense.I think these people like you know.
You could also say like computers don't work because I talked to my desktop computer while it was unplugged and it didn't answer me.Therefore, computers are worthless and it's just a big expensive paperweight.
But yeah, if you misuse a tool or you don't understand how it works, then of course you're not going to get very much out of it.
And I think that's where a lot of that AI skepticism comes from, is that they do not really look into how these things work.They don't understand the use cases.They don't really understand where you would or would not use it.
And therefore, they think it's worthless. With any new technology, you don't pay attention to what does the average person get out of it, or what does the least effective person get out of it.Its use is defined entirely by what is the best use case.
That's the only use case you actually care about.And then you care about how it scales, how its costs decline as it scales, and therefore how many new use cases get opened up by that.
Is there a similar sort of a pocket scenario in crypto where you could also believe that the dollar is not going to be the reserve currency or something? but not short the market or otherwise.
That is also very narrative dependent.
So if your view is crypto is going to build up this robust financial ecosystem that does everything that the fiat system does just more efficiently and without trusted third parties, then you can imagine a scenario where
There's a crypto transition that is gradual enough that it's not super disruptive and that just continuously makes the economy more efficient.And therefore you should be long.
You could also be a crypto optimist who says that the crypto is basically like the Soros break the bank of England trade, where as money flows into crypto, it devalues all Fiat currencies that causes massive disruption in the Fiat ecosystem and eventually huge sections of the economy blow up.
But the thing that we have left after all of that actually works really well. And that is sort of like a 2008 view, because 2008 was this case where there were parts of the economic system that were just not functioning correctly.
They were continuously accumulating excess risks in a way that was totally irresponsible and somewhat it would blow up.
And then it did blow up and what it got replaced with was actually a system that is a lot less flexible and a lot less fun, but that is actually a lot more stable.Like the big banks are far better capitalized than they used to be.They take
a different set of risks than they used to, but it's on net probably a set of risks that is less disruptive.
That has led to the growth of alternative financial ecosystem that does a lot of things that the big banks did in the 2000s of warehousing risk and repackaging it and selling it and making directional bets on assets.
and knowing that they can make or lose money.But that new part of the financial system, it, one, it is also just much better capitalized than banks.
And you really, you just shouldn't have these prop trading firms that are buying all kinds of really illiquid stuff.And they're also 30 times levered.And most of their leverage is very, very short-term funding.Like that's just an insane thing to do.
You can have that kind of leverage with certain strategies, but you do want to isolate them into little pockets of the financial system.And
What you want to do if you're designing a financial system right, you just want to make sure that everyone knows roughly what risk they're taking.
And the most important thing is everyone knows what is money and what is not money and what is on the continuum of being kind of like money and kind of not like money. And how how do things shift along that continuum?
Because that's what really blows people up.And that's what actually nukes financial systems is when you have people who think that some asset is money, like they think it is always worth 100 cents on the dollar, but it gets them an excess return.
It's not money.It's something else.And it becomes not money when everyone needs money.And that's the time when you really wish you had it.So that is something that our financial system is just less likely to have in the near future.
So it is a more stable system.And yet that I think that that kind of shift, like 2008, it would destroy a lot of market value, cause a lot of disruption, cause a very serious recession.
And then if crypto maximalists are right, then what we're left with is actually an economy that grows faster and or is more stable with smaller boom and bust cycles.So we actually end up better off and in the long run, stocks go up even more.
But you could almost say that Satoshi wrote a white paper to solve the same problem that Bernanke and Geider and all the rest were also trying to solve in the fiat system, and that fiat got to good enough before perfect got to scale.
That's what we end up with.You have these multiple critiques of the same system, kind of like the Reformation, where you do actually, you have people just totally splitting off from the original institution.
We also have people within that institution saying, yeah, we actually went a little bit overboard on some of this stuff.And we need to make sure that that never happens again, that these criticisms are right.
But also that because of that, we have to like define ourselves a little bit more sharply, because we're now defined partly in terms of this other alternative thing.
One thing you also write about is how for the Magnificent Seven, AI is sort of underrated in the impact that it's having.
Zooming out a bit, 20 years from now, when we look at this AI time, similar to how we look at the time of the internet, certainly the 2000s or late 1990s, will we have a similar reaction in that, hey, there was a bubble, there were some overpriced companies or some companies that weren't in fact valuable or too early for their time, but
Is this, in fact, was a revolution and, you know, the beginning of much bigger things to come?Or might we look at it, you know, something in a different way or even more powerful than what the Internet was?
Yeah, that is a really hard question.So like right now, the scaling laws tend to broadly hold true and haven't really seen evidence of them slowing down.So it does seem like AI just keeps on going, keeps on getting better.
And we rapidly reach the point where the future is just very unpredictable.
Well, in the same way that there's this tipping point where you have more mechanical power than human and animal power, and that mechanical power can be used to create more devices that have mechanical power or extract more energy sources from mechanical power.
And it's very hard for basically someone who looked at like the first steam engine and looked at it pumping water out of a coal mine in order to get more coal that you could feed to a steam engine.
they could not have conceived things like railroads and steamships and the ubiquity of steel and stuff like that.It just would have been totally beyond their imagination.
And similarly for us, if there is some tipping point where the cheapest source intelligence is a machine and not higher person, then we don't have a good way to model what the future looks like past that point.
There's just some fundamental input that used to be gated by number of humans and is now gated by something else.
If things peter out a little bit, there are plenty of scenarios where AI turns out to be a good business decision, even though it peaks at something slightly better than the current level where you can replace some of the people on your team with AI agents, but you can't replace all of them.
And the AI agents don't get that much better from here because they're trained on the outputs of humans and there's blended average.
they can get really, really good relative to the average person, but then they can't get to the point where they're doing just extraordinary things that have never, never been done before.
I, I try to be agnostic on that, but it's probably because there's just, there are the, I think there's some value in thinking through what the scenarios look like in both those cases, but also that it's not like you can plan that far ahead.
And I don't think there are a lot of things that you can do today to prepare for either of those worlds other than kind of hunker down, pay close attention, like
try to use these tools as much as you can so that you get used to interacting with artificial intelligence at varying levels and figuring out where to deploy different kinds of it.Yeah, I think it's hard to go, it's hard to model things beyond that.
But what I did write about recently, yeah, was that looking at the most recent couple MAC-7 earnings, like people, the companies are actually getting material revenue from AI. and their investing material amounts in AI.
I think it was Microsoft where they said their results were a little bit disappointing, but it's actually because of supply issues.
And there was a nice callback to the quarter before where you had, I think both Microsoft and Meta say that they were more worried about errors of omission.They were more worried about under-investing in AI than over-investing in AI.
And this, like last quarter, that actually was the correct worry.Like Microsoft did not get quite enough, did not jam quite enough money into NVIDIA's pockets and therefore was not able to sell quite enough API calls.
We, you could imagine scenarios where we just run out of use cases for near human intelligence or like better than human intelligence in very narrow domains, et cetera.But we haven't really done that with humans.
Like we still do have a shortage of smart people.There are still a lot of ideas that are looking for a brilliant person to spend a couple of years on them.So I don't actually think that there's a demand side problem longterm.
There is just a, a market, like a matching problem of
how do we develop the right kinds of artificial intelligence, how do we identify the right sets of problems, and then how do we reprioritize when there are some problems that require intelligence to solve, but now that intelligence can be orders of magnitude cheaper and can also be run in parallel and in bursts and things like that.
Yeah, that's a good, good overview.I do want to also make sure we get to your housing post because you talked about how expensive housing isn't as clear cut as we, as we imagine it.
And it has sort of more, you know, indicators or implications that are, than we, that we understand.So we're going to outline what you were saying there.
Yeah, so I think the first way to caveat that is to say that one of the big reasons that housing is expensive in the developed world is that there are lots of unnecessary regulations that prevent new housing from getting built, especially in very desirable places.
We could have denser big cities.We have done it in the past.There are denser places that work just fine.
There are a lot of problems and a lot of complexities with that, but there are also a lot of problems and a lot of complexities with a lot of these cities being much too expensive for most people to ever conceivably live in.
And so I like the category of complexities. how do you collect a lot more garbage and how do you have a lot more cops on the corners if more of San Francisco is high-rise apartments instead of single-family dwellings.
And so that's the initial caveat that, yeah, there are a lot of unforced policy errors that make housing more expensive than it should be.
But what I was thinking about was what happens to housing prices relative to rents if the only thing that changes in the economy is a It's like one point increase in growth expectations.
And what you assume from that is, OK, people are earning one one percent more compounded every year indefinitely on average, and therefore their ability to pay that rent also goes up by that amount.
And so if you are a building owner and you're thinking about buying a building, You can basically say that your total return is rental yield today plus long-term growth in rents.
If long-term growth in rents goes up, you can actually get a lower immediate rental yield and it's perfectly offset by the future growth. And this happens with stocks too.It's a pretty simple dividend discount model.
And if that's the case, then whenever growth expectations go up, the housing price to rent ratio and housing price to income ratio will also go up.So housing actually gets more unaffordable in a growing economy.
And that is bad in the sense of it does mean that people are paying more for a product whose quality has not changed at all, but it's a sign of something good, which is people have more money to pay.
So I think the, and then when you like going back to that, the housing supply restriction thing, it does just make sense that in a denser city, there will be more restrictions on building things because there are simply more parties that have to come to some agreement on what gets built.
You, if you build a new building and it needs some extension of the plumbing and electricity, electricity systems and telecom systems, et cetera, like. things have to get dug up on the streets and sidewalks that have high foot traffic.
And something gets built next to someone who was happy with what was next to them before.They just they have some interest.They may have too much of a political interest than is really economically justifiable.
They have some interest in affecting that.Whereas if you are building a new suburb on the outskirts of a city and there just wasn't anything there, there are just fewer constituencies who can say, no, there are fewer people with vested interests.
even in an environment where there is a lot more freedom to build and a lot more freedom to improve properties without getting permission from everybody, that there will still be more restrictions in these dense cities that are also pretty high income, pretty high rent places.
So in those places, you'd actually expect the ratio of housing prices to rents to go up.And so that means it takes longer to actually buy a home.
You are better off, but just over the course of your life, you would spend a larger fraction of your total consumption basket on housing in places that are growing fast.
And that is historically the case, like housing as a share of total spending has gone up over time as the economy has gotten bigger. And I think that will continue.
This is kind of related to a point that I made in the diff in early 2020 about how when you look at housing price to income ratios, one thing to keep in mind is that if someone moves from a low cost of living city to a high cost of living city and their after tax and after expenses income goes up, it can be the case that they are better off even though they're paying a larger fraction of their income as rent.
So If you were paying say $2,000 a month in rent in one place and you move to a city where you're paying $5,000 a month in rent, but. you also get a $4,000 a month raise.
You're better off, assuming all your other costs don't change, you're better off, but also probably the fraction of your income that you spend on rent goes up.
So part of what we see if we track long-term trends in housing prices versus income is the economic dominance of coastal cities that have very high costs but also pay really, really high wages.
So I guess you could say that all of this loops back to our initial discussion on polling and correlations and things.
You always want to think about the mechanics of how the numbers you measured actually came to be and what are some of the trends that are cutting in different directions that actually lead to these outcomes.
And it's very rare that you have some time series that measures something interesting and is really only measuring the effect of one specific cause.
The more interesting the time series, the more likely it is there are a bunch of different causes, a bunch of different feedback loops, and also that people are looking at that time series in order to make the next decision that also affects what the next number in that time series is.
Yeah.I mean, it is amazing how advanced our sort of, you know, understanding and our methods of sort of, you know, assessing, creating information markets.
And yet we're still like, you know, 50, 50 coin flip, you know, nobody knows what's, what's going to happen.Isn't that kind of striking?
It seems like the more certainty you have, the less important it is to have elections. Like if each party nominates someone and Nate Silver puts up his estimate, he says 99.7% chance that it's this side.
And on the betting markets, you just can't find someone who is willing to bet on the other side, et cetera.Like, do you need to bother?Like, should we, should we all be taking an extra trip on a Tuesday in November? Probably not.
The meaningful election should be close to 50-50.If they are meaningful, it's because there are strong opinions on both sides, and both sides feel like it's really, really important that they win.I think people always say that.
I think the first election I paid attention to was 2000, and that one, it was supposed to be a really big deal, and it was supposed to be this transformative thing, and then 2004 was even more so, 2008 even more so, and so on and so on.
It just kind of kept ratcheting up.But I think it's also just objectively true that there are much more meaningful differences between Harris and Trump than there were between Bush and Gore or between Obama and Romney or something like that.
Like it is it is a different election, but it's also if you have one one side where or if you if you have more questions about each side's ability to competently execute on the things that it says it wants to do.
And you also have questions on how different they will end up being, given that at least on things like tariffs, there's a fair amount of continuity on tariffs and on China policy.
Generally, there's a fair amount of continuity between the Trump and Biden administrations.
Like, maybe part of what you're picking up on in these really important elections is a shift in voter priorities that's going to be expressed through which politician wins.
And certainly a politician who identifies those shifts early and accelerates them will do better than a politician who doesn't.
But if the shift is just a change in people's priorities and the change in the salience of different issues, then it's going to have an effect and it's going to make these elections feel high impact.
But the really high impact shift was the change in the underlying opinion of the electorate.
Now, if if this election does end up being decided by the squirrel used to promote an OnlyFans account or something bizarre like that, like that, that will be disappointing.
But it is also the case that the more if it's a really important election, it's a 50 50 outcome, really, really high stakes.That does make it more sensitive to totally random stuff.
That does mean that this is an election where things like whether or not it rains on election day in particular areas in swing states does actually determine who the leader of the free world is in a couple months.
fun times, but you can't escape the contingent nature of history.
History would be very, very boring if you had a little JSON blob describing your starting parameters of the country and you can just run the simulation forward and find out the exact name and policies and opinions and scandals of whoever's going to be president 100 years from now.
That would be very boring, but the trade-off to history is interesting is sometimes it's interesting in just really, really stupid ways.
Is there an explanatory power or predictive power to Elon's quip, the most entertaining outcome is the most likely, or is that just a glib?
Yeah, I want to say some of it is survivorship bias, but there is
So there's this argument that humor has this sort of information theory interpretation, where when you tell a joke, you are continuously giving people evidence that fits one conclusion.
And then you give people one little piece of evidence that delightfully, completely obliterates their expectation, and they realize you're talking about something else.And this is also something that music does, where you
like a lot of classical music, they will, they will set up the theme.They will do variations.They will do, you know, part of the theme and then switch to something else that it's in a minor key or something like that.
And it's fun because you get, you get surprises that it makes sense that they're surprising.It's not totally random.There are patterns, but it's not just endless repetition of the same pattern.It's asking.
given this pattern, what are some fun things we could do with it?Or if you expect this pattern, what's an interesting surprise that you can get next?And so in that sense, maybe history is just more information dense when it is kind of funny.
And when the expected thing happens, there's just There's not much, that much to say about it.There's not a lot to write in the history books about it.
Like no one is going to write a history of a random congressional campaign where the incumbent started out, you know, somebody who got 80% of the vote last year.And the big question is, did they get 81 or 82% this year?
No one's going to write a book about that.It's super boring.
So it doesn't, it's a small piece of history, whereas a really big chunk of history will be things like very close fought campaigns where it looked like one side would win, then it looked like another side would win.
There were various assassination attempts and scandals and legal cases and things like that.Like that stuff is, is a lot more interesting and a lot more worth writing down.
So you could flip the Elon claim around and say, if it's worth getting into the history books, it's going to be funnier than the average thing that happens.But that doesn't quite change the likelihood.
It just, I think, changes the retroactive salience of that event.
There is this sense of history is going to give us a lot of interesting ironies and weird paradoxes and just cases where things could have gone very differently if some minor coincidence had happened or had not happened.
Yeah.I, before we get out of here, there's a couple of other things I want to get to.One is the, the idea of smaller social networks.
Well, actually first before I get to that on a point on the expensive house pricing, I heard about expensive sort of real estate.
I heard of an idea that someone had to, you know, decrease housing prices, which is don't let shine, you know, foreign investment.Don't let China own real estate.Maybe it was you, I can't remember, but with that.
be a good potential outcome of prices might drop if we don't let China sort of do heavy investment into real estate and why do we want them owning all this real estate anyway?
Yeah, I like there, there's definitely a solid sort of economic populist argument that I think is true of like, if you have a finite amount of space in your country, and you're not building quite enough housing that you are really compounding the problem if you have people who are treating this housing as a speculative asset.
And I think it's it's less China money in the US.Sometimes it's been more Chinese money in in Canada, where People did seem to be willing to just own a condo in Vancouver that they were not renting out.
So they are just continuously bleeding a little bit of money, but it means that if they ever need to leave China and go to Canada, they have an asset they can sell.They will have enough money to get by.
That doesn't seem to serve the interests of the average Canadian all that much, and so it does make sense to question that.
But there's also – if people are buying housing as an investment and they're renting it out, then it doesn't change the total supply of housing.In fact,
There was this recent paper arguing that for the large institutional landlords, they actually basically effectively increase the availability of housing because they just manage things a lot more efficiently.
They leave their buildings empty less than small-scale landlords do. So it might actually net decrease housing costs.
Although if you look into it, like one of the one of the reasons they have a cost advantage is that they pay lower property taxes because they're more aggressive about disputing those taxes.
So it is like sort of redistribution from like towards average people in their capacity as homeowners, but away from average people as they're in their capacity of recipients of local services that are funded through property taxes.
So there's there's a little bit of ambiguity there depending on where, what part of the system you're actually isolating in order to analyze.
I think if you do care about, like the place to care about overseas investors owning lots of real estate in the U.S.
is, and this is something that was actually, it's getting more media attention lately, that there are entities that appear to be linked to China that are really, really interested in buying farmland that has a line of sight to U.S.military bases.
And that is not the kind of thing that China would allow, I don't think that Russia would allow an American firm to just buy lots of land all around Russian military bases.
Even if they didn't catch us spying, they would look askance at that and assume that there's a motivation there.
For all I know, there is some number crunching Chinese real estate private equity firm that has done a bunch of regressions and realized that actually
Buying land near military bases is the new buying land in suburban Redmond because Microsoft is adding so many employees and they're all just going to buy these houses that give them a 30 minute commute to Microsoft HQ.
There are theses like that, but this is probably not that.This is probably
some kind of longstanding gradual, maybe kind of clumsy spying campaign, or at least it looks suspicious enough that I would love to hear a really, really good explanation for why this phenomenon exists.
But if overseas investors are overpaying for property in big US cities, I think a better solution is just make it easier to build lots of housing.
If New York made it much easier to build housing, and if San Francisco did that, and if one of the results of that was that
A lot of corrupt oligarchs from various countries that do not get along too well with the US suddenly have one of their big assets mostly wiped out or at least have its value significantly impaired such that they give up and sell it and just stash their wealth in gold bars and fine art or whatever else they like to buy.
Maybe that's a win for everybody.We get more housing, we get to slightly soak the rich, but it's not America's rich who are all wonderful and virtuous.It's actually the enemies of America's rich and who knows what they're up to.
So maybe that is probably probably a better solution.I don't I just don't get too worked up about foreign investors buying lots of property in the US because I don't think it's a huge impact on the overall supply of property.
They don't they don't like to buy the houses that a newly married couple is going to move into so they can start having kids.They're they're buying really, really expensive penthouses and things like that.They're buying luxury properties.
And so it's not the people getting priced out or just not that politically sympathetic.And I
I really don't worry that much if there is some distressed credit trader who would be able to pay 8 million for this condo, but has to pay 10 because there's a Russian oligarch who bought the one next door and pushed up the prices.
Is that economically optimal?No.But is that the biggest problem that we face or anywhere near that?No, also no.
Do you have a hard stop right now or should I let you go and we do it?Okay.Maybe we will.Last question that we'll get out of here.The social network part.
Do you think these small social networks can be venture scale businesses to harken back to our venture scale conversation?You know, there, there's this bit of fragmenting in social media, but TBD of how big these, these, these can get.
And we're talking about, you wrote about blue sky that raised a series a just as a.
Yeah, Blue Sky is weird.Like I do think that small social networks like that, they are an option on big networks messing up in identical ways.
And Reddit benefited a lot from, like one of the original reasons, one of the original big influxes of new Reddit users was when DVD, it was a DVD decryption keys got leaked and Digg started banning people for posting them.
And so people just flooded the homepage with variants on these keys and.
reddit allowed people to post the keys and so a lot of people just moved to reddit and so reddit was in retrospect partly a bet that dig would mess something up and then everyone would move there on the other hand reddit has had a bunch of cases where users say they hate reddit and they want to just move on to some other service and reddit's source code a long time ago was open source you can put together something that is you know it's not reddit but it's it's reddit like but none of those have really gotten huge traction there are a few that exist and some of them
have gelled around particular preexisting communities, but they they there hasn't been some existential threat to read it from stuff like that, even though their user base does get really, really ordinary from time to time.
So I think that those when you think about small scale social networks, I think there are two ways to think about it.One is like, is this small scale of like it's doing a similar thing to a big network?It's just not big yet.
And those can be economically viable.Every big social network started out as a small one.Since none of the big ones that are out there now are among the first, we know that the small ones can outgrow the big ones.They can come to dominate them.But
that doesn't have to keep happening forever.
And then if you go a little bit smaller scale, if you say like, this is a network that's small because there is a particular set of people who'd be interested in it, and they largely have some kind of loose real life social network already, and this is just a place for them to convene.
Like those can exist, those can survive for a really long time, but they're not venture scalable, and they really shouldn't work as a private company.They are closer to an actual community than like a platform that hosts communities.
So I think those like when they exist, it's usually a labor of love and it's usually something where whoever is running it is one not making any money and two is still dealing with tedious moderation issues.
But the issues are even more annoying because it's not just these two random people got into a slap fight in the comment section.It's like,
my two of my friends or at least loose acquaintances are fighting and I'm in the middle of this and I have to make a decision and I can ban both of them or like punish one and not the other.And no one's going to be happy with me no matter what I do.
So you have to really, really want the community to exist to to ever bother creating one of those things.And there's not a good way to monetize it.
Even if there is economic activity happening in that network, it's it's not really happening in a way where there's an easy way to capture the value, because since these people already know each other, like
even if you allow people to transact, they will probably just go around whatever the transaction mechanism is if they have to pay extra fees to to transact directly on platform.
So yeah, I don't I don't think it works as venture scale, but I think it's actually a nice compliment to being in venture that if you have this community where you're just expanding your surface area.You will hear more rivers.
You will hear about more interesting ideas.You will get introduced to more people.And so you just get that surface area that you need to know everything that's going on so you can invest in it in an opportune time.
So not venture scale, but a thing venture capitalists should do.
Totally.Well, that's it.That's something I definitely resonate with and try to do myself.Well, we're at time.This has been a fascinating discussion.Hopefully in our next conversation, we will know who the next president is.Who knows?Yeah.Yeah.
Happy election to everybody.Happy election day.
Thanks for listening to the riff.Please go follow and subscribe.Give us five stars and check out Burns.Excellent newsletter.The diff, if you haven't already.
Sign in
