VALUE: After Hours (S08 E35): The Singularity Is Priced In (That’s the Problem)

During their recent episode, Jake Taylor and Tobias Carlisle discussed:

• AI Ran 916,000 Backtests — Here’s What It Discovered

• Why the 60/40 Portfolio Doesn’t Always Protect Investors

• Why the Biggest Market Risk Comes When Investors Feel Safe

• Is AI Already Smarter Than Humans? The Singularity Explained

• Why the Next Stock Market Boom Won’t Look Like the Last

• Jake’s Veggies: Why Doomsday Predictions Keep Getting It Wrong

TRANSCRIPT

[Tobias] We’re live. This is Value After Hours. I’m Tobias Carlisle, joined as always by Jake Taylor.

No special guest today. We had a late withdrawal, but that’s okay. It’s JT and I talking about my new book, His Veggies.

How are you, JT?

[Jake] Whatever else we can come up with to kill an hour. It should be fun.

[Tobias] We were talking about this just before we came on, so let’s keep on talking about this a little bit. I’m doing some research for the next book after this one.

[Jake] It’ll be out next Tuesday.

[Tobias] I looked up the definition of the singularity and it was this Werner Vinge guy and he’s borrowing on the idea of the event horizon in a black hole, that there’s a point beyond which it’s exponentially so much better that you can’t predict the future. Not that we’ve ever been able to predict the future in the past, but you can’t predict the future. Then when we were in China last year, we had this definition of the singularity, which was when you make a machine that can make a more intelligent machine.

At that point, in 1965, there was this short story by this guy by the name of Goode. He was a statistician and a science fiction writer. He came up with the ultra-intelligent machine, which is a machine that makes another machine more intelligent than itself and so on until you get to this singularity.

The singularity is making the ultra-intelligent machine. It looks like right now OpenAI and Anthropic seem to be using machines to make the machines. Are we in the singularity?

Are we crossing the singularity?

[Jake] Yeah, maybe. I don’t know. I feel like if it’s only code, this is one of the problems I think is you get is, you kind of need the feedback loop so that the machine knows when it’s getting better or getting worse at what it’s doing, so then it will keep working on it and optimising.

There’s so much of the world that doesn’t really give you the feedback loop very easily. Coding, it does. Maybe obscure math problems, it might.

It’s obviously going to keep pushing in those directions where it can keep brute force bootstrapping itself upward, but I don’t know if you can get purchase on some problem sets that allow you to keep doing that.

AI Ran 916,000 Backtests — Here’s What It Discovered

[Tobias] I think it’s interesting in the context of backtesting. I was saying to Jake before we came on that I have lots of backtest data and I fed it all into Claude and then I asked Claude to run these various backtests and I gave it access to my backtesting software as well. Where I would run dozens maybe over the course of an afternoon, it ran 916,000 backtests testing 1,217 different parameters.

It’s a data mine.

[Jake] Yeah.

[Tobias] Then I said to it, it’s a data mine.

[Jake] You’ve been data mining with a spoon and now you’ve been given an earth mover. Right.

[Tobias] Exactly right. I have some rules for data mining. One of them is that you data mine in the first half of the set and then you test in the second half of the set.

That’s got its own problems as well, but it’s better than nothing. It’s very hard to find anything that persists from one set to the next. I think that’s kind of interesting, but it optimises something in the first set, it doesn’t work in the second set.

I was looking at it in terms of asset allocation because Ray Dalio has that view that you take uncovered later return streams and you get enough of them and then you spread your bets over those and there’s always something that’s working.

[Jake] Lever the bejesus out of it. That’s right.

[Tobias] And then there’s the Harry Brown who wrote fail safe investing and out of that comes the permanent portfolio. And he says that there are these four states of the world where you have prosperity, inflation, stagflation. Is that one of the states?

Yeah.

[Jake] Deflation with economic difficulties.

[Tobias] Monetary tightness?

[Jake] Yeah, maybe that’s it. Something like that.

[Tobias] And so his sort of match up with Dalio’s views of the world where he’s like growth is positive or negative and inflation positive or negative. And so those bad states of the world are negative growth and positive inflation. And if you take that mix and you can see the assets that they say that you need to combat.

So you need gold for inflation, you need small cap stocks, large cap stocks, global stocks. You can have T-bills, you can have 30 year bills, you can have anything in between. You’ve got this huge mix of assets that you can choose from.

And I just said to the machine, look from 1970 to 2000, I think, or whatever halfway through the data is 1998 or something like that. And tell me what the best mix is. And then have a look at that mix in the second half.

And so something goes from one. So there’s like 1,217 different mixes that you can have. Goes from one to like 900 in the second set.

And the ones that’s number one in the second set is like 800 in the first set. And then the one that fits through both, it’s 400 in the first set and 300 in the second set. So you’re a long way from…

And this is optimising for different things too. I can say optimise for sharp, optimise for absolute return, optimise for lowest drawdown. You can do all of these different…

That’s how you end up with 916,000 different tests because each one’s sort of running through all 1,200, optimising for something different every single time through two sets of data and then over the full set. So that’s why it comes up with like so many different things. And then you can ask it to look in different countries as well.

So there’s like 11 developed market countries that have got data going back that far. There’s no good answer.

[Jake] What does it say? Just smoke started coming out of your computer?

[Tobias] I mean, I can give you the answer that is the least worst of all of the options.

[Jake] Right. It’s got to be really unsatisfying.

[Tobias] Yeah. Yeah, it is. But it’s…

Harry Brown’s intuition… Harry Brown’s just put a quarter into bonds, gold, stocks, and something else. T-bills or something.

Cash, I guess.

[Jake] And that did just fine?

[Tobias] I mean, for just a naive, no idea of what’s going to happen, it does fine through the whole thing. It’s positive. It doesn’t have very many losing decades.

And that’s the other problem that you have is like in 1971, gold comes off the gold standard and then has this like explosion where it revalues to market value. Like is that representative of that decade of the seventies is where a lot of the gold bugs rest there. Case for gold.

And then you’ve got a 40 year falling rates from 1980 to 2020. Is that representative of what bonds are going to do? Where we’ve had this incredible bull market in the US for equities.

Is that representative of what global equities are going to look like? So even if you do all of this backtesting and you come up with an answer, is the answer meaningful?

[Jake] Well, and the sad reality is that the actual end of the things that really matter is still very meagre. It’s a low population sample of true catastrophic market drawdowns. Yeah, of course, there’s been lots of corrections, but actual ones that really, really mattered when you start tipping into like Bernstein’s deep risks, there just haven’t been that many of them.

And so to draw too many conclusions from a small sample size is bad stats, right?

[Tobias] Yeah, for sure. So after all that testing, that was sort of what I wanted to do a book on all of that testing. And that’s not this book, but this book is one of the things that I considered while I was doing it, which was, I looked at how to avoid the crashes for the most part.

And there’s basically the answer is there’s no way to do it satisfactorily. Just let me just give everybody the answer from the outset.

[Jake] All right, well, we’re done.

[Tobias] But I’ll tell you everything that I did consider.

[Jake] Yeah.

[Tobias] So I looked at valuation for one, so CAPE as a cyclically adjusted PE as a measure of CAPE’s got problems with this calculation, particularly in the current climate where earnings have raced so far ahead, like the implications of CAPE are that earnings are much, much lower than they really appear. But then it matches up to like Tobin’s Q and Buffett’s gross market capitalisation at GNP. So the answer’s like not that wrong.

Even the market above its own trend gives you roughly the same answer. Like you don’t need to do all of these sort of heroic valuation tricks to get to the answer that the market’s expensive anyway. And the implications of an expensive market, lower returns and a little bit of volatility in the future.

So then I looked at the 10-3 inversion, which is something that I talked about for a long time, from 22 to 24.

[Jake] Yeah, until it stopped working. Then we stopped talking about it. Well, it seemed to have such a good strike rate.

Yeah, it was like nine for nine, wasn’t it, up until this?

[Tobias] It had a great strike rate. And then it just didn’t work this time. Or the NBR didn’t declare leverage.

Or has it worked yet? Well, it’s getting to the point where you’re really stretching the definition to say… Because we un-inverted last 12 months ago, something like that.

And then, well, we went sideways for a little while. I don’t know. The un-inversion doesn’t seem to work.

[Jake] Well, I mean, do you put an asterisk on it when you have what are we doing? $2 trillion of deficit spending per year plus the AA CapEx boom that just maybe kind of kicked the can 12 months down the road? I don’t know.

Will historians go back and look and say like, well, yeah, that one still counts? Maybe. I don’t know.

I think the important maybe, tell me if this is true of what your findings were, is that none of these are useful much for timing.

[Tobias] Nothing’s useful for timing. Valuation is not a useful timing tool at all. Valuation doesn’t work as a timing tool.

There’s no timing tool. The other one I was going to say is the equity risk premium, which we talk about a little bit, which is the surplus of earnings the 10-year government treasury. And where that’s negative, stocks are expensive.

Where it’s positive, then there’s a margin of stocks over risk-free rate. And it’s not predictive at all. But I like it because when you do a valuation, like you’re pulling in, I pull in the 10-year as part of my valuation.

It’s a discount rate. And it’s a threshold. You should be able to earn more than the 10-year over a reasonably foreseeable period of time.

Otherwise, you should be in the 10-year. That would be a much simpler thing to do. But it doesn’t work.

It’s not predictive at all. The only thing that predicts is the absolute PE or the absolute earnings yield. High earnings yield leads to pretty good returns.

Low earnings yield leads to pretty poor returns across a big data set. It obviously doesn’t apply in every single nine.

[Jake] How much time, on average, do you need to arrive at that conclusion? Is it seven years after, 10 years after?

[Tobias] 10 years. It’s a long period of time. It’s not predictive.

[Jake] I have seen the… It was CAPE ratio, and it showed subsequently one year after, it’s just a scatterplot, just total mess cloud. There’s nothing to it.

And then I forget if it was three or five year, it starts to… The dots start to line up into a trend that looks like, okay, yeah, high CAPE, low five-year forward return. And then you get to, I think, 10 years, and it’s still okay.

I think if I remember right. But then 15-year and 20-year, it’s over again. And the line flattens back out because basically it probably has more to do with the business quality than it did the entry valuation or something.

[Tobias] Well, I thought it was most predictive at 20 years, but that point remains that that’s a very At some point, starting valuation matters less after 20 years hence. So I looked at all of those tools for like… But then there’s different types of crashes too.

So I went through all the crashes since 1929, over the last 100 years. So 29, who really knows that maybe that’s an outlier, kind of like a strange… There’s less sophisticated market in a very different way.

But then there’s… 1987 crash is very rapid. 2000s of valuation question, so expensive.

We’ve never seen that as expensive as that before. Then 2008-9, it’s like a financial crisis. Not necessarily valuation, more financial crisis.

So it wasn’t like it was expensive on a cape rate basis, but it was not as expensive as 2000. And then we’re back to 22, which was 2020, which is like a crash, a genuine crash and recovery, but it was over very quickly. 22 was a much slower grinding bear market.

[Jake] Not even a full year’s worth. So none of these sound like they match each other. It’s all like one off.

[Tobias] They’re all totally different.

[Jake] All totally different.

Why the 60/40 Portfolio Doesn’t Always Protect Investors

[Tobias] And so in 2008, if you had used… So the classic 60-40 portfolio is 40% bonds, 60% equities, with the idea that when the equities go down, there’s this flight to safety, the bonds, the bonds rally, you rebalance out of the bonds into the equities. So you’ve got more money into the equities at the bottom.

It recovers. Didn’t work in 22 because bonds sell off too.

[Jake] Yeah. CLT did worse.

[Tobias] And then if you look at 60, 60-40s only worked for 20 years. There’s a short period of time where bonds and equities were uncorrelated. Most of the time they’ve been pretty correlated.

[Jake] Isn’t it interesting how these rules of thumb, heuristics, bromides come from small sample sizes often. Maybe it was 20 years it was true, and that was enough for a whole generation to say, well, 60-40, or stocks always outperform bonds as a thing. Like, well, yeah, in some narrow circumstances, that’s been true.

[Tobias] You’re always confronting that idea that there’s the past. What worked in the past is the past. And how far back in the past are you going to go?

20 years, that seems like a reasonably, that’s modern history, 20 years. Beyond 20 years, it’s ancient history. Why would that idea apply anymore?

[Jake] That feels wrong to me. That makes me feel old when you say that.

[Tobias] I’m not saying that’s my view of the world. I’m saying that’s a pretty widely held, that’s a reasonable view of the world.

[Jake] It still makes me feel old to hear.

[Tobias] The median age is 28.

[Jake] Of what?

[Tobias] Humans.

[Jake] Right now?

[Tobias] Yeah.

[Jake] Okay.

[Tobias] Might be 38 in a developed world, but I think that’s right.

[Jake] That’s pretty good.

[Tobias] Look that up, Hyde Mind. Tell me if I’m wrong. 20 years, that’s a significant, you’re a child.

20 years ago, if you’re 28, you’re a child, and everybody else is older than that.

[Jake] I was going to ask, do you know what the median age is in the finance industry?

[Tobias] Younger.

[Jake] It’s younger than that?

[Tobias] I don’t know.

[Jake] I don’t know either. But to say that if an entire industry, or a big chunk of the industry has one worldview that was moulded during a particular time, they’ve never really seen the out of sample, have they?

[Tobias] Right.

[Jake] I saw someone the other day say, my biggest worry- Do you want to get rich, or do you want to be a market historian?

[Tobias] Yeah. I think Rudy Havenstein said the other day, my concern is all the 45 year old PMs on desks who’ve never seen a, maybe a, whatever happened 45 years ago, who’ve never been in the market, haven’t been in the market long enough to see a bear market in bonds, maybe.

[Jake] Sure.

[Tobias] Yeah, maybe. Bear market in bonds, that would be 45 years since 1980.

[Jake] Well, but even if you’re 45 on the desk, let’s say- Oh yeah, you weren’t, you were a kid.

[Tobias] Yeah, you need to be, you’re 65, you’re retired.

[Jake] Yeah. No, no, I would say, I mean, under 30 for sure, it’d be hard to imagine that you’ve, you didn’t really live through 2008. And I don’t, I’m not sure I count 2020, it was so fast a recovery.

22 is also kind of relatively mild. Have you really taken the test yet?

[Tobias] Well, that’s one of the reasons that I like to go back through the data and have a look and see what happened, what, how you might react to it in the data. Well, the other, because I’ve tested other ideas in there as well. Like one of the ideas that I tested is the moving averages.

You know, lots of people swear by the 200 day moving average, the 50 day moving average, the crossover between the 50 or the 200. Do you use a simple moving average? Do you use an exponential moving average?

All of these various different things. I didn’t look at all of that in the book. I just looked at moving averages generally.

And the problem that you have is this, it doesn’t matter which one you choose. They all have the same problem. And it doesn’t, two in concert doesn’t help either.

You have this problem where if it’s Japan in 1990, and you’re like, you believe in this moving average and it kept you in right all the way to the peak. And then it plucked you out right at the very top and didn’t let you get back in until the bottom. And so as far as you’re concerned, like if Japan’s been a very mild market to invest in, the problem is that a lot of markets go all the way down to their moving average before they rally again.

So you’ll risk off. And then it depends on how often you’re checking. You’re checking monthly or weekly or daily.

It just creates more opportunities for whipsaws and over-trading and there’s no good solution. And so moving averages are trend strategies. And that’s an asset class all by itself that you can include, but it doesn’t solve all of your problems.

It works sometimes and it doesn’t work other times and it’s expensive.

[Jake] Does it work in the times where your other stuff is unlikely to be working?

[Tobias] It can. Yeah. So it might earn its place as- Diversifier.

As a diverse, as one of the uncorrelated streams that Dahlia looks for, but it doesn’t necessarily solve your problem.

[Jake] Does that, how robust is the data on like CTAs for that? Are there, is it like- I think if anything, it flatters CTAs. Where you get the 1970s create this very outsized perception.

[Tobias] Well, I think it flatters them because there’s so many that, the only ones who survive are the ones who win. The ones who don’t work, don’t survive. There’s a real survivorship bias in those, I think.

But then you have, so then what about hedging, buying options? So there’s lots of different ways you can do that. VIX calls, HYG, high yield debt puts.

[Jake] We don’t need to talk, Toby, about how much money we’ve wasted on VIX calls.

[Tobias] Well, I’ve lost it. I’ve lost it not only in VIX calls, I’ve lost it in HYG puts. And then index puts, like out of the money index puts as well.

And in every single one of those instances, I’ve had the event, I’ve been in the event, had the big event occur, had it be like a third of the portfolio worth of value. This is just in my PA. This is not a public strategy.

[Jake] And not investment advice.

[Tobias] And not investment advice, yeah. And in every single instance, they have expired worthless. And one of them was like, I can’t even remember now, like Christmas 2018, I was there with my dad.

My dad was like, geez, maybe you should take some of that off. I’m like, listen, you don’t understand. This is when it gets exponential.

You should take some off.

[Jake] Yeah. Well, the problem with the VIX calls was always that they were the European style. So they were only money good on the day of expiration.

And the whole rest of the market never moved the price really of the call because they all just assumed it would all die down before the end of the expiration. And they were right. So you ended up never getting, like you’d be money good and then actually not end up cashing any real checks, which is- That’s one of the problems with vol funds, right?

[Tobias] That you’re in the fund and it pays off, but it takes so long to get the results at the end of the next month or midway through the next month. And then it’s how long to take a redemption. Then you rebalance long and the moment’s gone.

[Jake] And you’re like tantalus. You’re just like reaching for the fruit that’s right in your PA already and then you can’t get your hands on it.

[Tobias] Yeah. That’s exactly right. You get to feel good, I guess, while it happens.

You’re not as blown up as it could be, but you don’t get to rebalance.

[Jake] You get to feel very intellectually superior for a week there, right? Even if financially it doesn’t- Yeah.

Why the Biggest Market Risk Comes When Investors Feel Safe

[Tobias] That’s all you get. The psychic benefits of it. That’s it.

So the only sensible solution I think after having… I talk about a Hyman Minsky, the Minsky moment, which is that- Stability breeds instability. So people start feeling good, so they start levering up.

[Jake] Yeah. They feel safe.

[Tobias] They feel safe. So it’s always riskiest when it feels safest. And the corollary to that is it’s always safest when it feels riskiest.

[Jake] Ooh. Put that on a t-shirt.

[Tobias] That’s the headline of my next book. So after doing all of that, I think that the solution is something like you have to have some sort of acid allocation mix, which is like, it’s such a pedestrian response, but I think that it’s the only… I’ve basically eliminated everything else.

[Jake] So you went in a giant academic research circle to arrive at like, oh, this should look pretty boring and sensible, basically. Just try to avoid the first and bottom quartile and you end up okay?

[Tobias] I did discuss a little bit of the psychology of it in there, because I think that that’s important.

[Jake] I think that’s really important.

[Tobias] I’ve been through enough blowups to know how, particularly to… I started working in 2000 at the peak, and so I saw a blowup, but I was kind of insulated from it a little bit, because I just had salaried and didn’t have much or any savings at that point. And then 2008, 9, I felt a very…

It was like a chunk of years of working and saving that was in the market that got blown up, and that was a pretty sweaty. And there were people suggesting that that might be the end of the financial system at that point in time.

[Jake] Maybe the world, the global order on a larger base.

[Tobias] It felt pretty real. I don’t know if that was the sort of stuff you’d see. Did you go get money out of the ATM?

It felt that way in 2020. Do you think that felt that way in 2020? Like, oh, it could be a pandemic, could be the end of the world.

[Jake] Well, I’m sure we could go back and review the tapes of what we were talking about.

[Tobias] I was sweaty in some of them.

[Jake] Yeah. I mean, I had genuine biological concerns about… I thought like, God, what if my kids see bodies piled up in the streets, because we can’t keep up with the…

[Tobias] Well, I said that was what was happening in China. That was a real possibility.

[Jake] I know. And I was like, oh, gosh, that sounds horrific. But obviously, we didn’t, thankfully, get to that.

[Tobias] Right. And if you have a 1918 flu pandemic, then it’s still possible that that happens. And it’s happened lots of times.

It’s killed off large chunks of humanity. So that was an existential risk. So there was end of the financial system, end of humanity.

The next one, I don’t know. It’s probably…

[Jake] Oh, now we got AI, right?

[Tobias] AI. Yeah. End of humanity again.

It’s going to feel pretty real in the moment. I think you have to proceed as if it’s not the end of the world. And then if it ends up being the end of the world, then you have to adapt at that point.

But assume that it’s not the end of the world. What if it’s not the end of the world?

[Jake] You don’t know this, but this is a perfect lead in for veggies.

[Tobias] Awesome.

[Jake] Yeah.

[Tobias] Well, it’s just about time. Do you want to do it?

Jake’s Veggies: Why Doomsday Predictions Keep Getting It Wrong

[Jake] Should we just do it? All right. Yeah, let’s do it.

All right. Toby, I need to tell you something, and I’ll need you to stay calm. The world is going to end in 24 days.

And this is just not my opinion. It’s science said so. The Journal Science, with a capital S, that is.

And on November 4th, 1960, a physicist named Heinz von Forrester published a paper in that preeminent journal. And what it said, the title was Doomsday, Friday, 13th, November, 2026 AD. He took every estimate of world population from the year 1 to 1958, and he fit it to this mathematical curve.

And the math spat out that date, Friday, the 13th of November, 2026. And he told Time Magazine, and I’m quoting here, our great, great grandchildren will not starve to death. They will be squeezed to death.

And this guy wasn’t some crank. And for 30 years, actually, his forecast looked prescient. Reality was running ahead of the math of the curve.

And in 1987, there was a follow up in science that was titled World Population Still Ahead of Schedule. So then, around 1993, the real line just fell off the curve, and it never came close again. And the equation was saying that Earth would have about 76 billion people by then, hence the whole squeeze to death thing.

And instead, we’re more at like 8 billion. And so von Forrester was actually in good company of doomsday sayers. Paul Ehrlich famously in 1968 said that the battle to feed all of humanity is over.

He predicted hundreds of millions would starve in the 1970s. He would take even money that England would not exist in the year 2000. Then he bet Julian Simon famously that five metals would get more expensive over the 1980s.

And of course, all five got cheaper. And both of these scientists that we’re talking about, they took a real trend and they drew a line through it, and then they painted this dire picture from it. But they forgot that the line was also made of people, and people who could see that line as well.

And Ehrlich didn’t model breakthroughs like Borlaug’s wheat already in the ground in India when his book came out. And that process hasn’t really stopped. And in fact, Matt Ridley, who wrote a fantastic book called The Rational Optimist that I highly recommend, he highlighted an example just last month.

British scientists have gene edited tomatoes so that they accumulate a precursor that light can turn into vitamin D. And apparently it’s easier to reprogram a tomato than it is to improve the British weather, I guess. But human trials are testing whether eating them will actually improve people’s vitamin D levels, which is actually very important, especially for darker skinned people living away from the equator.

They have a hard time getting enough vitamin D. So basically, somebody identifies a problem, somebody tries to answer it, and then we test whether it works. And of course, more mouths to feed to get to 8 billion has been part of the equation.

But it’s also more minds that are going to be working on problems, especially when they can easily share what they’re learning. So why should we care? Every bare thesis is a doomsday equation of some sort.

And we’re, I think Toby, you and I are sometimes as guilty as anyone of pointing out troubling trends, just maybe dismissing others exuberance is foolish. And the audience might think that I’m a bearish curmudgeon, and I am sometimes, that’s fair. But I also want to make a full throated case right now that rational optimism for the long term for humanity is the logical stance and not just vibes based, let’s bring actual tangible progress.

So I’m going to give you three reasons. And maybe you can come up with even better ones. But the first one, humans bend curves.

Von Forrester was worried about birth rates. And basically, nobody ordered them lower except for the Chinese there for a while. And that’s messier than we have time to get into.

But global fertility went from 4.9 children per woman in 1950 to roughly 2.25 today. Now, TC, I know you’re going to say, well, the new doomsday is depopulation, right? Korea, Japan, Italy, more adult diapers than baby diapers is the common quip.

But fine, I’ll admit that maybe that’s true. But notice that the doom sign flipped inside of one lifetime. So any kind of system where the dominant fear can invert in 60 years is not probably something you should model with only one curve in either direction, I would say.

Number two, when the warnings get real enough, we do tend to act on them finally. So do you remember the depleting ozone layer when we were kids? Like nobody under 30 does because that’s what fixed looks like.

Two chemists in 1974 said that CFCs were eating the ozone layer and they were right. But 13 years later, we had the Montreal Protocol. And today, about 99% of those chemicals are no longer used and the ozone is on track to recover.

Same thing with acid rain, Y2K, these scary headlines that end up being solved problems. And number three, which I think is the most interesting, most of the team hasn’t even taken the field yet. For all of history, the number of people who could work on hard problems was relatively small.

You needed to be in the right place, maybe with the right father, near the right library. There’s a lot of coincidence that was required. In 1913, as a great example, a clerk at the Madras Port named Ramanujan mailed nine pages of theorems to a Cambridge professor he’d never met.

This professor hardly then read them and he couldn’t believe his eyes. Like there were these mathematical breakthroughs that just poured off the page. That was the whole system back then, this lone Indian genius with a stamp and an Englishman who happened to open the envelope and be smart enough to recognise that genius.

Well, how many Ramanujans never found the stamp or just toiled on a rice paddy for their whole lives? In 1820, one adult in 10 could read. Today, it’s more like nine in 10.

The pool of people who can even enter the conversation to solve problems went from basically a rounding error to maybe now most of the species. And in the last 15 years, close to a billion people in India came online, a billion in China. The median African is 19 years old and nearly every one of them carries a tutor in their pocket that doesn’t sleep, doesn’t really bill them, doesn’t care who their father was.

And you can learn almost anything nowadays and it’ll keep getting better from here, I think. So even with the promise of AI and robots on the horizon, the singularity, I think maybe don’t count out the human brain just yet. We have tonnes of human capital that’s about to get to work solving our biggest problems.

As a funny aside, Toby, of course, you remember the matrix and you remember how the machines had enslaved the humans and put them in those pods of goo and they were using them as batteries basically. Well, apparently in the original script, the machines were actually using the humans for both energy and compute. They were harnessing the human brain for compute.

And it’s interesting to think about that now as we’re racing towards building all these data centres and we talk about the supply of compute being so constrained. So anyway, indulge me, picture it’s 2056, it’s 30 years from now. A 16-year-old in Kano, Nigeria, studies a rare disease that’s affecting her family and she has access to scientific papers and AI tutor, a digital lab that she can use remotely and her work actually helps produce a treatment for it.

And we may be on the verge of solving all kinds of these sort of problems that we didn’t really have enough resources to get to yet. Now, I don’t know whether that particular story happens, but I do think that we’ll get discoveries from people who previously never had a chance to contribute on a big scale. And that makes me very optimistic.

And there’s a lot of human talent out there that really hasn’t even been given a fair shot yet. So I’m going to go on the record, today’s October 20th. I’m going to boldly predict that the world will not end on November 13th due to 76 billion people crushing each other.

And if I’m wrong, then I’ll be happy to issue a correction in a December episode.

[Tobias] Good one, JT.

[Jake] Thanks.

[Tobias] I forgot to give a shout out to everybody, but then I’ll come back. Breckenridge, Boise, what’s up? Toronto, Courtrick, Belgium, that was a good one.

Bendigo, Victoria, Tallahassee, Toronto, Pettit, Tigfar, Israel, Snohomish, Scotland, Lausanne, Switzerland, Tombal, Texas, Shetland Islands, good one. One we haven’t had before, I think. Serbaton, Riyadh, Saudi Arabia, Jupiter, Florida.

I think I got everybody. Sorry if I missed you. Bellevue, I think I got it.

I’m with you. I’m hugely optimistic. We sort of started what we were talking about at the start.

I don’t know if we got to the definition of the singularity is when the ultra intelligent machine makes it a more intelligent machine. And at the singularity, we’re already seeing like it’s solving math problems. It’s decoding the human genome.

So that’s going to be incredible for medicine and for ageing maybe.

[Jake] Yeah. Gosh, and we need it. Like we spent in the US 20% of GDP on trying to keep everyone healthy.

[Tobias] Just imagine if you could- With bad outcomes, with not great outcomes.

[Jake] I don’t know if that’s entirely true. I think we have like reasonably good outcomes, but not on a kind of per dollar spent basis.

[Tobias] But on the age, what do you call it? The total life expectancy, yes.

[Jake] Yeah, I think we had, I think the opioid epidemic really like put a big, like sort of weird- It’s still one of the biggest killers.

[Tobias] It’s still the biggest killer by a mile. I saw a statistic the other day on Twitter. Yeah, it’s much, much bigger than everything else is vanishing relative to the opioid epidemic.

I don’t know what that means.

[Jake] I think those things might be addictive.

[Tobias] That’s probably what it means.

[Jake] That’s what it means.

[Tobias] Sorry, did I interrupt you? Did I keep going?

[Jake] No, no, I was just saying that we imagine if you could carve out 10% of US GDP out of that 20 and put it back towards things you actually wanted, like, I don’t know, electronics or vacations or, I don’t know, better schools or a million other things if we could get a better handle on our expenses. There are other countries that get roughly equivalent healthcare outcomes that spend like half as much as we do. And the whole system, I’m not super optimistic that we’re gonna be able to solve it from the government doing a better job with it.

There’s just too many entrenched interests and too many really big, powerful lobbies. But from the technology side, if we can just chip away at it and be like, well, this is the new standard of care because it’s so much better and it’s cheaper to administer and personalised medicine can probably help a lot there, then I think that’s kind of our only really hope of fixing the tapeworm as Buffett’s called it.

[Tobias] Did you read Fermat’s Last Theorem?

[Jake] I did.

[Tobias] I read that, I don’t know how long ago, 20 years ago or something like that. And as a non-math person, I was sort of surprised at the lengths you had to go to make a proof, like pulling in those unusual conjectures to sort of demonstrate the idea. But they’ve recently made all of these breakthroughs in math because of anthropic or open air.

One of them has done all of this work on those math problems and mathematicians are upset. And I saw somebody saying that the solutions were alien in the sense that these are not connections that a human would have made.

[Jake] Oh, really?

[Tobias] Which I found really, I found that an odd thing. I don’t know, that may be true. I don’t have enough of an insight into whether that’s true or not, but I found it odd because when I read Fermat’s theory, the connections that the mathematicians were making there to make those solutions also seemed alien to me from the outside.

But I wonder if a human had made that connection, would you say that’s an alien connection? Or would you say that that’s a brilliant insight to have, he’s across these two different regions of math and he’s able to combine them together to come up with this insight?

[Jake] You know, that’s a good point. I think that I kind of dropped from the original write-up that I had for the veggies today, which is that Matt Ridley also has this idea that innovation is when ideas have sex. So basically you take two kind of disparate ideas and you put them together and all of a sudden, like something cool gets unlocked.

And that combinatorial power, you know, if you think about AI as the, you know, even you running a thousand back tests that you previously would have had done by hand, well, just imagine the random generator of like throwing AI at problems where it takes two things that seems like have never been put together before. And like, what if it’s all just, what if there’s like a million suitcases with wheels that are just waiting to be solutions put forth for us, ready to go? I think that’d be really cool.

And I, you know, hopefully we can make some real progress with that.

[Tobias] Well, by extension, the person who was making the point that the math solution was alien because these connections are so strange to put them together. Like they extended that then to like, what if you make these alien connections in other? So that’s exactly what you’re talking about now.

[Jake] Yeah. That’d be awesome.

Is AI Already Smarter Than Humans? The Singularity Explained

[Tobias] We must, we must be. So the singularity is where, you know, the Werner Vinge definition was that everything, everything technology increases at such a rapid pace that you just, you lose sight of where you are. Everything happens instantaneously from your perspective.

And you can’t predict then beyond the singularity. Not that we’ve ever been able to predict. Like that’s the, like we’ve been able to predict what happens in the future.

[Jake] Yeah. Like you can see past, you know, two weeks from now, with any kind of, come on.

[Tobias] But it seems to be that, it seems to me like we, open AI and Anthropic are making the machines that make the machine smarter. Like they’ve either, they’re either doing that now or that’s, they’re going to be doing that this year. That’s the definition of ultra intelligence.

That’s the singularity. Like we’re living in the singularity right now. And that just means, rather than all of these individuals.

[Jake] I have to still take my garbage can down to the road? Like what happens?

[Tobias] Well, that’s the funny thing, right? When we were in China last year.

[Jake] Yeah.

[Tobias] Like the, aside from like maybe the Star Trek machine that can beam you up and down. Like the things that you would say that you’ve seen in the future, flying cars or autonomous cars. Oh, I mean, probably not autonomous cars, probably flying cars is what everybody wanted.

But flying cars, humanoid robots and- Got it now. And artificial intelligence. And like, that was what they were working on.

We’re at that point in time where like the, what was science fiction is now just, we are really going to get that done. The autonomous cars, one way or another, it’s close-ish. Whether it’s Musk saying it’s this decade or whether it’s this super intelligent machine figures it out.

I mean, Musk saying it’s five years ago, 10 years ago. Clearly it’s still 10-ish years away for Tesla. Although I think that, do you think Waymo’s got there?

[Jake] Well, I think- Have you ridden in a Waymo? Yeah, I have. I mean, of course they’re, they focused on tighter problem set, like a single city that’s fully mapped out.

Less edge cases, I guess is the right way to think about it. But it seems like statistically they’re doing well as far as like accidents per mile driven, that kind of stuff.

[Tobias] Once it’s lower than a human.

[Jake] Better than a human, yeah. I mean, I don’t know. Do you feel safe?

Would you put your kids in one and let them go drive around without you?

[Tobias] Yeah. I think humans make lots of mistakes behind the wheel. So I think you just go for a drive and you see humans making mistakes behind the wheel.

[Jake] I think they’re getting worse, actually.

[Tobias] Well, everybody’s on their phone. Everybody’s on their phone all the time. It’s amazing.

Most cars seem to be on their phone. Particularly if somebody comes to a stop at a light, they pull out their phone.

[Jake] Oh gosh. Yeah, we need like Faraday cages or something in the car to keep- We need autonomous driving.

[Tobias] We need that soon. But the things that the hard tech, the high tech Chinese VCs were investing in, this sort of stuff that’s way out in the future was just stuff that was like, well, it’s a humanoid robot. I’ve seen lots of YouTube videos on hands and they’ve got Robot Olympics where they run now.

There’s a robot fighting league where they fight each other. It’s getting close. They’re not in the home yet and who knows how much one of these things actually costs, but they’re close.

[Jake] It is interesting how quickly we sort of take what felt like magic and then it becomes just like your birthright to have it. Internet on an aeroplane, right? To be able to have data streaming back and forth to something that’s 30,000 feet in the air and moving 600 miles an hour and you can, whatever, stream Netflix with it.

That seems like a miracle to me. Now, if it’s down for five minutes, you’re ringing your help button to be like, what’s going on here? How come the internet’s not working?

This is bullshit.

[Tobias] I used to like flying because you didn’t have to work. There’s this period of time where you’re disconnected from the internet.

[Jake] Yeah, I’ve heard authors, some guys would just book cross-country flights to get some work done. They would just write on the aeroplane because they were fully isolated from the noise.

[Tobias] But I keep on thinking that the singularity is here. Like, I don’t know what that means.

[Jake] Well, even if it was, like what if it sort of just was like, yeah, whatever, that’s fine. Like these things are pretty good at doing what they do.

[Tobias] If it takes away the admin, like if one can do my taxes for me.

[Jake] You’re in, you’ll fault cosine.

[Tobias] Take the tax, take the tax plan away, take the admin away, take the compliance away.

[Jake] Does that make these two big frontier labs, two, $3 trillion companies?

[Tobias] I don’t think so. Because every time I see somebody with the bleeding edge, the state-of-the-art, I guess people are going to need those state-of-the-art machines. But it seems to me that we’re already at the point where older models are quite good at doing a lot of stuff that people are going to need.

And you basically get them for free.

[Jake] Yeah. That’s one of the two- Just got to give your data the zuck basically to- But that’s the challenge, right?

[Tobias] It’s finding the, like, where is the bottleneck? Is it going to be power? Is it going to be computing power?

Is it data centre space? You’ve got to think that all of those problems are like solvable problems that they’re not that difficult to solve. They just take a little bit of time.

[Jake] Maybe it’s computing space. Might it be trust?

[Tobias] Well, that’s the natural, then you wouldn’t trust any, like, you’re going to get a telephone call from someone pretending to be someone who you know, and it’ll be believable. The AI is close.

[Jake] Yeah. Well, I would meant more like, let’s say you were going to let AI totally run your life.

[Tobias] I see, yeah.

[Jake] Who do you trust in the tech world to have absolutely all of your data to just go- There’s got to be a way to silo that off, right?

[Tobias] You silo the parts that are important off, air gap it.

[Jake] Sure, buddy. You don’t think that all these tech companies have like profiles of you already built from every interaction.

[Tobias] Ray Perk says, have you seen the video game mashups that Opus 5.5 has enabled via binary decompilation to source? I’ve seen the games that they’ve made. I don’t know if it’s via that method.

Binary decompilation to source. You make me say that twice, because you know what it was.

[Jake] I think that they’ve been able to take software- You’re just taking like, you’re showing video games that are like out of context.

[Tobias] I think they can just, I think they can take the, I think that the AI can take the binary, the output, and then turn it back into the underlying software. I think that happened over the weekend or the week before with basically all software.

[Jake] But with like, you know, some other mod on it automatically.

[Tobias] Or you can just put Word, Microsoft Word in, and it’ll give you a functioning version of Word built backwards from what it can see.

[Jake] But I’ve already got Word.

[Tobias] Then what are you going to pay for that software? If somebody else can go backwards.

[Jake] So distillation basically of your software.

[Tobias] All software is cracked. Is that- Well, I think it’s consumers. It’s all consumer surplus.

Who makes the money in this world? If you’ve got a robot that can do all the physical labour, and you’ve got this ultra intelligent machine that it’s smarter than you are, it doesn’t make mistakes. It works 24 hours a day.

There’s 10,000 of them that it can access at any point in time, a million of them that it can access. What’s the human day?

[Jake] What are we doing?

[Tobias] Not physical labour, not the thinking labour. What else do we do?

[Jake] I don’t know.

[Tobias] Make more humans. That’s the only thing we can do.

[Jake] Are we going to be artisanal investors?

[Tobias] Maybe we don’t have to work anymore.

[Jake] Maybe. I mean, that’s what’s been promised at lots of other- Halmarx was right.

[Tobias] The robot revolution is here.

[Jake] I was thinking more of the nuclear power in the 1950s where it’d be too cheap to metre.

[Tobias] Too cheap to metre. So I looked that quote up because I wanted to include that and I’ve included that in the opening of my book. What is interesting is that the guy said that government departments shouldn’t make these prognostications about the future.

And he said, but in the future, nuclear energy means that energy will be too cheap to metre. And then that hasn’t happened. Because we haven’t got nuclear, it’s too expensive to build nuclear.

[Jake] Is that what happened?

[Tobias] What do you think happened? I haven’t actually- I’ve just got a blank space there in the book that I’m writing right now where I’m trying to figure out what happened.

[Jake] I haven’t looked it up. TBD. I don’t know what did happen.

I mean, certainly there was a lot of regulation in energy, probably for good reason because it lends itself to monopolistic characteristics. Like we can’t have 10 different lines maybe going into your house for different carriers. And there’s certainly economies of scale where you would think like, yeah, you want to have one big utility servicing a location.

[Tobias] Every home gets a personal nuclear reactor.

[Jake] Well, maybe. I mean, distributed generation has always been like a thing that people wanted and thought was going to be- Stick it in the substation.

[Tobias] So it’s like, it doesn’t have to be at your house if you’re worried about that. Or you can have it off grid like people have generators have a little nuclear reactor, little modular nuclear reactor.

[Jake] Maybe.

[Tobias] And that powers your home AI.

[Jake] Sure. Local model. Now we don’t have to trust these maniacs with our data.

[Tobias] You’ve got a solar panel on the roof.

[Jake] As well as- If I’ve got a nuclear, I’ve got the sun in a box in my garage. I don’t need the solar panel, Toby.

[Tobias] Charge your car from the nuclear reactor.

[Jake] Sure. Why not?

[Tobias] Internet comes in via satellite. So maybe all of the, I like the idea of compute in space, like AI in space. So all that comes down to earth is the intelligence, like the output is what comes down.

[Jake] Sure. Yeah. You have a dumb terminal.

A little like, maybe we’ve just become very, very local. And like your food is grown locally. Your water supply is provided locally because energy unlocks all those things relatively easily, right?

That’s the biggest prime mover is energy. If you have actually abundant energy, you can have a lot of other stuff.

[Tobias] So energy is the big bottleneck.

[Jake] Probably. Probably always has been throughout humanity’s history.

[Tobias] Household nuclear reactor. A human being, the entire energy consumption of a human being over the entire course of a life is the size of a Coke can of nuclear waste.

[Jake] Is that right?

[Tobias] Yeah.

[Jake] Wow.

[Tobias] And think about how much- That’s what big nuclear wants you to believe. I don’t know. That’s a statistic that I haven’t made.

[Jake] Camco statistic. Well, I mean, think about, gosh, how much wood did you have to chop down to make an ancient energy living before? I mean, that was hard work.

[Tobias] Is there enough nuclear material in the world for everybody to have a Coke can of material?

[Jake] I don’t know. There must be. I don’t know.

Can we grab some out of space? Is that the- Is there a nuclear meteor?

[Tobias] That’s what SpaceX is going to lasso, the meteor as it comes past. One of the things that I was, in the book that I was, the crash test dummy book, because I did say it’s crash in the first instance, but then it’s being, you have to be able to rebalance for the boom. That’s the second part of the, it’s the third part of the book.

And so I wanted to make that clear that it’s not just a crash book. There is this boom part to it. And I fully think that there have been so many crashes in history.

There’s a crash about once every seven years. We’re highly likely to get another crash. It’s got nothing to do with the market being overvalued or AI.

You don’t have to identify the reason for why there’s going to be a crash. It’s just a feature of markets that they crash. So you don’t have to bend over backwards to try and figure out the source of it.

Why the Next Stock Market Boom Won’t Look Like the Last

They just tend to happen. But then there’s also a boom that comes after that. And so the worst thing that you can do is blow yourself up in a crash.

You just want to make sure you’re positioned for what comes next. And this is the final point that I just wanted to make about the book. The thing that worked in a preceding boom is rarely the thing that works in the next boom.

So in the seventies, it was gold. Gold went sideways. In the eighties, it was private equity, leveraged buyouts.

Nineties was tech. Japan, yes. Emerging markets.

Emerging? Japan? Maybe it’s developed.

[Jake] Nah, developed.

[Tobias] And then the 2000s, probably tech, maybe private equity again. EM.

[Jake] Big EM.

[Tobias] 2010s, tech again. 2020s, still tech. Meme stocks and now AI.

Everybody’s still got their board, I got club NFTs.

[Jake] Ooh. Yeah. How are those doing?

[Tobias] Not cheap enough to buy yet. They’re holding their value. Uncorrelated.

[Jake] Yeah. Oh boy. I don’t know.

Not so sure about that one.

[Tobias] So folks, I got all of my, I said to Jake, I’ve got all of my crash anxiety into this one book. And I looked at every single thing that I’ve ever looked at from like all the way back to greenbacked, looking at valuation, moving averages, instability, hedging, different asset classes in different countries and what has worked and what hasn’t worked. And I put it all into one book and that was the, that’s Crash Test Dummy.

So if you’re interested in what causes crashes, what happens when crashes happen, what the other side of the crash looks like, then take a look at Crash Test Dummy and let me know what you think about it.

[Jake] We’re going to have to send a copy to Russell Napier at the Library of Mistakes in Edinburgh. That’s where it’s, it needs to have a copy there for sure.

[Tobias] I’ll personally deliver it.

[Jake] You should. It’s a good visit.

[Tobias] Yeah. So I’m, and then I want to do, I want to do one more book too. I want to do something on the singularity because I get this feeling that the singularity is like, by the definition that we’ve discussed today, the singularity is close if it’s not already here.

And this is the point that every single science fiction writer since like 1965 has pointed to as being this like momentous event beyond which humanity’s irrevocably changed. And it seems to me that it’s like, it’s happening now, it’s happening this year, it’s happening very soon.

[Jake] What do you think of the, it does, it feels like to me, some of the, there’s like a bit of religious overtones that kind of creep into some of these discussions. You know, like a rapture and maybe even like, I mean, I’ve heard Peter Thiel talking about like antichrist as a thing right now. I don’t, do you have any thoughts on that?

Or is that still a big blank spot in this chapter?

[Tobias] I haven’t, I don’t think that’s going to make it in the book. I don’t think that’s, might be beyond my pay grade.

[Jake] Yeah. Okay. Fair enough.

[Tobias] I think that politics and religion, generally, I try to avoid them. Um, for the most part, publicly.

[Jake] Yeah.

[Tobias] Because it’s just too, the upside downside skew is bad. Yeah. And it’s, there’s, it’s partially, it’s not rational.

So it’s not something that you can really bring to bear reason to sort of pick the outcome or determine what’s going to happen. Like it’s, it’s not that anything is predictable, but it’s entirely unpredictable. I don’t know if that’s like a different type of predictability, but it’s just not predictable at all.

[Jake] 10 people will be like, love to hear what you said. And then a thousand people will send you hate mail.

[Tobias] Right. Or even if it’s 10 to one, 10 to one’s the ratio. I don’t like the one.

[Jake] Still bad.

[Tobias] It’s equally not a good ratio.

[Jake] You, so you don’t want you, you like 10 people agree with you, but if one person sends hate mail, that cancels out 10.

[Tobias] Well, I wouldn’t even see the 10. They just like nodding behind the screen. And then one person’s like.

[Jake] Idiot. Yeah.

[Tobias] Yeah.

[Jake] That’s fair.

[Tobias] What do you think?

[Jake] I mean, that’s also.

[Tobias] Yeah.

[Jake] Yeah. Above my pay grades. I think anytime.

We have. A really hard time looking past two weeks into the future. We look for other explanations that maybe go outside of.

You know, normal explanation, whether that ends up being mythology or. You know, we would. You know, crop failures.

We would say it was because, you know, Apollo was mad at us or, you know, you didn’t do the right sacrifice. And I think we’ve slowly been chipping away at those for most of our history.

[Tobias] We’ve just got a scientific explanation, a scientific sounding explanation for exactly the same, like exactly the same feeling.

[Jake] But I think we’re, we’re, we’re wired to create stories out of things. And we’ll match, we’ll force fit whatever data we can find or facts we can find into a narrative. And, you know, that, that lends itself for good reason.

Like we, you know, our evolution was such that we. I’m like telling each other stories around the campfire helped us survive. And so stories feel good.

They, they make sense for us. And then they, that, that conveyed some, you know, survival advantage. When some story helped you survive that, you know, we were wired for story.

And so we shouldn’t be surprised then when big things that are happening around us. We look to attach stories to them.

[Tobias] In the science fiction book by Good, where he talked about ultra-intelligent machines building the next more intelligent machine. He said that it would be the last tool that humanity builds. So we’ve been good at building tools.

We’ve got to this point where we’re no longer gonna be needed to build tools. But I still think ultimately the machines serve our wants and needs and desires. And so as you sort of come up with new, they’re a limitless run.

What do you want? There’s a limitless, limitless desire, pneumatic desire. Somebody gets something you’re going to want it to.

[Jake] And so we’re going to be powering the- Is it a future of there’s no more rivalrous goods anymore?

[Tobias] Possible.

[Jake] That can’t be true though, because geography creates like- Abundance, right?

[Tobias] Super abundance. That’s the claim. Yeah, but if- That’s what big, that’s what big AI would have you believe.

[Jake] Yeah, but if I’m going to a sports game or a concert in person- That’s true. There’s only so many spots that you can sit. I guess you can go into like 3D.

[Tobias] Yeah, they look pretty good, those bars where you’re like right up close to the screen.

[Jake] Yeah, maybe. But I think you’ll still, like the human will, might still want to have actual real experiences in the real world with other humans. I don’t think that necessarily goes away.

[Tobias] I mean, let’s hope.

[Jake] Let’s hope.

[Tobias] That’s it, folks.

[Jake] I didn’t think we were going to make it, but- Neither did I, but we did it. We stretched it out with enough bologna to make it for an hour.

[Tobias] Any final words?

[Jake] No, other than just pick up Toby’s book, give it a read. Let us know what you think.

[Tobias] And yeah, if you love it, leave me a five-star review. If you don’t like it, send me an email and tell me why. I’d really appreciate it.

[Jake] Send a quiet email.

[Tobias] It’s hard to get reviews. And reviews are highly prized by the Amazon algorithm. So even if you just get to the end and click the five stars, that’s very helpful.

If you feel compelled to write something, then that’s even better.

[Jake] So- Don’t even read it. Just click five stars.

[Tobias] I do read the reviews. There’s one review up there. Thanks, Jeff Horn, whoever you are.

That was a nice review.

[Jake] Oh, what a guy.

[Tobias] I appreciate it. And I didn’t push it until I had some reviews up. So now that the reviews are up, I’m letting people know.

And I appreciate them. Thanks, guys. We’ll see you next week.

We’ll have a guest next week. See everybody then. See ya.