Wait is this real? Did a random 25yo get 45B under management because he got some interviews after getting fired from OpenAI? I missed all of this and it almost seems like performance art to me.
Yes, it’s real, and the failure is much more about risk management than about having wrong fundamentals.
His fundamental thesis is as an AI maximalist, with expertise on who will be the winners.
His fund completely blew up, even though his fundamental thesis still may be correct.
He made a ton of money early, but then he just kept increasing his bet size via leverage. The problem is that these stocks may be going up, but they are EXTREMELY volatile. When the whole market went down with some of the Iran news, his margin levels dipped too low, and his lenders issued margin calls. He had to sell his stakes at losses, and his fund collapsed.
Now, part of the problem is he was not a legacy hedge fund with great relationships with his banks, because if he was, he might have been able to postpone the margin calls. Had he been able to do that, he would have been ok, because his positions actually recovered within a few weeks.
It didn’t matter, though, because he was forced to sell at the low point.
In other words, even if your long term bet is correct, if you are maxing out your margin to max out your bet, you are one volatile dip away from blowing up.
No matter how good the bet, you have to be able to ride out the volatility to survive.
It’s like if someone offered you a bet on a coin flip, where they will pay 5x your bet if you win. How much should you bet on each flip?
Well, by expected value you should bet everything you have, because your EV is 2.5x, so the more you bet the higher your expected value.
However, if you bet it all, you have a 50% chance of losing all your money and not being able to make any more bets.
He got 45B under management because he published his investment thesis and a lot of tech bros agreed with it. They thought he had a novel investment idea and gave him money to enact it.
Everyone has some sort of category error to an extent. But more than that, the part about safety guardrails really resonates with my experience. I inevitably have to go down to the low-level to inspect memory, but because of safety issues, I can barely use Codex as it keeps prompting me for 'trusted access.' AI is actually really good at catching bugs, so I'd love to use it, but it does this every single time...
> It’s much easier to say someone else’s job is going to be fully replaceable by AI when you don’t actually know what they do.
Too true. This isn't limited to AI, either. The most obvious example in my lifetime was during peak blockchain hype, when people who had never worked in finance convinced themselves that blockchain was going to act as the backbone for how money gets moved around. As if the problem that needing solving was Bank of America doesn't trust Capital One to update a number in their database.
The nice thing about AI, at least, is I can always push back and tell people, "Sure, we can do this with AI. I just need you to use Claude or ChatGPT manually to prototype how it would work." This normally results in the requestor realizing that there's human judgment calls involved in the inputs, process, or outputs that require meatbag intelligence.
Every time I've done that, the people were so ignorant that they still didn't recognize that expert judgment was required... they just had the LLM make some slop.
Yeah, ive seen the same thing in construction and agriculture. People take the most strikingly visible part of the job, try to automate it, then wonder why nobody wants to spend $1,000K buying the automation for the easiest and simplest and solved parts of the job.
"You don't got to frame walls anymore!", something that is the most enjoyable part of residential construction, takes like a day of work at most, and can be done while on autopilot. "We can instantly generate wall plans for stud location from top-down blueprints!", oh that thing people do in their head because they just space it all on 16 or 24 inch centers and it doesn't actually matter which side you start on? "We can wirelessly drive your tractors!", the easiest task on the farm which all the weeks of maintenance and planning and setup led up to?
Now don't get me wrong, there are many things that can be improved in those industries certainly, but you aren't going to know what those things are until you actually go on the job site and learn how they currently do it daily. Most all of the low hanging fruits have already been picked one way or another and a fancy step stool is not going to be industry changing.
I am trying to fathom a more catastrophic failure in communications than the lab CEO's using this narrative. The reality ends up being closer to Radiology jobs [1], but we're never able to fully realize this because the press is running around like headless chickens and unloading the fear on a public that then reaches for any leverage they can get (see: Datacenter NIMBY-ism).
Any short term FOMO valuation bumps they got from this narrative has to now be liquidated into paying premium for server racks in a warehouse.
Reminder folks: There is no evidence that AI is not normal technology [2]
> As if the problem that needing solving was Bank of America doesn't trust Capital One to update a number in their database.
As I understood it, the attraction of blockchain was more that the money could get moved around without Bank of America and Capital One knowing about it, and therefore without those financial institutions reporting the transactions to the government as they are required to by law. Which of course screams "money laundering" and "tax evasion" to the government. Yes, there are possible legitimate reasons for wanting to be able to move cash around anonymously and invisibly, but they're basically rounding error compared to the ones mentioned above.
In other words, blockchain on this view was a classic example of the old saying that you can't apply a technical solution to a social (or in this case political) problem.
> Climbing Mount Everest and getting a PhD are both hard. I wouldn’t say that doing one means you can do the other “no problem.”
and here:
> Surgeons and electrical engineers may be extremely smart and accomplished, but that doesn’t mean they have the ability to weigh in on climate science.
Feels tangentially like a version of Nobel disease[0], discussed here[1] a couple of weeks ago.
> It’s much easier to say someone else’s job is going to be fully replaceable by AI when you don’t actually know what they do.
This exactly. As someone who once lived in Silicon Valley and then ended up interviewing 700+ CxO's of profitable (or nearly profitable) SaaS businesses in various domain specific fields, it amazes me how many prominent "experts" in SV believed they were consistently "right" simply because they made enough bets to eventually get one right. Maybe not in enterprise value terms, but in volume of companies there are FAR more successful and profitable SaaS businesses that you've never heard of because they were started by entrepreneurs in their respective domains (the MD who started an EHR platform or the truck driver who started a TMS solution, etc.).
It's actually pretty easy to read. I wasn't sure how to at first, but I asked my hyperswarm of Claude Mythos bots, and after several hours they recommended I use my eyes to look at the words, and then my brain could read them. As an AI skeptic I wasn't so sure, but it indeed worked.
Does the actual outcome matter or does the promise/what is sold/pushed upon society matter?
Was social media a clean win for human social interaction? Was fast food the replacement of food that was promised in the 70s?
Were doctors and shrinks in phone apps rather than real life a clean win for diagnosis, understanding and improvement of life?
But society was completely shaped by the market in these questions anyways in a way that is impossible to deny.
While the author might be right about the arrogant PhDs imposing their view on other fields they know jack shit about, it doesn't make it less true when you have billions backing it and making it a reality, it will eventually be it, even if it's arguably worse.
The failure of an over-leveraged hedge fund run by someone with no investing background who was just regurgitating what other people have been saying does not mean that labor markets are going to survive. The author has intellectual arrogance of his own to consider.
Is this really 'when genius fails' or actually another 'we're very bad at identifying genius'?
It's a word that's thrown around a lot (particularly in tech) that rarely has much to back it up. Often, it seems to be used for 'confident young man', or even 'grifter', and then when the idol falls, the same people quietly focus on lionising someone new.
There’s a tremendous amount of naivety on display by the big labs.
A bit of that is to be expected from any disrupter, but there’s now serious questions about if these startups have the basic competencies required to survive long term.
This kind of turbo confident naivety is quite common among software engineers.
I know people preaching about LLMs being the Next Industrial Revolution, who couldn’t place the industrial revolution in the correct century.
I have people at work talking about jobs and technologies getting replaced by AI who don’t understand how a lawn mower works, who don’t know who Lenin was or what a differential equation is, because they’ve simply been writing JavaScipt since high school.
Ahh well. Let them fail if it's so obvious they will. Let the financiers fail too since they're all so dumb.
There's idiocy and overconfidence everywhere. When more money is involved, I guess it's more entertaining to watch. At least they're trying something. Maybe there's more effective ways to fleece people of their money that yield greater societal benefits, but I think there's some beauty in someone thinking they have it all figured out being able to lose their money on a shitty bet.
Now once they cry for government help, that's where I care.
They are going to run out of real world data very soon. AI labs need to hire non-cynical domain experts to build physical hardware/automation that can generate experimental data. They are hard to find though. Most domain experts are so tunneled-visioned within their own domain that they can't think outside of it.
The intellectual arrogance of the AI labs is just the next from of intellectual arrogance from Silicon Valley. I've seen the exact same arrogance from everything in SV for the past 15 years; it's probably been going on longer, but I wasn't in a position to notice then
Working in AI was once reserved for intellectually gifted people. But with the recent advances in compute performance, working in AI today is basically not much different from pounding on a black box as long and as hard as necessary to make its output correspond to what you want it to be.
Obviously everyone likes the think "this time it's different" but with LLM's and AI it certainly seems to be the case.
Having a machine that's even as smart as most human's + cheap robotics, does seem to lead to us having no jobs in near-ish future eventually. (5-15 years)
Though obviously AI progress could hit a brick wall tomorrow.
The worst materialization of this are the slew of startups that are going to “watch your employees work so they can learn their job” as if random clicks on a computer can be reversed engineered to automate someone’s job.
This sort of audacity is a large part of the reason people increasingly hate AI
These hedge fund guys aren't dumb. They're just milking the boom while they can, even if they won't admit it to themselves.
Hedge funds often get 20% of the upside. Pocket it, and then when it goes bust just walk away with the money. Defend yourself by saying "everyone went bust that year."
37 comments
[ 0.22 ms ] story [ 34.1 ms ] threadHis fundamental thesis is as an AI maximalist, with expertise on who will be the winners.
His fund completely blew up, even though his fundamental thesis still may be correct.
He made a ton of money early, but then he just kept increasing his bet size via leverage. The problem is that these stocks may be going up, but they are EXTREMELY volatile. When the whole market went down with some of the Iran news, his margin levels dipped too low, and his lenders issued margin calls. He had to sell his stakes at losses, and his fund collapsed.
Now, part of the problem is he was not a legacy hedge fund with great relationships with his banks, because if he was, he might have been able to postpone the margin calls. Had he been able to do that, he would have been ok, because his positions actually recovered within a few weeks.
It didn’t matter, though, because he was forced to sell at the low point.
In other words, even if your long term bet is correct, if you are maxing out your margin to max out your bet, you are one volatile dip away from blowing up.
No matter how good the bet, you have to be able to ride out the volatility to survive.
It’s like if someone offered you a bet on a coin flip, where they will pay 5x your bet if you win. How much should you bet on each flip?
Well, by expected value you should bet everything you have, because your EV is 2.5x, so the more you bet the higher your expected value.
However, if you bet it all, you have a 50% chance of losing all your money and not being able to make any more bets.
Too true. This isn't limited to AI, either. The most obvious example in my lifetime was during peak blockchain hype, when people who had never worked in finance convinced themselves that blockchain was going to act as the backbone for how money gets moved around. As if the problem that needing solving was Bank of America doesn't trust Capital One to update a number in their database.
The nice thing about AI, at least, is I can always push back and tell people, "Sure, we can do this with AI. I just need you to use Claude or ChatGPT manually to prototype how it would work." This normally results in the requestor realizing that there's human judgment calls involved in the inputs, process, or outputs that require meatbag intelligence.
> Too true. This isn't limited to AI, either.
That's so true there's a term for it: engineer's disease/engineer's syndrome.
https://en.wiktionary.org/wiki/engineer%27s_disease
https://ask.metafilter.com/297591/Origin-of-the-term-Enginee...
"You don't got to frame walls anymore!", something that is the most enjoyable part of residential construction, takes like a day of work at most, and can be done while on autopilot. "We can instantly generate wall plans for stud location from top-down blueprints!", oh that thing people do in their head because they just space it all on 16 or 24 inch centers and it doesn't actually matter which side you start on? "We can wirelessly drive your tractors!", the easiest task on the farm which all the weeks of maintenance and planning and setup led up to?
Now don't get me wrong, there are many things that can be improved in those industries certainly, but you aren't going to know what those things are until you actually go on the job site and learn how they currently do it daily. Most all of the low hanging fruits have already been picked one way or another and a fancy step stool is not going to be industry changing.
I am trying to fathom a more catastrophic failure in communications than the lab CEO's using this narrative. The reality ends up being closer to Radiology jobs [1], but we're never able to fully realize this because the press is running around like headless chickens and unloading the fear on a public that then reaches for any leverage they can get (see: Datacenter NIMBY-ism).
Any short term FOMO valuation bumps they got from this narrative has to now be liquidated into paying premium for server racks in a warehouse.
Reminder folks: There is no evidence that AI is not normal technology [2]
[1]: https://www.apollo.com/wealth/insights-news/insights/daily-s...
[2]: https://knightcolumbia.org/content/ai-as-normal-technology
As I understood it, the attraction of blockchain was more that the money could get moved around without Bank of America and Capital One knowing about it, and therefore without those financial institutions reporting the transactions to the government as they are required to by law. Which of course screams "money laundering" and "tax evasion" to the government. Yes, there are possible legitimate reasons for wanting to be able to move cash around anonymously and invisibly, but they're basically rounding error compared to the ones mentioned above.
In other words, blockchain on this view was a classic example of the old saying that you can't apply a technical solution to a social (or in this case political) problem.
> Climbing Mount Everest and getting a PhD are both hard. I wouldn’t say that doing one means you can do the other “no problem.”
and here:
> Surgeons and electrical engineers may be extremely smart and accomplished, but that doesn’t mean they have the ability to weigh in on climate science.
Feels tangentially like a version of Nobel disease[0], discussed here[1] a couple of weeks ago.
---
[0] - https://en.wikipedia.org/wiki/Nobel_disease
[1] - https://news.ycombinator.com/item?id=49166918
This exactly. As someone who once lived in Silicon Valley and then ended up interviewing 700+ CxO's of profitable (or nearly profitable) SaaS businesses in various domain specific fields, it amazes me how many prominent "experts" in SV believed they were consistently "right" simply because they made enough bets to eventually get one right. Maybe not in enterprise value terms, but in volume of companies there are FAR more successful and profitable SaaS businesses that you've never heard of because they were started by entrepreneurs in their respective domains (the MD who started an EHR platform or the truck driver who started a TMS solution, etc.).
I just skimmed it, so I'm hoping someone with more fortitude can summarize.
Was social media a clean win for human social interaction? Was fast food the replacement of food that was promised in the 70s? Were doctors and shrinks in phone apps rather than real life a clean win for diagnosis, understanding and improvement of life?
But society was completely shaped by the market in these questions anyways in a way that is impossible to deny.
While the author might be right about the arrogant PhDs imposing their view on other fields they know jack shit about, it doesn't make it less true when you have billions backing it and making it a reality, it will eventually be it, even if it's arguably worse.
It's a word that's thrown around a lot (particularly in tech) that rarely has much to back it up. Often, it seems to be used for 'confident young man', or even 'grifter', and then when the idol falls, the same people quietly focus on lionising someone new.
Interesting.
A bit of that is to be expected from any disrupter, but there’s now serious questions about if these startups have the basic competencies required to survive long term.
I know people preaching about LLMs being the Next Industrial Revolution, who couldn’t place the industrial revolution in the correct century.
I have people at work talking about jobs and technologies getting replaced by AI who don’t understand how a lawn mower works, who don’t know who Lenin was or what a differential equation is, because they’ve simply been writing JavaScipt since high school.
There's idiocy and overconfidence everywhere. When more money is involved, I guess it's more entertaining to watch. At least they're trying something. Maybe there's more effective ways to fleece people of their money that yield greater societal benefits, but I think there's some beauty in someone thinking they have it all figured out being able to lose their money on a shitty bet.
Now once they cry for government help, that's where I care.
Having a machine that's even as smart as most human's + cheap robotics, does seem to lead to us having no jobs in near-ish future eventually. (5-15 years)
Though obviously AI progress could hit a brick wall tomorrow.
It’s more important to keep an open mind, and readjust your assumptions regularly.
This sort of audacity is a large part of the reason people increasingly hate AI
Hedge funds often get 20% of the upside. Pocket it, and then when it goes bust just walk away with the money. Defend yourself by saying "everyone went bust that year."