There's an interesting econometric side angle here when it comes to trying to set the tolerances in the bet that will represent truly revolutionary economic transformation while still making people be willing to take the bet. For example, a lot of people have objected that my 18% unemployment figure is so high that there's no chance I'll lose on that one. But a) I thought the whole point was that AI was imminently going to dramatically undercut the labor market and put a ton of people out of work and b) just 5 years ago we had a 15% unemployment rate due to an entirely non-AI reason. I found a genuinely difficult to set thresholds that would not make me lose the bet due to non-ai catastrophes!
I tend to agree, although I do think software engineering will be dramatically different by 2029, although I'm not sure how that will look from an econometric standpoint. There's another fragility in the terms, in that "productivity" on a real wages/worker basis has already been taking off, and it's got more to do with the cycle+demographics than AI. (Plus, some ppl think that AI transformation will lead to collapsing margins and widespread service deflation (ie Baumols in reverse)
My longer comment is on your blog, Freddie, just as I was about to cancel my subscription to start paying for Claude (& keep my wife’s budget constraints on subs,)
There are 23 major SOC categories, in the form of xx-0000. The lowest, #45 farming, has 1 million folk.
I’m pretty sure you’re expressing Alarmism Fatigue. Especially climate & Trump alarmism. The alarmism is about big changes, too fast to process. Your list of 17 objective metrics is both excellent and a great reminder that Econ revolutions are much slower than political/ government revolutions. Also that the big macro numbers are very unlikely to change a lot more than prior historical range. I think it’s a closer bet, only 50% chance you win, than Arnold’s 60% estimate.
Please consider asking your readers what they think your likelihood of winning is, in increments of 10 or 5.
Lots of technologies get adopted slower than they "should". A real world economy isn't a frictionless "perfect competition" model. I think AI will be a big deal in ten years. But not nearly as much in three.
Another difficulty is the weirdness of incentives on any bet concerning the macroeconomy. Assuming Scott holds stock, he'll be getting a payout much larger than $5k anyway if his side of the bet wins over yours. And assuming you own stock, the $5k loss would be a drop in the bucket from your returns.
Of course with this kind of bet, the most important incentive is reputational so it probably doesn't matter so much.
Too late to finesse it now, but for future bets one could just exclude "economic catastrophes not mostly attributable to AI disruption" scenarios.
Wording bets well outside of well-defined and mostly predictable contexts is not easy, which is why contract jurisprudence has some basic "all bets are off" triggers. Apparently polymarket just went through this issue this week with regard to their markets involving predictions about a "government shutdown".
Previously, shutdowns involved the bulk of the government, and so drafting bet language based on that assumption seemed reasonable. But this time, it's just DHS, which is unprecedented. Polymarket made the test for "government shutdown" hinge on whether OPM would report a lapse of appropriations and federal offices generally closed on their official status website / app. But since it's just DHS, it looks like OPM won't do that. On the other hand, DHS is huge and does tons of things and a DHS-only shutdown is still obviously a major government shutdown, even if not a "whole government shutdown".
Kalshi is apparently interpreting a DHS-only shutdown as qualifying as a "shutdown" (and the bettors on there are predicting another long one). But folks on Polymarket apparently came to the opposite interpretation, but apparently this wasn't clear until Thursday night. And so suddenly all the high-priced "yes shutdown" bets lost everything.
Ha, I know exactly where that spot is on Pearce Lane. Didn't even need AI. Interesting things going on in Bastrop, for example, that's where the Boring Company is headquartered. If you can't go across it, go under it.
Doing it wrong. Uuugh. I appreciate the principled desire to step outside your door and begin exercising without tedious driving preamble, but, yuck … Put your bike on your vehicle and drive to Bastrop State Park, cycle around there or on the park road, rare in Texas, that leads to small Buescher State Park.
Parts of Bastrop SP burned in a massive fire I dunno 15 years ago, and are finally beginning to recover.
It’s thought to be a remnant stand of pines, so no one was sure how well they would even come back under very different conditions than those that produced them.
Cycle around Austin sprawl and you learn nothing except what sprawl looks like anywhere, which is to say nothing. Also you are likely to get killed or yelled at.
If you’re living in central Austin north of the river - it *used to be* pretty easy to get down to Town Lake on a path along and underneath Mopac. Not sure how that changed with the last redo of Mopac.
You, sir, have a tough butt. 😊 I enjoyed my five years in Austin, and I see that you are also learning its charms. If the future was unpredictable before LLM Ai, it is certainly even more unpredictable now.
Well, I am 79. I never walked 5 miles a day in my life. But I guess if I had walked 5 miles a day until I was 76, it would have worked out the same anyhow.
If I wanted to throw my leg over a bike seat, I think I would forget the leg throwing altogether and just get a girl's step-through bike (if they still sell such a sexist thing, though it may have gone the way of side saddles) . . . with or without streamers on the hand grips.
They are no longer identified as girls' bikes. They are step-through bikes. Thought about it. Decided against it for other reasons, including the risk of falling for old bones.
I contemplated that for about a year. That would get me on and off the bike just fine. The problem of having a fall while on a bike remains. I decided walking will just have to do.
If only adult film industry was monitored separate from other forms of film or entertainment in the BLS!
The last 7 years have been weird for the industry, because there's a sharp rise from an alternative (only fans), and COVID caused big problems on the supply side production. Still, I was trying to find numbers the other day and it very much looks like they've lost over half their employment numbers since 2019 and AI is big part of this story.
Even absent AI, automation - eg the robots that are going to replace Amazon workers - has and will change the economy - and the important, less-talked about point is that that will *change the culture*. It has changed and will change the structure of the American populace.
That's like saying porn will never replace real sex. It's surely true in some philosophical way. But among young people, there's considerably more porn than there used to be and considerably less sex.
Or as Tyler Cowen--and thousands of economists--might say, "On the margin, porn is replacing real sex."
There's an interesting econometric side angle here when it comes to trying to set the tolerances in the bet that will represent truly revolutionary economic transformation while still making people be willing to take the bet. For example, a lot of people have objected that my 18% unemployment figure is so high that there's no chance I'll lose on that one. But a) I thought the whole point was that AI was imminently going to dramatically undercut the labor market and put a ton of people out of work and b) just 5 years ago we had a 15% unemployment rate due to an entirely non-AI reason. I found a genuinely difficult to set thresholds that would not make me lose the bet due to non-ai catastrophes!
I tend to agree, although I do think software engineering will be dramatically different by 2029, although I'm not sure how that will look from an econometric standpoint. There's another fragility in the terms, in that "productivity" on a real wages/worker basis has already been taking off, and it's got more to do with the cycle+demographics than AI. (Plus, some ppl think that AI transformation will lead to collapsing margins and widespread service deflation (ie Baumols in reverse)
My longer comment is on your blog, Freddie, just as I was about to cancel my subscription to start paying for Claude (& keep my wife’s budget constraints on subs,)
There are 23 major SOC categories, in the form of xx-0000. The lowest, #45 farming, has 1 million folk.
I’m pretty sure you’re expressing Alarmism Fatigue. Especially climate & Trump alarmism. The alarmism is about big changes, too fast to process. Your list of 17 objective metrics is both excellent and a great reminder that Econ revolutions are much slower than political/ government revolutions. Also that the big macro numbers are very unlikely to change a lot more than prior historical range. I think it’s a closer bet, only 50% chance you win, than Arnold’s 60% estimate.
Please consider asking your readers what they think your likelihood of winning is, in increments of 10 or 5.
Lots of technologies get adopted slower than they "should". A real world economy isn't a frictionless "perfect competition" model. I think AI will be a big deal in ten years. But not nearly as much in three.
Another difficulty is the weirdness of incentives on any bet concerning the macroeconomy. Assuming Scott holds stock, he'll be getting a payout much larger than $5k anyway if his side of the bet wins over yours. And assuming you own stock, the $5k loss would be a drop in the bucket from your returns.
Of course with this kind of bet, the most important incentive is reputational so it probably doesn't matter so much.
Too late to finesse it now, but for future bets one could just exclude "economic catastrophes not mostly attributable to AI disruption" scenarios.
Wording bets well outside of well-defined and mostly predictable contexts is not easy, which is why contract jurisprudence has some basic "all bets are off" triggers. Apparently polymarket just went through this issue this week with regard to their markets involving predictions about a "government shutdown".
Previously, shutdowns involved the bulk of the government, and so drafting bet language based on that assumption seemed reasonable. But this time, it's just DHS, which is unprecedented. Polymarket made the test for "government shutdown" hinge on whether OPM would report a lapse of appropriations and federal offices generally closed on their official status website / app. But since it's just DHS, it looks like OPM won't do that. On the other hand, DHS is huge and does tons of things and a DHS-only shutdown is still obviously a major government shutdown, even if not a "whole government shutdown".
Kalshi is apparently interpreting a DHS-only shutdown as qualifying as a "shutdown" (and the bettors on there are predicting another long one). But folks on Polymarket apparently came to the opposite interpretation, but apparently this wasn't clear until Thursday night. And so suddenly all the high-priced "yes shutdown" bets lost everything.
Ha, I know exactly where that spot is on Pearce Lane. Didn't even need AI. Interesting things going on in Bastrop, for example, that's where the Boring Company is headquartered. If you can't go across it, go under it.
Doing it wrong. Uuugh. I appreciate the principled desire to step outside your door and begin exercising without tedious driving preamble, but, yuck … Put your bike on your vehicle and drive to Bastrop State Park, cycle around there or on the park road, rare in Texas, that leads to small Buescher State Park.
Parts of Bastrop SP burned in a massive fire I dunno 15 years ago, and are finally beginning to recover.
It’s thought to be a remnant stand of pines, so no one was sure how well they would even come back under very different conditions than those that produced them.
Cycle around Austin sprawl and you learn nothing except what sprawl looks like anywhere, which is to say nothing. Also you are likely to get killed or yelled at.
If you’re living in central Austin north of the river - it *used to be* pretty easy to get down to Town Lake on a path along and underneath Mopac. Not sure how that changed with the last redo of Mopac.
I regret being so cynical that I also have to bet against Freddie. I have seen the light, and holy cow, it’s so very damn bright.
You, sir, have a tough butt. 😊 I enjoyed my five years in Austin, and I see that you are also learning its charms. If the future was unpredictable before LLM Ai, it is certainly even more unpredictable now.
I used to bike 50 miles or so almost every Sunday in the summer, but I'm Arnold's age and just can't any more. It's not the butt; it's the legs.
😂 I’m 77. until last year, I walked at least 5 miles a day. I haven’t been able to throw my leg over a bike seat for at least three years.
Well, I am 79. I never walked 5 miles a day in my life. But I guess if I had walked 5 miles a day until I was 76, it would have worked out the same anyhow.
If I wanted to throw my leg over a bike seat, I think I would forget the leg throwing altogether and just get a girl's step-through bike (if they still sell such a sexist thing, though it may have gone the way of side saddles) . . . with or without streamers on the hand grips.
They are no longer identified as girls' bikes. They are step-through bikes. Thought about it. Decided against it for other reasons, including the risk of falling for old bones.
Get a woman's bike? :)
I contemplated that for about a year. That would get me on and off the bike just fine. The problem of having a fall while on a bike remains. I decided walking will just have to do.
Or a really good bike seat . . . or both.
😂 Yut
If only adult film industry was monitored separate from other forms of film or entertainment in the BLS!
The last 7 years have been weird for the industry, because there's a sharp rise from an alternative (only fans), and COVID caused big problems on the supply side production. Still, I was trying to find numbers the other day and it very much looks like they've lost over half their employment numbers since 2019 and AI is big part of this story.
There are always canaries in the coal mine.
Even absent AI, automation - eg the robots that are going to replace Amazon workers - has and will change the economy - and the important, less-talked about point is that that will *change the culture*. It has changed and will change the structure of the American populace.
What kind of bike do you ride? Drop bar road bike? (Guessing so if you went 50 miles.)
a 21-speed hybrid that I bought here.
That's like saying porn will never replace real sex. It's surely true in some philosophical way. But among young people, there's considerably more porn than there used to be and considerably less sex.
Or as Tyler Cowen--and thousands of economists--might say, "On the margin, porn is replacing real sex."
For this we are erasing - stomping on - all prior efforts at securing a more stable energy future.
Might as well crank up the AC and throw open the windows and air condition the world.