William Baumol

How many data centers do you have to build for 10%, um, uh, annualized growth? It's hard to build houses in parts of this country.

How many physical structures do you need for 10% annualized growth? The implications even for something like Baumol's cost disease, if you have productivity growth that is so massive in the digital economy, if the overall economy is growing at 5 to 10%, like, doesn't that imply that, like, m- movie tickets and concert tickets and, like, restaurant meals are seeing just absolutely berserk inflation because we have to increase the wages of less productive workers to keep up with the productivity of the rest of the economy? Like, you're talking about a level of societal change that I think would be, in some cases, quite wrenching and extremely unpopular.

So I tell them, like, if you want to be optimistic, like actually optimistic about this technology, you want it to add half a percent, a half a percentage point of GDP growth on an annualized basis.

That is the Goldilocks spot of America's getting richer, but the world isn't transforming so quickly that we get some kind of, you know, um, what is it called? You know, Butlerian Jihad that, that rises up against, uh, AI.

They are the same problem.

To understand why government deficits are expanding at such a rapid pace, we have to look back to the 1960s, when an economist named William Baumol asked a fairly simple question.

He wanted to understand why a string quartet playing Beethoven should be paid any more today than they were a hundred years earlier.

The instruments, he argued, exactly the same.

That's going to be a thing.

And then there's also another phenomena that occurs that's a little bit more complex to explain, but it's called Baumol's cost disease.

And it basically means that as our society gets more affluent, stuff that involves one-on-one interpersonal attention that can't be automated becomes more expensive, which means that college professors are probably going to become more expensive as time goes on.

So there's a lot of things that are going to be hitting universities all at once.

So they're different.

And the really interesting here, in the nineteen sixties, there was an economist named, uh, William Baumol, and he found something interesting about a string quartet.

He says it takes four musicians the same amount of time to play a Beethoven quartet today as it did in eighteen hundred.

You cannot get more efficient at playing the notes.

But those musicians still need wages that keep up with the rest of the economy, or they'll go do something else.

So the price of live music rises relatively to everything else, forever.

Baumol called this cost disease.

So think about this.

So the key assumption is when we automate, there's some stuff we've automated in the past, there's some, some stuff that labor's still doing.

Among the things that labor is still doing, we tend to automate the tasks that have high labor costs rather than low labor costs, okay? And equivalently, we tend to automate the things that labor's bad at first rather than labor's good at, and this will be related to something called Baumol-Tobin paradox.

So you'll see that assumption.

Um, fortunately, I think it's testable, so we don't test it, but someone could test it.