Hacker News Board Hacker News, browsed like a classic message board
Live data from Hacker News

The Productivity J-Curve [pdf] (2018)

ide.mit.edu

Post reply on HN
Print view View original

21–28 of 28 posts

Re: The Productivity J-Curve [pdf] (2018)

#21

Post by zbentley »

jmalicki wrote: post #8
metalliqaz wrote: post #6

Earlier quoted context omitted.

are you saying that previous technologies had effectively infinite supply?

It's not that supply was actually infinite, but you didn't realistically have situations where you said "I want to buy GPUs for a data center" only to be told "there's a 3 year waiting list." You might have two months after NVidia 3090s came out where they were short, but it is nothing like today.

Citation needed. Industries that faced multi year supply constraints in recent memory include: nuclear power, battery manufacturing, flagship commercial aircraft models, late-stage pharmaceutical safety certification, certain luxury cars, and more.

Re: The Productivity J-Curve [pdf] (2018)

#23

Post by stymaar »

alephnerd wrote: post #11

FYI about terminology before people who don't read the paper comment 1. GPT means general purpose technology or any sort of new technology that has a compounding effect on productivity, not the OpenAI model. 2. Productivity in this case means economic output , not the colloquial definition that means "hard work". If it takes 5 automotive factory workers to assemble a car manually but 2 with industrial automation, the…

> 2. Productivity in this case means economic output, not the colloquial definition that means "hard work".

The colloquial definition doesn't mean “hard work”, it also means you “produce” more (optionally, for the same amount of time). The difference is how you measure what's being “produced”.

With the “economic output” definition a barista in a posh bar in SF is “more productive” than their counterpart in a popular place in rural Minnesota, because it generates more revenue, even if the later serves twice as many patrons while also keeping the restroom clean (which would colloquially make the later “more productive”).

“Economic output productivity” can also decline if consumer spending do so, not because workers are “less productive” (in the colloquial sense), but because unsold goods or services don't count as “economic output”.

(IMHO overloading things that have a well-defined colloquial term is a very bad habit of economists and it makes things needlessly confusing for laypersons)

Re: The Productivity J-Curve [pdf] (2018)

#24

Post by jmalicki »

zbentley wrote: post #21
jmalicki wrote: post #8

Earlier quoted context omitted.

It's not that supply was actually infinite, but you didn't realistically have situations where you said "I want to buy GPUs for a data center" only to be told "there's a 3 year waiting list." You might have two months after NVidia 3090s came out where they were short, but it is nothing like today.

Citation needed. Industries that faced multi year supply constraints in recent memory include: nuclear power, battery manufacturing, flagship commercial aircraft models, late-stage pharmaceutical safety certification, certain luxury cars, and more.

Fair!

At the same time, a lot of interest in this paper is specific to the AI industry (even if it's not only about that), and outside of high-end labor, the other constraints are new there.

Re: The Productivity J-Curve [pdf] (2018)

#25

Post by gwerbin »

ok123456 wrote: post #7
metalliqaz wrote: post #6

Earlier quoted context omitted.

are you saying that previous technologies had effectively infinite supply?

No. I'm stating where the paper's assumptions are clearly violated. AI companies are intentionally trying to monopolize the supply of inputs needed for R&D. This violates homogeneity of degree 1.

Isn't this true in a lot of situations? Basically anytime maximum production capacity of an input is limited, or scale-up time is very long, firms are large relative to the supply so they are not price takers.

Re: The Productivity J-Curve [pdf] (2018)

#27

Post by curio_Pol_curio »

gwerbin wrote: post #25
ok123456 wrote: post #7

Earlier quoted context omitted.

No. I'm stating where the paper's assumptions are clearly violated. AI companies are intentionally trying to monopolize the supply of inputs needed for R&D. This violates homogeneity of degree 1.

Isn't this true in a lot of situations? Basically anytime maximum production capacity of an input is limited, or scale-up time is very long, firms are large relative to the supply so they are not price takers.

PP's point has (mostly?) been addressed in

https://www.aeaweb.org/articles?id=10.1257/mac.20180386

Inputs that can be monopolized are manifestly _not_ the intangibles; price-taking is independent of their effect

Re: The Productivity J-Curve [pdf] (2018)

#28

Post by gwerbin »

curio_Pol_curio wrote: post #27
gwerbin wrote: post #25

Earlier quoted context omitted.

Isn't this true in a lot of situations? Basically anytime maximum production capacity of an input is limited, or scale-up time is very long, firms are large relative to the supply so they are not price takers.

PP's point has (mostly?) been addressed in https://www.aeaweb.org/articles?id=10.1257/mac.20180386 Inputs that can be monopolized are manifestly _not_ the intangibles; price-taking is independent of their effect

I don't think they implied that those inputs are the intangibles, only that the functional form assumption in the 2018 paper is unrealistic.
Post reply on HN