Minor pedantic point. The articles uses “Black immigrants” and “African immigrants” interchangeably.
The majority of Black immigrants don’t immigrate from Africa. The biggest source of Black immigrants is the Caribbean - Jamaica and Haiti, specifically.
>Officials explained that the 10,000-yen levy will apply to hotel stays costing 100,000 yen or more per night
I can't imagine the average Kyoto tourist will be spending $650 a night - when I went years ago we spent about $250 for a 4 star hotel. It'll be good for revenue but I have doubts about crowd control.
I imagine as passive investment grows to become a larger and larger share of market activity, the opportunity for exploiting pricing inefficiencies also grows, increasing the potential reward for price discovery. I imagine there is an equilibrium somewhere that most markets will settle around.
This used to be true in CCP (the leadership leans heavily on technocrats with elite education) but Xi has changed all that.
Xi is from the Cultural Revolution generation that had their education cut short, and he distrust the “elites”, so the leadership now is heavy on loyalty and low on competence (yeah we know what that’s like)…
I too also felt that it was never a compelling argument to say we should sell China our high-end chips in order to keep them dependent and to prevent them developing their own AI hardware stack. It always seemed obvious to me that they would try to reverse engineer our stuff whether or not we were selling it to them.
But I’m not a well known smart economist, so I figured I must be naive and missing something.
Also not to mention, during the Cold War nobody tried to make the case that we should sell our advanced military hardware to the USSR in order to disincentivize them from developing their own.
Regarding your antitrust post… We need to distinguish between economic sectors in which it’s possible to grow the pie and those in which the pie is more or less fixed and all we’re trying to do is figure out how to divide it.
It probably won’t work to use antitrust legal enforcement against things like social media, crypto, AI, computers, etc. As soon as we corner one of them, someone invents a New Thing that renders the old debate obsolete. At best, we can kneecap today’s Villain of the Week and free up space for an even more nefarious villain to take its place.
The same logic DOES NOT apply in sectors which are roughly capped in size—agriculture, for instance, or meat production, or lumber. In sectors such as these, the presence of oligarchs lead not so much to innovation as monopsony: a relentless squeezing of margins and Gilded-Age dynamics.
We need to develop a legal framework (and economic analysis) nimble enough to make this distinction and not try to shove everything into a one-size-fits-all framework.
Very nice set of topics.... Especially enjoyed the "selective immigration" topic. Selective immigration intensity seems to correlate heavily with unicorns/capita in US cities.
Google and Microsoft maybe are driving innovation, but lets stop pretending that Meta has anything to offer please, have you ever clicked on the AI link below a Facebook post before? Its the very definition of junk learning from junk, its embarrassing to a level that makes ChatGPT hallucinations look like the work of Einstein. It is well past time that everyone accepted that with Facebook, Zuckerberg had one great idea (which is fine, thats 1 more great ideas than most people have) and he was particularly skilled at monetising that one idea, but he is not an original or innovative thinker, he has zero insight into the future and what it will look like (the Metaverse anyone, how about NFTs?) and stop pretending that he has anything to offer when it comes to building the future
The new Albrecht & Decker paper (and Noah Smith’s take) shows no clear link between higher markups and the fall in U.S. business dynamism. If I understand his synopsis. I am no economist but I feel uncomfortable about something. Are we wrong to feel that a few mega-platforms hold too much sway. The data test industries; the lived experience comes from cross-industry infrastructures like Amazon, Google, Apple, and Meta. Both truths can coexist: the aggregate numbers look benign, yet the architecture of digital power still shapes markets, speech, and politics.
Noah, these are all particularly interesting topics. Tourism has expanded so much and where I live there is overtourism and the locals hate it. I wonder what the social impact of all the cities that have gotten overrun with tourism is. We used to go to Jackson, WY frequently, did not for ten years then went back and it was awful. But I suppose it fits Yoga Berra's "nobody goes to Jackson any more, it's too crowded" joke. It did lose its charm though, but new visitors don't realize it so on it goes.
We should not expect Chinese models to fail to reach parity with American models. Older GPUs do exactly the same math as newer GPUs, they simply do it a bit slower and using a lot more electricity. The evidence suggests that engineering better LLMs is not as hard as engineering better semiconductor manufacturing machines. This is a rapidly moving field. If the Chinese failed to reach parity given enough time that would be surprising. The goal of the restrictions aren't to prevent the Chinese from matching an American model released six months ago, it's to prevent them from matching an American model released today. Maintaining more than a six month lead due to compute seems unrealistic given the historical trends.
The more interesting question is, "If export controls aren't working, why aren't China's models exceeding American models?" We can imagine a whole slew of hypotheses here, and I'd be interested if you have any data that tests them.
Minor pedantic point. The articles uses “Black immigrants” and “African immigrants” interchangeably.
The majority of Black immigrants don’t immigrate from Africa. The biggest source of Black immigrants is the Caribbean - Jamaica and Haiti, specifically.
>Officials explained that the 10,000-yen levy will apply to hotel stays costing 100,000 yen or more per night
I can't imagine the average Kyoto tourist will be spending $650 a night - when I went years ago we spent about $250 for a 4 star hotel. It'll be good for revenue but I have doubts about crowd control.
I agree, that is just not enough more to stop anyone who wants to visit Kyoto. For many this is a once in a lifetime trip.
I imagine as passive investment grows to become a larger and larger share of market activity, the opportunity for exploiting pricing inefficiencies also grows, increasing the potential reward for price discovery. I imagine there is an equilibrium somewhere that most markets will settle around.
> China’s leaders, being smarter than, say, a gerbil...
I wonder what having a government like that feels like. Must be nice.
Offset by the secret police
This used to be true in CCP (the leadership leans heavily on technocrats with elite education) but Xi has changed all that.
Xi is from the Cultural Revolution generation that had their education cut short, and he distrust the “elites”, so the leadership now is heavy on loyalty and low on competence (yeah we know what that’s like)…
I too also felt that it was never a compelling argument to say we should sell China our high-end chips in order to keep them dependent and to prevent them developing their own AI hardware stack. It always seemed obvious to me that they would try to reverse engineer our stuff whether or not we were selling it to them.
But I’m not a well known smart economist, so I figured I must be naive and missing something.
Thank you for the validation, Noah.
Not to mention, we saw what they did to us with rare earths
Also not to mention, during the Cold War nobody tried to make the case that we should sell our advanced military hardware to the USSR in order to disincentivize them from developing their own.
Because that would be silly, right?
Excellent point
Nobody should get any help due to the race or is their ancestry
Any help should be based on class. If you're poor getting the leg up is helpful
That families, communities, and history exist means that many people get help due to their race or ancestry.
Regarding your antitrust post… We need to distinguish between economic sectors in which it’s possible to grow the pie and those in which the pie is more or less fixed and all we’re trying to do is figure out how to divide it.
It probably won’t work to use antitrust legal enforcement against things like social media, crypto, AI, computers, etc. As soon as we corner one of them, someone invents a New Thing that renders the old debate obsolete. At best, we can kneecap today’s Villain of the Week and free up space for an even more nefarious villain to take its place.
The same logic DOES NOT apply in sectors which are roughly capped in size—agriculture, for instance, or meat production, or lumber. In sectors such as these, the presence of oligarchs lead not so much to innovation as monopsony: a relentless squeezing of margins and Gilded-Age dynamics.
We need to develop a legal framework (and economic analysis) nimble enough to make this distinction and not try to shove everything into a one-size-fits-all framework.
Very nice set of topics.... Especially enjoyed the "selective immigration" topic. Selective immigration intensity seems to correlate heavily with unicorns/capita in US cities.
Great article, Noah!
Google and Microsoft maybe are driving innovation, but lets stop pretending that Meta has anything to offer please, have you ever clicked on the AI link below a Facebook post before? Its the very definition of junk learning from junk, its embarrassing to a level that makes ChatGPT hallucinations look like the work of Einstein. It is well past time that everyone accepted that with Facebook, Zuckerberg had one great idea (which is fine, thats 1 more great ideas than most people have) and he was particularly skilled at monetising that one idea, but he is not an original or innovative thinker, he has zero insight into the future and what it will look like (the Metaverse anyone, how about NFTs?) and stop pretending that he has anything to offer when it comes to building the future
The new Albrecht & Decker paper (and Noah Smith’s take) shows no clear link between higher markups and the fall in U.S. business dynamism. If I understand his synopsis. I am no economist but I feel uncomfortable about something. Are we wrong to feel that a few mega-platforms hold too much sway. The data test industries; the lived experience comes from cross-industry infrastructures like Amazon, Google, Apple, and Meta. Both truths can coexist: the aggregate numbers look benign, yet the architecture of digital power still shapes markets, speech, and politics.
Noah, these are all particularly interesting topics. Tourism has expanded so much and where I live there is overtourism and the locals hate it. I wonder what the social impact of all the cities that have gotten overrun with tourism is. We used to go to Jackson, WY frequently, did not for ten years then went back and it was awful. But I suppose it fits Yoga Berra's "nobody goes to Jackson any more, it's too crowded" joke. It did lose its charm though, but new visitors don't realize it so on it goes.
If export controls are working why are China’s open source models reaching parity with closed source ones that use high end Nvidia GPUs?
We should not expect Chinese models to fail to reach parity with American models. Older GPUs do exactly the same math as newer GPUs, they simply do it a bit slower and using a lot more electricity. The evidence suggests that engineering better LLMs is not as hard as engineering better semiconductor manufacturing machines. This is a rapidly moving field. If the Chinese failed to reach parity given enough time that would be surprising. The goal of the restrictions aren't to prevent the Chinese from matching an American model released six months ago, it's to prevent them from matching an American model released today. Maintaining more than a six month lead due to compute seems unrealistic given the historical trends.
The more interesting question is, "If export controls aren't working, why aren't China's models exceeding American models?" We can imagine a whole slew of hypotheses here, and I'd be interested if you have any data that tests them.
Great points.