The Simple Mathematics of Chinese Innovation | Hacker News Reader

The Simple Mathematics of Chinese Innovation

marginalrevolution.com

33 pointssurprisetalk33 comments

It is actually a very simple formula, one which has already seen a number of iterations - Japan, ROK, ROC all followed the same path: Strategic endurance provided by single party governance over several decades, investment in education, central planning on core industries, protectionism to nurture, before expanding globally to dominate. PRC is really just a scaled up version of the 'East Asia Tiger', albeit one with monstrously sized domestic market.

That will never work for an economy the size of China. If China really starts producing it will easily satisfy the full demand of the entire world. So this can't happen.

But, China is a centrally planned economy. Every centrally planned economy died (or went into a state so bad people starve, like North Korea) because it gives higher ups an incredible incentive to lie about everything. This is made easy by the fact that centrally planned economies suppress dissent.

> If China really starts producing it will easily satisfy the full demand of the entire world. So this can't happen.

(flips over random stuffs nearby) ... yep it can't happen. It never happened.

China is not a centrally planned economy.

There are five year plans. The 14th Five Year Plan is just finishing up, and the 15th Five Year Plan is in final discussions, with formal approval in October. Comments from outside China indicate that about 80% of the goals of the 14th plan were achieved. High speed rail, energy, electric cars, mechanization of agriculture, and consumer goods worked out very well. There was overbuilding and a crash in housing and commercial construction. Areas of technical trouble are jetliners and high end semiconductors. That's about what was expected. The five year plans have been ambitious but realistic in recent decades. Up until the 1990s, they were totally unrealistic. Famines resulted.

These plans mostly drive capital allocation. It's not like the USSR's GOSPLAN, which actually set production quotas for factories. That worked badly, especially with a one-year update cycle.

Some economic planning is not central planning. Otherwise the US in the 30s and post-war France were both centrally-planned economies.

For a while, both were. Certainly once the US entered the war it was a centrally-planned economy.

I don’t think you actually disagree with the person you replied to. The point of saying China isn’t a centrally planned economy wasn’t to say there is no central planning at all, but that it’s invalid to conclude that the level of planning China does do condemns it to failure.

correct - it is a centrally directed economy, where the implementation is successively and 'pyramidically localised'

As long as 4b in developing world slowly develops the pie will continue to grow and PRC going to be in position to capture most of it, even/especially if it steps on toes of western incumbants.

Also we're in 12g surveillance state era, central gov can soon monitor every production line and get granular metrics via variety of oblique indicators, we're a few years away from heaven is high and the emperor is far away, to camera and microphone is in your face and emperor is omnipresent because every thing a bureacrat says will be cross checked by sensors and llm. The thing about centralized power is it actually cares about truth (which is orthogonal to suppressing it) because ultimately their heads on the line.

VS US polity where polticians lie and no one really gives a shit, this is made easy by the fact that democracy basically means government by the people of the people for the people. But the people are retarded and fundmentally they don't care enough that higher ups have incredible incentive to lie about everything.

Single party effectively shuts out poor people from demanding a welfare state. The west kind of accomplishes the same goal by just keeping poor people outside its borders and outsourcing the single party governance of poor people to China and rest of wold. The filibuster effectively creates a single party in America though. you just need a little bit of top down governance, a la Trump, or Xi Jin ping. It’s when you have opaque , behind the curtains governance, behind Biden / Kamala that you get a dis coordinated, hard to rationalize mess.

The whole article boils down to "more people is good enough", a point that feels trivially disproved by the existence of India; clearly there's a lot more going on than just a large population.

well said, just numbers with no plan or investment is worse even

The differences in macroeconomic policy between America and China has been described as a difference between optimizing for GDP growth and optimizing for productivity growth: America will be satisfied with increasing GDP at the expense of productivity, and China will be satisfied with with increasing productivity at the expense of GDP. The Chinese state owned enterprises are able to burn through billions of dollars of investment with little or no returns so long as the end result increases total productivity in the economy at large, and not just profit at the level of the firms.

> China will be satisfied with with increasing productivity at the expense of GDP

Yeah I don't think that's correct. They've got explicit GDP growth targets which implies they do care about GDP growth.

https://www.forbes.com/sites/ywang/2025/03/04/china-sets-202...

Not only that, but they care about them so much that they feel the need to lie about GDP figures. https://en.wikipedia.org/wiki/Li_Keqiang_index

Their own former VP had his own measurement he used as a proxy for GDP growth since he didn't believe the official figures.

Is it a good idea to start learning Mandarin in 2025 to be able to follow these Chinese scientific innovations?

This reason alone is unlikely to provide sufficient motivation to keep grinding away. Language acquisition can be a slog even though it's (eventually) super rewarding.

Let's put the question differently: let's imagine we already know Chinese. How much will we be potentially be able to capitalize on that knowledge in the coming years?

How much do you capitalize on reading papers that are written in English right now?

I would argue even for scientific collaboration the gains will be minimal. The LLMs are more or less capable of translating content in natural sciences biderectionally, and most researchers already have a passing knowledge of English. On the contrary I would not expect this to be applicable in humanities, since translating (for example) psychology material is much more nuanced.

On the other hand, IMO this would be quite influential in finding contacts for business or manufacturing.

I was looking through stuffs on vacuum tubes the other day. And all the interesting ones were already in the ever so slightly arcane versions of the language I'm most familiar with... that felt weird. But I also learned that ITO on glass was a common thing for a long time.

> let's imagine we already know Chinese

This is the case for many people. What else do you bring to the table?

Let's look at an example of a pretty closely related skill: imagine we're already literate. What's that worth?

a) important papers are published in English? and b) don't we have AI that's really good at natural language translation, especially within constrained contexts such as scientific papers?

China is a great place to do business, especially at the cutting edge.

by inertia science is done and will be done in English. Notwithstanding the Chinese/Japanese/Korean scientific journals - anything "big" gets translated. With LLMs the barrier for researchers to write in English is reduced

In the programming sphere on the other hand.. I get the impression there is a giant OSS Chinese community that doesn't do anything in English..

My Chinese is too crappy to follow technical topics - so it's possible my impressions are wrong

> anything "big" gets translated.

> With LLMs the barrier for researchers to write in English is reduced

Anything "big" gets translated to English. Not from English. With LLMs the barrier for researchers to read in English is reduced. Not to write. Researchers in China has no motivations to think or publish in English, and so they won't. It has always been that way, I think?

Not in my experience - but maybe it depends on the field. Publishing for major international journals is good for your career everywhere. Publishing for a small regional journal is the fall-back option.

> Anything "big" gets translated to English.

Do you have any specific examples of noteworthy recent research that was not originally published in English? I've honestly never come across it. (this used to be a thing a million years ago.. Soviet publications and really old stuff in German)

I'm not sure if your observations and my arguments contradict. My point is, lots of science happen in local languages and gets translated to English, rarely from English. English is the lingua franca as in LANG=C is always the fallback, not as in en-US.UTF8 is the interface default.

Germans discovered chemical formulas in German, Soviets did metallurgy in Russian. It's consistent that guys in China discuss AI and OSS bugs in Chinese. They support fallback modes and exported data is often converted to standard formats for the best exposure, but their internal formats are not always in English.

No, I don't have any examples to show, I just think "English is lingua franca of science" types of statement is misleading and create false impressions that the original researches always happen in English, not just the data exchange parts.

The original post was asking if he needs Chinese to "follow these Chinese scientific innovations". That would pretty much entail reading publications from Chinese scientists - which for the most part will be in English.

> It's consistent that guys in China discuss AI and OSS bugs in Chinese.

So if they internally at the lab/office talk in Cantonese or whatever seems irrelevant. If you want to work with Chinese researchers, then it's likely helpful to learn Chinese. But that wasn't the question

> Germans discovered chemical formulas in German, Soviets did metallurgy in Russian.

They published in German and Russian. Now-a-days they do not

Most code is write only , even in English.

Most important papers are already in English, are they not? Also I can't imagine how long it would take to learn the language to the point of being able to parse scientific texts. Maybe it's skill issue on my part but despite learning the english language since 2nd grade (i am almost 30 now with a degree earned in english) I still sometimes struggle with scientific papers

I feel like the time you would spend on the language is better spent on learning the actual science

If your goal is to get a foreign language to a level where you can meaningfully understand scientific works published in it, you're going to spend an incredible amount of time on it. (And will not likely meet it without immersion.)

If your goal is to follow them at a superficial level, waiting for someone else to translate the important bits, or using machine translation is going to be a much safer bet.

But don't just believe what I say. Ask any >50 year old PHD from a non-English speaking country (preferably an ex-Soviet block one, or an Asian one), they'll tell you how much of a nightmare reading English papers was for them.

government is not good for everybody，but new tech does

Reply on news.ycombinator.com