AI & Tech Issue #139 ·

The Price of Not Looking at China Properly

Two large-scale studies show Western scientists barely cite Chinese papers — and Korea has the same blind spot in batteries and chips.

Oswarld (Kwangseob Ahn) · GTM & AI strategist working the seam between technology and business

The Price of Not Looking at China Properly

Opening

Dear reader, in 2025 Chinese researchers published roughly as many papers as researchers in the US, UK, Germany, and Japan combined. 8 of the top 10 institutions in the Nature Index1​ are Chinese, and Harvard has ceded the No. 1 spot to Zhejiang University. The research output of the Chinese Academy of Sciences (CAS) alone is 2.4 times Harvard’s.

And yet Western researchers barely read these papers. Two large-scale analyses published this year put numbers on that fact. In fact, there is a running joke in the AI research community:

AI is built by Chinese in China and Chinese in the United States.

This is not a matter of academic politics. It is a structural problem: the same experiments get run independently on both sides of the planet, breakthroughs spread more slowly, and in the end the side that isn’t looking pays the price. And to give you the conclusion up front — this ‘selective blind spot’ is not just a Western problem. Korea shows exactly the same pattern.

The Neglect in Numbers: Citations Stop at the Border

Two studies published this year laid out the phenomenon in data.

The first is an NBER working paper published in January by Abhishek Nagaraj of UC Berkeley and Randol Yao of MIT. Analyzing citation data from tens of millions of English-language academic papers from 1980 to 2022, they found that only about one-third of citations to Chinese papers came from outside China. Over the same period, the US and EU drew roughly half of their citations from abroad. The gap barely narrows even when you restrict the sample to papers in top-5% journals or to the top 1% most-cited papers. Even papers in good journals go unread.

The second is an analysis a Chinese-Dutch team posted in April to the preprint server arXiv2​. This study goes a step further: even after controlling for variables like paper output, it shows that Western researchers cite Chinese papers markedly less than expected. The asymmetry was striking. Chinese researchers cite American papers more than expected, while American researchers cite Chinese papers less than expected.

Of course, these figures shouldn’t be taken entirely at face value. The practice of Chinese researchers excessively citing one another to inflate metrics may have distorted the first result, and the second study could not fully control for differences in average research quality across countries. A related study by MIT professor Pierre Azoulay’s team, published in the Journal of International Economics, shows that once you correct for home bias3​, China’s global citation rank drops from 2nd to 4th. China’s domestic citation inflation is real.

But that is not the whole story. In the same study, America’s home bias also came in 16% above expectations. The fact that China’s home bias is large does not excuse the West’s under-citation. Both phenomena exist at the same time.

The field-by-field differences are the interesting part. When Nagaraj and Yao ran a field-level breakdown at The Economist’s request, domestic citation was especially high in fields like chemistry and engineering, where Chinese labs account for most frontier research. In those fields there are simply fewer peers abroad worth citing. In other words, behind the statistic ‘high domestic citation share’ may lie the reality that ‘the frontier of that field is already in China.’ What’s unfolding is a more complicated terrain than mere neglect.

Why Don’t They Look? A Trio of Trust, Ignorance, and Geopolitics

Three layers of causes overlap in this ‘selective blind spot.’

First, a trust deficit. According to Retraction Watch data, as of February 2025, 30,977 papers by China-affiliated researchers had been retracted — 55.3% of all retracted papers worldwide. Between 1996 and 2025, a paper by Chinese authors was roughly 6 times more likely to be retracted than one by US or UK authors. The Chinese government banned per-paper bonuses and publication quotas in 2020 and has been cracking down on paper mills4​, but a reputation takes 1 year to collapse and 10 years to rebuild.

Second, cultural ignorance. Oxford professor Dorothy Bishop points out that Western researchers don’t know the hierarchy of Chinese institutions and can’t tell similar-sounding names apart — the difference between Zhejiang University (浙江大学) and Zhejiang University of Technology (浙江工业大学), or the pecking order among the dozens of institutes under the Chinese Academy of Sciences. Without that context it’s hard to judge how much to trust a paper’s provenance, and when judgment is hard, people tilt toward not reading. According to one study, the most effective way to reduce citation bias against papers with a Chinese surname as first author was, ironically, ‘masking the author information.’ Switching the affiliation to a US institution also cut the bias by 16%, but rendering surnames as initials only reduced it even more. It is telling that the problem is milder in journals like Nature that use numbered citation formats.

Third, geopolitics is shrinking the points of contact themselves. The international co-authorship rate of Chinese papers fell from an average of 24% in 2000–2019 to 18% in 2024. The US National Institutes of Health (NIH) foreign-interference investigations created a chilling effect on collaboration with China-born researchers, and a 2024 study found that US researchers with Chinese collaborators actually became less productive than peers with collaborators from other countries. Political risk raised the cost of academic collaboration.

These three don’t operate separately. Retraction records breed distrust, distrust breeds indifference, and geopolitics shrinks the points of contact — a vicious cycle. And as contact shrinks, the very opportunity to ‘see the other side properly’ disappears. The fewer chances there are to meet at conferences, co-author research, and visit each other’s labs, the more foreign the names on papers become — and papers with unfamiliar names go increasingly unread.

Korea’s Selective Blind Spot: Exaggerate or Ignore

I can’t tell this story without bringing up Korea.

Here’s something I’ve been saying for a long time: the area Korea strangely exaggerates is China’s semiconductors, and the area it inexplicably underestimates is China’s rechargeable batteries.

Chinese semiconductors generate an oversupply of ‘crisis’ and ‘catch-up’ narratives, courtesy of US export controls. Every time CXMT or SMIC’s 7nm process makes headlines, reports pour out claiming “Samsung is being overtaken” — while the concrete context of yields and mass-production capability is left out. Meanwhile, Chinese battery makers (CATL, BYD, and others) surpassed Korean firms in global market share long ago, yet the frequency with which that number gets addressed head-on in the Korean press is astonishingly low.

The West ignoring Chinese scientific papers and Korea underestimating Chinese batteries are the same structure: seeing only what you want to see and looking away from what’s uncomfortable. Exaggerate the fields that feel threatening, avert your eyes from the fields where you’ve already fallen behind — and in the end, never see the other side whole.

The same pattern repeats in the West. When DeepSeek appeared, the first reaction was “Chinese software is ultimately imitation” — but it delivered genuinely original results in inference cost efficiency. Nvidia’s stock dropping 17% in a single day was the moment the market ‘acknowledged’ that achievement. Belittling China’s software capability, not reading its papers — they are all variations on the same selective blind spot.

The costs of this blind spot are concrete. When both sides run the same research independently, the research efficiency of humanity as a whole drops. One side starts from scratch, unaware of results the other has already established. On global challenges like climate change, infectious disease, and food security, this duplication is not mere inefficiency — it is wasted time.

Meanwhile, Korea currently ranks 7th in the Nature Index. Its share grew 4.1% versus 2023, overtaking Canada, and in the physical sciences it climbed to 4th. While the shares of major Western countries — the US, Germany, the UK — fell by more than 7%, Korea grew. In the biological sciences its share rose more than 11%. As the West declines and China soars, Korea is quietly climbing.

What matters more is Korea’s position. According to US National Science Foundation (NSF) data, Korea’s international collaboration index (ICI) with the US stands at a meaningful 1.1, while it simultaneously receives close-to-expected citations from China. This ‘bridge’ position — maintaining academic contact with both the US and China — is Korean science’s greatest strategic asset. But riding along with the selective blind spot (looking at one side and ignoring the other) means throwing that asset away.

Oswarld’s Take

I look at this problem less through the lens of ‘science’ than through the lens of GTM (go-to-market strategy).

There’s a pattern I’ve seen countless times while building GTM strategies. No matter how good the product is, if it doesn’t win the market’s trust, it doesn’t get distribution. And trust never arrives simultaneously with quality improvement — it always follows with a lag. What Chinese science is going through right now is exactly this ‘trust lag’ problem. Quality has risen to No. 1 on the Nature Index, but its reputation is still trapped in the shadow of that 55% retraction share.

What this citation data reveals, though, is a structure scarier than that. Normally, even with a lag, the market eventually recognizes quality. But when geopolitics intervenes, the lag doesn’t shrink — it widens, because the points of contact themselves shrink. The international co-authorship rate falling from 24% to 18% means the very channels for seeing each other properly are narrowing.

I see Korea not as an observer of this blind spot but as a party to it. Exaggerating semiconductors, underestimating batteries, belittling Chinese software — all of it is a variation on the same structure of ‘not looking properly.’ And the price of a blind spot is always paid by the side that isn’t looking. That’s true in markets, and it’s true in laboratories.

That Korea sits at 7th in the Nature Index while keeping relationships with both the US and China also means it is positioned to fill in both sides’ blind spots. Whether Korea uses that strategically or free-rides on one side’s blind spot — I believe that will decide the path of Korean science and industry from here.

Closing

The West ignoring Chinese papers, Korea underestimating Chinese batteries, the West belittling Chinese software — they are all variations on ‘not looking properly.’ And while you’re not looking, the other side keeps advancing.

In a world where 8 of the Nature Index’s top 10 institutions are Chinese, ‘not looking’ is not disdain — it is a cost. The cost of running the same experiment separately on both sides, the cost of breakthroughs spreading slowly, and the cost of not even realizing the board has changed. The window in which Korea can serve as a bridge between the US and China is not infinite. Now is the moment to choose.

What part of China are you deliberately ‘not looking at’ in your own field? Tell me in the comments.

References & Further Reading

Primary sources

  • Abhishek Nagaraj & Randol Yao, “The Geography of Science”, NBER Working Paper #34694, 2025. : The core study analyzing citation data from tens of millions of papers, 1980–2022. Its finding that only one-third of citations to Chinese papers come from abroad is the starting point of this newsletter.
  • Shumin Qiu, Claudia Steinwender & Pierre Azoulay, “Paper Tiger? Chinese Science and Home Bias in Citations”, Journal of International Economics 157, 2025. : Shows how China’s global citation rank drops from 2nd to 4th after correcting for home bias. The comparison between home bias in trade and home bias in scientific citation is especially interesting.
  • “More than half of all retracted papers are from China, analysis finds”, Chemistry World, April 2026. : The latest reporting on Chinese paper retractions, based on Retraction Watch data.
  • “Nature Index 2025 Research Leaders”, Nature Index, June 2025. : The original data behind the country rankings — Korea’s rise to 7th, China’s widening lead at No. 1, and the decline of major Western countries.

Background

The author, Kwangseob Ahn, is a professor of business administration at Sejong University and lead consultant at OBF (Oswarld Boutique Consulting Firm). He teaches statistics and data analysis — business data management, business analytics — at the university, while in the field he leads GTM strategy and AI strategy consulting, designing the interface between technology and business. He has published academic research on memory architecture for AI dialogue systems (HEMA) and runs Daily Arxiv, a project curating global AI papers every day. He completed a master’s program at Korea University’s Graduate School of Management of Technology and its KMBA. He is the author of Homo Brainless: The People Who Outsource Their Thinking.

Footnotes

  1. Nature Index: A ranking by Nature Portfolio that scores the research performance of countries and institutions based on papers published in 145 leading natural-science and health-science journals. Because it computes each author’s fractional contribution (Share) rather than raw paper counts, it is more refined than simple volume comparisons. ↩

  2. arXiv: A preprint server where academic papers are posted publicly before peer review. Widely used in physics, mathematics, and computer science, it is the fastest channel for accessing the latest research. ↩

  3. Home bias: The tendency of researchers to cite papers from their own country more than expected. It works much like the preference for domestic products in trade, and the same pattern shows up in scientific citation. China’s home bias, at 57.2%, is the highest among major countries. ↩

  4. Paper mill: An operation that mass-produces and sells fake or extremely low-quality academic papers. It gets them into journals through a combination of fabricated data, plagiarized structure, and bought peer review. ↩