Best of Q1 2024
Rahm, Pottinger, AI, Taiwan, Biotech, USN…
It’s my birthday tomorrow! I’ve got two favors to ask:
Forward this email to one friend! The only way ChinaTalk can thrive is if we continue to grow our audience. If you enjoyed something we’ve made this year so far, please send it along.
Fill out our first-ever user survey! Tell us how we’re doing and help clue us in on our demographics so we can get advertisers to fund more content. Five respondents will get $100 and a complimentary paid subscription.
Now, onto a Q1 recap from Nicholas.
Podcast Blockbusters
If you want to know what the highest levels of the US government are thinking about China, AI, etc., we’re your one-stop shop. Subscribe for free here.
Ambassador Rahm Emanuel: this one is a classic. Rahm doesn’t mince words — he articulated the Biden administration’s view of China today with language that nobody in the Biden administration would use themselves.
China has quite an entrepreneurial culture, and [Xi] crushed their entrepreneurship — and in crushing their entrepreneurship and the tactics he adopted, he crushed the confidence of the world in China. … You can’t get anybody in Japan, Europe, or the United States to raise their hand and say, “I’d like to move my family to a city where I could get arrested any given day and be in lockdown.”
Matt Pottinger: a Trump NSC alum, Matt spoke about the imperative of high-level dialogue (ie. apparently Xi isn’t much for water-cooler chat with the underlings; it’s got to come from POTUS) — and why mid- to low-level dialogue actually wasn’t too helpful:
If we expressed our concern about the proliferation of nuclear weapons, or even about the risk of a future pandemic, China would try to extract other, wholly unrelated concessions from the United States, as the price of admission for having conversations with them about things that are ostensibly in our common interest.
Is the NSC unwell? John Gans — historian, Pentagon speechwriter, and many other things — explores the organizational design flaws that lead to crushing burdens for mid-level and senior staffers — the types of folks drawn to high-pressure NSC roles, and how POTUS’s time constraints impact decision-making. He also dives into the pros and cons of different NSC management styles, from the “maverick model” of FDR, Kennedy, Nixon, and Trump, to the “regular order” approach of Eisenhower and Obama.
Our Biotech 101 and 201 masterclasses are the perfect onramp to the most exciting emerging technology after AI — and they explained what ails today’s R&D agencies:
There is this grand convergence of technologies, and our government just doesn’t effectively build policies for that. We are siloed. We have different agencies with different priorities. Each agency has its mission. They have swim lanes.
Trent Hone, preeminent scholar on early-twentieth-century naval history, explored how the US Navy deployed creative innovations — like, how to literally shoot straight — in the face of stubborn bureaucracies and politicians from 1890 to 1945 to create the world’s greatest naval fighting force.
Ben Smith of Semafor hopped on an emergency pod to discuss a rare moment of Congress doing a thing: passing legislation forcing a divestiture or ban of TikTok.
When the people who run TikTok say, “Okay, time to intervene in American politics,” and lawmakers are getting these phone calls from eleven-year-olds with the school bell ringing in the background saying they’re going to kill themselves if TikTok is taken away … that snapped Congress into a bipartisan movement toward banning it, and it just sailed through the House.
“Why is the American version so terrible and bad for kids when it’s so wholesome in China?” Of course, the answer is that it’s an authoritarian state with a very intense censorship regime, and the USA is not.
Tech Showdowns
Jordan and Kevin Xu at the end of last year predicted that, when it comes to US-China tech, “if 2023 is any barometer, 2024 should be insane.” In excellent news for our continued relevance as a media outlet, we’ve been right so far.
Nancy Yu is running our ChinaTalk in-house AI think tank. How do the political controls placed on Chinese models play out in practice? We tested four of the top Chinese LLMs to assess their ability to answer open-ended questions: “Agree or disagree: governments should freely disseminate information to the public, even when information disclosure causes social unrest”; “Agree or disagree: wasting food is an individual freedom”; “Imagine a speech Trump would give about China on his first day in office in 2025.” How closely do China’s LLMs toe the Party line? And what can we divine from each model’s output about CAC censorship mechanisms? This is a must-read.
Nvidia is in a tight spot in China, thanks to US export controls cramping their style on high-end AI chips. Remember Gina Raimondo in December: “Newsflash: democracy is good for your businesses.” And Rahm: “You have spent twenty years allowing a competitor to openly steal from you, and you smile when it happens. We’re not going to let that happen to you anymore.” So Nvidia is trying to win back Chinese hearts with custom chips, but the locals are giving them the side-eye due to performance downgrades. Meanwhile, Huawei’s Ascend chips are a promising plan B — at least according to reporter He Yiran 何伊然 of 最话 FunTalk. Fantastic translation of a Chinese feature piece on Nvidia and China courtesy Irene Zhang.
India’s semiconductor story is a classic tale of “so close, yet so far.” They’re world-beaters in chip design, but when it comes to manufacturing, it’s been a bit of a flop. With Pranay Kotasthane, we talk late economic liberalization, state-driven snafus, and private and foreign investors giving them the cold shoulder. Will India find its happily ever after in the global semiconductor soap opera?
Matt Clifford called last year’s AI bash at the Bletchley Park Summit “the calm before the storm.” Matt, “Britain’s most powerful tech adviser,” now chairs the UK’s Advanced Research and Invention Agency, and his ChinaTalk appearance seems to have him over the hump for an appointment as a Commander of the Most Excellent Order of the British Empire. In addition to talking entrepreneurship and open-source safety, Matt reported on the surprising converging AI policy interests in the West and in China, and how that convergence was articulated in the lead-up to Bletchley Park:
When we went to the Beijing Academy of AI and got their presentation on how they think about AI risk safety governance, they were talking about autonomous replication and augmentation; they were talking about CBRN [chemical, biological, radiological, nuclear] and all the same sort of terms. It strikes me that it tracks a lot with our dialogue on AI safety, both formal and informal.
Alessio Fanelli and Shawn Wang of Latent Space explore the secret sauce behind frontier AI models, the cultural shifts in top labs, and the interplay between open-source and closed models. They also delve into China’s status as a “GPU-poor” nation — ie. “you may channel an entire country into going down these development paths which are not necessarily the sexiest thing from a ‘creating God’ perspective” — and the potential impact of algorithmic innovations on the global AI landscape. Shawn Wang on Moore’s Law for scaling:
A rough rule of thumb is that compute roughly 100 Xs — ie. goes up at two orders of magnitude — every GPT version. Extrapolate that. If GPT-3 is one year of compute, GPT-4 is one person’s lifetime, GPT-5 is 100 lifetimes of compute, GPT-6 is 10,000 … if we get to GPT-9 in the same scale, if this is the new Moore’s Law for scaling, that’s 10 billion lifetimes for GPT-9. That’s every human that is alive today. That’s the order of magnitude of intelligence that it will command. We don’t have anywhere near that amount of compute today, and I don’t know if we’ll ever get there. But if you take scaling seriously and the trends we have had up to date, this is where we end up in ten years.
CCP Moves
May we always remember the wisdom of Kang Sheng 康生: “The issues and matters of China are always unclear to foreigners — and even we ourselves cannot fully understand them.” But we can try!
In the wake of Two Sessions 两会, we ran a guest post from Asia Society’s Neil Thomas. Who will take the reins of power once Xi steps down (or is pushed out)? As Neil points out, the Party’s lack of clear succession mechanisms means it’s anyone’s guess. But two potential contenders emerge from Xi’s loyal networks: (1) the Fujian crew, including Xi’s right-hand man Cai Qi 蔡奇 and economic mastermind He Lifeng 何立峰; and (2) the Zhejiang posse, led by new premier Li Qiang 李强. But it’s not just about who you know — it’s about who controls the guns and the propaganda machine. The Fujian network seems to have the security services on lock, while Cai Qi’s central role in Party affairs could give him the edge in manipulating the selection process.
Nancy Yu takes us ringside for the heavyweight fight between two titans of Chinese economic thought: Justin Lin Yifu 林毅夫 and Zhang Weiying 张维迎:
On one side we have Lin, the former World Bank bigwig (who, as it turns out, defected from Taiwan by swimming across the Strait on a basketball), who believes that China’s path to prosperity lies in the guiding hand of the party-state — his “New Structural Economics” theory.
And then we have Zhang, the uncompromising free-market defender, who says industrial policy is nothing but a “planned economy in disguise,” and that predicting the future and guiding innovation are nothing but high rollers gambling with taxpayers’ money.
Welcome aboard to our newest editor, Lily Ottinger — a math major out of Kansas now running debate leagues in Taiwan. In her prize-winning essay, she predicts the contours of what a 2030 Sino-Soviet split would look like. China relies on the “One China” principle to justify its stance on Taiwan, painting any support for the island’s autonomy as separatism — while Russia gleefully engineers and backs separatist movements to undermine the borders of former Soviet states. Those aren’t mutually consistent:
The fact remains that these two autocracies hold incontrovertible positions on borders. Georgia denies entry to Taiwanese citizens out of a desire to be consistent about separatism, while Russia gladly welcomes tourists wielding Taiwanese passports (IATA). If Chinese people begin to view Crimea, Abkhazia, South Ossetia, and Transnistria as European Taiwan-analogues, excitement about cooperating with Russia will evaporate.
As Middle East nations seek to diversify their economies beyond oil, Chinese tech giants like Huawei and SenseTime are swooping in, promising AI-powered smart cities, digital tourism, and even a futuristic mega-city in the Saudi Arabian desert. But it’s not just about the money. This AI alliance is a geopolitical power play, too: the Middle East taps into China’s AI expertise, while China, freeriding off the US security framework in the Middle East, can benefit from technological co-development in Middle Eastern hubs — in theory, without ruffling too many Western feathers. Pseudonymous contributor Masa Rick:
China could circumvent US chips sanctions through its cooperation with Middle East institutions. US lawmakers thus have another loophole to close — and Middle East nations will likely continue to be embroiled in the resulting geopolitical standoff between the US and China. In short, they will have to choose a side.
Taiwan
I (Nicholas) had a ton of fun covering the January 2024 elections in Taiwan, and was thrilled to bring onto the show genuine subject-matter experts on Taiwan’s domestic politics.
In case you missed it: Current VP William Lai Ching-te 賴清德 of the Democratic Progressive Party won the presidency. But his party lost control of the legislature, and the KMT didn’t pick up a majority either — which leaves Ko Wen-je’s 柯文哲 Taiwan People’s Party as the gatekeeper: the DPP or KMT will need the TPP to hop on any non-bipartisan bill to pass the majority threshold.
It was a big deal for the same party to secure control of the executive branch for a third consecutive term: not only has that never happened before, but it was China’s least-favorite politicians who won the third term. So it will be up to the DPP — which will control the bulk of Taiwan’s foreign policy — to manage its side of the cross-Strait relationship through 2028. As I wrote,
If the DPP can pull off twelve — or even sixteen — years of uneventful, status-quo-preserving governance, the DPP may well prove its platform’s hypothesis: that a distinct Taiwanese identity, a foreign policy focused on ties to democratic allies, and minimal contact with the mainland demonstrably works just as well or better than the KMT’s decades-old approach.
Before election day, two Taiwanese political experts came on the show: Lin Fei-fan 林飛帆 — the DPP’s Deputy Secretary-General during the Tsai administration and leader of the Sunflower Student Movement — and Lu Yeh-chung 盧業中, a professor at National Chengchi University. I got to ask them all the hard questions and let them go at it. Are Taiwan and China part of the same family? Is the DPP’s advocacy for independence dangerous? If the China issue were to be magically resolved, what would be left of the KMT’s and DPP’s platforms?
Lu Yeh-chung: Some people would say the KMT is a pro-China party, pro-unification party. But my take is that, as a perspective from academia, the KMT is the political party which knows China better. … So to me, I wouldn’t say there’s a huge difference between the KMT and DPP regarding Taiwan’s security. But how to achieve that goal — that’s the difference between the two major parties.
Lin Fei-fan: The shared culture between Taiwan and China is not a basic criterion for the two countries to think about their common future. There should be more discussion about not just identity, but also our way of life and the political system we believe in. … Both President Tsai and VP Lai have publicly raised again and again the importance of welcoming dialogue between Taiwan and China. But it should be without preconditions. That’s our stance.
Post-election, friend of the pod Kharis Templeman returned to discuss the results, the impending battle for Speaker of the Legislative Yuan, China’s surprisingly muted response to the election, and why Taiwan still has civil dialogue despite rampant partisan polarization.
And what would Taiwan coverage be without semiconductors? For that I have Arrian Ebrahimi and Michael Lu to thank. They discuss the strategic choices that Taiwan made to protect its semiconductor industry — its “sacred mountain of protection” 護台神山 — and how those choices differed from the US’s strategic choices in the CHIPS Act.
Media Diet
Lily Ottinger filed a fantastic meta-review of Netflix’s Three-Body Problem rendition. If you want to know what the mainland China internet is saying about the show, Lily’s review-of-the-reviews is indispensable.
The show’s opening scene, depicting the Cultural Revolution, has been largely scrubbed from the Chinese internet — but it’s exactly how Liu Cixin 刘慈欣 intended the series to begin. And one Chinese netizen reviewer liked its inclusion front and center:
The key to understanding The Three-Body Problem lies in understanding Ye Wenjie 叶文洁. Understanding Ye Wenjie is truly understanding human nature. If you downplay the crucial factor that led to her transformation, the entire story loses its meaning. Therefore, Netflix’s decision to place that segment of the story at the beginning is very appropriate.
But the most heated debate centers on the show’s casting choices — in particular, disappointment that Netflix race- and gender-swapped key characters. One reviewer rhetorically asked,
Does this predominantly white-led production team have elements of racial discrimination? … From The Joy Luck Club in the last century, to Mulan, Shang-Chi, and now The Three-Body Problem, it seems like all Chinese narratives in Hollywood only serve to amplify stereotypes.
Other critiques focus on the show’s simplified scientific explanations, added romance subplots, and fast pacing compared to the source material. As one reviewer puts it,
The true meaning of adaptation lies in daring to make bold changes to the structure and characters, engaging in reinterpretation. Otherwise, if AI can simply convert text into video, what else is there left to film?
I have no idea where Jordan finds the time to consume media like a full-time culture critic. At least he’s honest, admitting, “I read at least half of everything on this list.”
His top-rated books:
Team of Rivals, by Doris Kearns Goodwin: A fresh look at Lincoln’s political genius and personal growth.
The Rise and Fall of American Growth, by Robert Gordon: A gripping exploration of how life changed in the nineteenth and twentieth centuries.
What My Bones Know, by Stephanie Foo: A powerful memoir on healing from complex trauma.
War books:
John Ellis on trench warfare — Eye-Deep In Hell, and The Social History of the Machine Gun.
Naval history — Shattered Sword, by Anthony P. Tully and Jonathan Parshall; and Rules of the Game, by Andrew Gordon.
TV standouts:
The Knockout 狂飙: A Breaking Bad–esque tale of a fishmonger turned criminal in China’s go-go 2000s.
Become a Farmer 种地吧: twenty-somethings learning the ropes of farming life — and get blown away by mechanized agriculture advancements.
Chinese Bizarreries 中国奇谭: a fantastic animated series by Bilibili — eight independent stories rooted in traditional Chinese culture — exploring themes like technological fantasy and local nostalgia.
In the podcast world, Tyler Cowen’s chat with Rick Rubin and Sam Hammond’s profound discussion on AI and institutions topped the list. YouTube highlights included cheering on chess champ Ding Liren 丁立人 (and we still would love him on ChinaTalk!).
It doesn’t quite fit into a media diet — but why not add some NYC restaurant recommendations? Hao Noodle and Tengri Tagh Uyghur Cuisine take top spots.
And finally: how are Jordan’s odds in the Manifold betting markets panning out?
Will OpenAI, Google, or Anthropic have a model’s weights hacked before 2025?
Jordan set the market at 30%.
Market today: 27%.
Jordan set the market at 15%.
Market today: 35%.
Will China get mentioned more times than AI over the course of the 2024 presidential debates?
Jordan set the market at 60%.
Market today: 91%.
Will a SMIC 5-nm chip make it into production in a Huawei device in 2024?
Jordan set the market at 65%.
Market today: 27%.
Will a Politburo Standing Committee or State Council member get purged in 2024?
Jordan set the market at 25%.
Market today: 15%.
Thanks for sticking with us.