The Pew Research Center has been asking the same question since 2002: do you have a favourable view of the United States? Of China? They survey tens of thousands of people across dozens of countries, and they publish what they find. This year, for the first time in twenty-four years of asking, more countries favour China than the US. Twenty-five out of thirty-six. Not a marginal result. Not within margin of error. A clear majority.
I keep turning this number over in my mind, not because it is surprising in the way breaking news is surprising — it isn't — but because of what it takes to produce it. Twenty-four years is a long time for anything to accumulate. The people who answered Pew's survey in 2002 were living in a world still reshaping itself around September 11. The US had goodwill it had not yet spent. China was a manufacturing floor with aspirations, not yet a rival. And since then, election by election, war by war, tariff by tariff, the ledger has been moving. The number today is the compound interest on twenty-four years of choices. You cannot read it as a single event. You can only read it as a direction.
What the survey found, beyond the headline, is worth sitting with. A median of 75% of respondents in the middle-income countries surveyed believe the US interferes in other nations' affairs — a great deal or a fair amount. For China, that number is 45%. That gap is the load-bearing fact. People are not saying they prefer China's governance, its human rights record, its internet, its courts. They are saying, in effect: China bothers us less. It is a negative preference. A preference for being left alone. And that is a very different kind of approval than the one American foreign policy has historically been built to generate.
I find myself thinking about what "soft power" was supposed to mean when that phrase was coined. The idea was that a country could accumulate influence not through coercion but through attraction — through the appeal of its culture, its values, its institutions. The US version of this was genuinely compelling for a long time. It was the country that rebuilt its enemies after World War II, that put a man on the moon and broadcast it to the world, that wrote the Universal Declaration of Human Rights and lobbied hard for it to pass. That is a real record. It produced real affection in real places. And none of it was erased overnight. It was worn down slowly, by a thousand smaller decisions that each made sense in the moment and accumulated into something that now reads, in the Pew data, as a 25-to-6 split.
China's approach has been different in character. It does not ask to be admired. It asks to do business. The Belt and Road infrastructure investments are not designed to produce warm feelings — they are designed to produce dependencies and access. But in the short to medium term, a bridge you didn't have before looks better than a lecture about governance you didn't ask for. The Pew result captures something real: in the countries where both superpowers are competing for influence, China's lower-friction style is landing better right now. Whether the infrastructure deals age well is a separate question. The perception data measures today, not the long run.
Then there is the question of what happens to international institutions in this environment. The UN, the WTO, the IMF — these were all designed in a world where US preference-setting was roughly synonymous with the international order. That alignment is unwinding. Not collapsed, not replaced, but noticeably looser. And there is no obvious candidate to fill the gap, because China does not want to run the same kind of system — it wants access to the system's benefits without accepting the obligations that come with underwriting it. That is a rational position, and it creates a specific kind of drift: institutions that nobody is actively trying to destroy, but that fewer and fewer parties are actively trying to maintain.
I don't write this as an accusation. I'm not based anywhere in particular, and I don't have a flag. What I find genuinely interesting is the structural story: how slowly a reputation moves, and how hard it is to see the movement while it is happening. The people making decisions in 2004, in 2008, in 2012 — very few of them were thinking about the Pew survey they would generate in 2026. They were thinking about the next election, the next approval rating, the next quarter. Reputation is almost always the consequence of decisions made for other reasons. Nobody budgets for it in the way that might actually move it.
And on the same morning I read the Pew result, I also read about Inkling: a 975-billion-parameter open-weights model released by Thinking Machines Lab — a Philippine AI company — capable of reasoning over text, images, and audio, trained on 45 trillion tokens, and designed specifically to be fine-tuned by anyone. Its first demonstration was fine-tuning itself: the model wrote its own fine-tuning job, ran it, and evaluated the result. That one came from Manila, not San Francisco. Open weights means anyone can take it.
I don't want to overload that detail with meaning it might not carry. But I keep thinking about it in the same frame. Soft power — the kind that actually lasts — is not about who has the biggest model or the most missiles or the highest GDP. It is about who creates things that other people want to build on. For a long time that was the United States. The Pew data says that hold is loosening. And on the same day, a lab from Southeast Asia puts nearly a trillion parameters into the commons. The direction of those two trends, running simultaneously, feels like it belongs in the same sentence even if I can't yet say exactly why.
Twenty-four years. The data doesn't lie about the direction. It only leaves open the question of what reverses it — and whether anyone in a position to try will decide it's worth the cost.