How Finding China’s Nuclear Sites Upset Pro-Beijing Trolls

This summer, my former colleague Decker Eveleth discovered that China was constructing more than 100 missile silos in the country’s remote interior. I had spent the previous summer working with Eveleth on mapping China’s nuclear forces for a project at the Middlebury Institute of International Studies and had encouraged him to take a second look. Both of us expected he would find something, but the sheer number was a jolt.

A week or so after Eveleth shared his findings, another researcher named Matt Korda reported finding a second field large enough to eventually accommodate another 100 or more silos. “I usually have to pay someone to do that,” the commander of U.S. Strategic Command said at a conference. “If you enjoy looking at commercial satellite imagery or stuff in China, can I suggest you keep looking?” (Another group has now found a third site that they describe as early construction for a third field of silos, although it is in a very early stage of construction and differs in some respects from the first two sites, which are extremely similar.)

Two hundred or 300 silos is, potentially, a pretty big development. After all, China is thought to have just about 100 nuclear weapons that can reach the United States—that’s up from just 18 when I was a graduate student two decades ago.

I started traveling to China in the early 2000s. If you happen to have flipped through a guidebook to China back then, you might well have come across an interesting story about an ethnic Hakka community in the province of Fujian. The Hakka there traditionally lived in communal roundhouses called tulou, and a lot of guidebooks told the same story about those houses, trying to lure the reader to take a long bus ride to a tourist site and cough up 50 yuan for entrance.

From the air, those roundhouses look like giant circles. So, the story goes, when U.S. intelligence analysts saw those circles in satellite images for the first time in the 1980s, they mistook them for missile silos. Ultimately—and this is how the locals claim they found out about it—the U.S. government had to dispatch a team to the area to investigate, only to realize their mistake but fall in love with the beauty of the place. (Hint, hint.) You can find this story in guidebooks like Lonely Planet or Rough Guide from the period, and it gets reprinted from time to time in state media like Beijing Review.

The first time I heard this story, I almost kind of believed it. It’s a good yarn. The problem is, it’s not true. The story was made up in a 1999 online Chinese magazine and went viral in China at about the same time I started traveling there. Local officials went along with it because it was good for tourism—and because it fit a national narrative of paranoid Americans misinterpreting a peaceful China.

It is, as I say, a good story. But even back in the mid-2000s, it was pretty clear to a lot of Chinese that the roundhouse story was ridiculous. One Chinese scholar pointed out a pretty obvious flaw: “in the 1980s, coastal cities had just implemented an opening-up policy, and the unopened mountainous villages did not allow foreigners to inspect at will.” He had published an academic article in an international architecture journal on those very same roundhouses in 1984. A Japanese architect read the article and was impressed enough that he wanted to come visit Fujian and see them for himself—but it took two or three years to get the visit cleared.

A different local scholar tracked down the authors of the original story, a father and his daughter who was in college at the time, as well as the local official to whom the story was attributed. The local official denied telling the story, while the father and daughter admitted that it was a completely fictional account. Another local, who was a big advocate for preserving the heritage of the area, pointed out that the story was an “unproven rumor and should not be hyped in tourism promotion.”

This vignette came back to me after Eveleth found those silos. Because today, unlike 20 years ago, I spend a lot of my time interpreting satellite images. And if you know anything about photo interpretation, you know the tulou story is ridiculous.

Interpreting a satellite image takes skill. It’s not just a case of saying, “Look at that round thing! Maybe it’s a missile silo?!”

You don’t have to take my word for it. The CIA has declassified many of the reports from the 1980s in which the U.S. intelligence community tracked the construction of missile silos in China. (And no, none of those silos are in Fujian.) The analysts already knew what a Chinese missile silo looked like. China built an experimental test silo at a place called Wuzhai and launched a missile out of it. When China began building identical silos elsewhere in China, the analysts spotted them immediately. (Also, a nitpick, but the silo covers in China at the time were square, not round.)

On a deeper level, the analysts weren’t just looking for the silos. They were looking for an entire supporting infrastructure: barracks for construction workers, launch control centers, and rail transfer points where specialized rail cars delivered the missiles. Photo interpretation is definitely a skill. That’s not to say that analysts didn’t make mistakes—you can see reports in which they revise previous assessments—but analysts don’t make the sort of childish error being fed to gullible tourists in this story.

We went through the same analytical processes ourselves in identifying the silo fields this year. Eveleth found the site with the exact same method that the CIA used: identifying a structure at a known missile location, in this case an air-supported structure used during construction at a place called Jilantai. He didn’t go hunting for round things. He went looking for new construction of a very unusual type. And once he found a candidate site, we looked for the supporting infrastructure, finding both a military base under construction and launch control centers with cable trenches connecting to multiple silo sites.

And, of course, we looked at all the stuff around the site. One thing both Eveleth and I noticed was that the silo field was west of a large wind farm. The big idea in photo interpretation is understanding what “normal” looks like in a specific place. Which objects in an image belong in this location, and which are out of place? We closely examined the wind farm, as well as other wind farms in China, just to be certain that we weren’t being fooled by a very weird looking renewable energy project. It didn’t take long to determine that the two kinds of sites look nothing alike.

All done, I started to package up the satellite images, annotating each one by adding explanatory labels. One of those images was a wide area shot showing the location of all 120 missile silos under construction in China’s Gansu province. I spent a fair amount of time worrying about that particular image. I wanted to make sure it showed the whole 700-square-mile silo field, as well the surrounding terrain. I spent more time than I’d like to admit going back and forth over what icon to use to mark the location of each silo. I tried yellow pushpins, but some of my colleagues hated that. Eventually, I settled on a black-and-white circle. And then I decided to add a few more annotations for context. I made sure to label the nearest prominent city, Yumen. And then, without much thought, I decided I would also label the Yumen’s Gansu Wind Farm.

It seemed like a reasonable thing to do. The wind farm is the most important piece of infrastructure in the area. We had looked at it carefully and concluded the silo field was not part of the wind farm, so what was the harm in labeling it? Why not say the silo field is west of Yumen, near a large wind farm? Boy was I wrong. Because pretty quickly, a high-follower Twitter user, a member of the cluster of far-leftists sometimes referred to pejoratively as “tankies” who back authoritarian regimes as long as they’re anti-American, took that image and tweeted:

“Come on this is just lazy. It says right there that that’s a windfarm lol.”

The tweet garnered more than 8,000 likes.

Suddenly, we were reliving the entire tulou silo business in real time. Literally: Multiple Twitter users replied to the tweet with pictures of the tulou, comparing the two stories. (Which is an apt comparison, although not in the way the posters imagined.)

Obviously, we pointed out that we were the ones who labeled the wind farm. That just made things weirder. My favorite response was “So you labelled it as a wind farm and assumed we would be gullible enough to not notice.” Yes, this was absolutely my plan: to label things so that the reader will not notice them. I am playing seven-dimensional go and you cannot comprehend my strategy.

When I started out studying China’s nuclear forces, I was really struck by the deep differences between how Chinese and American experts see them. I’ve written two books about China’s nuclear program and, for many years, taught a class where students compare declassified U.S. intelligence estimates on China’s nuclear program with historical Chinese sources covering the same developments. Instead of different perspectives on China’s nuclear program, it often seems like we’re discussing two totally different countries building the bomb—then and now.

So, I should have prepared for this discussion to get weird. And yet I was still surprised.

It’s easy enough to explain most of the blowback I faced on Twitter. If one scrolls through the accounts insisting that the silos are a wind farm, you will notice that many of the same accounts are also denying the ongoing repression of Uyghurs in Xinjiang. They consistently parrot the Chinese government line. Some of the accounts are Chinese state media or obviously part of bot networks; others are just the kind of deluded Westerners who cheered for Joseph Stalin and Mao Zedong back in the day. If the Chinese government admits the existence of the silos or boasts of the growth in its arsenal, they’ll adopt the new party line with equal vigor. I mean, the whole point of the term “tankie” is that their loyalty never wavers, not even when Moscow or Beijing sends tanks rolling over protesters.

The funny thing, though, is that the Twitter tankies assumed that Eveleth and I reported the existence of the silos as part of some bid to destroy U.S.-China relations. We published the story because it’s true—a rationale with which they may have little experience. But I also think it is important to have a public debate about how we reduce the nuclear danger between the two countries, which starts by knowing who has what nuclear weapons—and why. In the months since we published our findings, it has become clear that the U.S. government knew about the silos.

Interestingly, it seems that many people in the U.S. government don’t agree with the hypothesis that many of the silos may remain empty as part of a shell game similar to what the United States planned for the Peacekeeper missile. And I’ve long known, others don’t appreciate my suggestion that Washington’s long-term investments in nuclear modernization and missile defense bear some of the blame for China’s changing nuclear posture. I think China is trying to preserve the survivability of its nuclear deterrent in the face of massive U.S. investments in nuclear modernization and missile defense.

A lot of people in the U.S. government think it’s the other way around. They think that massive U.S. investments in these capabilities are needed because China’s is aggressively developing a capability for nuclear coercion. It’s very clear to me that the debate about what is motivating China to modernize its nuclear forces is different among academics than it is within the U.S. government. As far as I can tell, we academics have gotten some outside views a hearing inside the Biden administration, which I think is a good thing.

So, to the people defending China against our alleged slander, you’re not helping. Pretending China isn’t expanding its nuclear forces doesn’t get us closer to finding a way to work together to reduce nuclear dangers. Whether I am right or wrong that China’s changing nuclear posture is at least, in part, a reaction to U.S. choices, we urgently need to find a way to revive arms control discussions between the United States and China. By pretending the expansion isn’t happening, you’re just leaving the debate to the people who think China has planned all along to outrun the United States in a nuclear arms race. I don’t think that’s true, but if we can’t get a handle on the burgeoning arms competition underway, it will be.