The Chinese Nuclear Tunnels: Case Closed

Oh, crap. We painted them white just to bury them?

The Strange Case of the Chinese Nuclear Tunnels is, for the time being, closed. Professor Philip Karber and his students at Georgetown (where I earned my doctorate, I am proud to add) have given their 363-page report to the Defense Department and the arms control community. It has landed with a thud.

Karber, you may remember from earlier posts here, was the origin of a weeks-long story on tunnels under China. The issue, actually, wasn’t the tunnels, which have been there since the 1950s and which were not a secret. The real question revolved around what was in them, and Karber and his students claimed they knew the answer: nuclear weapons.

Now, no one’s disputing that the Chinese might be building tunnels to protect their nuclear forces. The nuclear-owning states have a history of doing weird things, and the Chinese and the former Soviets have an even longer history of out-weirding everyone.

But Karber and his student team said that China could be using those tunnels not to protect existing nuclear forces, but to hide extra weapons that no one knows exist. The current Chinese arsenal is estimated to be somewhere around 300-400 weapons; Karber claims there could be three thousand more hiding down there.

Where did that number come from? Unfortunately, it seems to have come from huge assumptions, sloppy scholarship — including a basic fact-checking error that would not have escaped even a rookie book editor — and highly elastic interpretations of open source material like Google Earth.

The answer from the professional arms control and defense analysis community, apparently, is unanimous:


Actually, the answer is a little more pointed than that.

“China has not produced enough fissile material to produce 3,000 nuclear weapons,” said Hans Kristensen, director of the nuclear information project at the Federation of American Scientists.”Nor do they have delivery systems for so many weapons. It’s just inaccurate on all fronts, that estimate.”

Meanwhile, over at the Union of Concerned Scientists, China expert Gregory Kulacki was brutal, calling the scholarship “incompent,” “lazy,” and below the standards a high school would set for abusing Internet evidence:

This is not competent scholarship. There are thousands of claims on Internet blog posts, supposedly based on US government documents, that aliens from outer space were recovered in Roswell, New Mexico in 1947. Just because a lot of people repeat them in their personal blogs does not give them credibility.


I'm going to appear every time this story is mentioned.

Kulacki also suggested that the tunnels might encourage the Chinese to have fewer rather than more missiles in the belief that they were better protected, which is an explanation that makes far more sense to me. (What’s the gain or purpose of having an extra 3000 weapons if they’re essentially invisible? As Dr. Strangelove said, if you have a Doomsday Device, you’ve got to tell someone.)

The DoD reaction has been frosty, to say the least, probably because their analysts were led on a wild goose chase:

The report has been reviewed by the Defense Department but has not led to a revision of estimates on the size of China’s nuclear force, said a senior defense official, who spoke on condition of anonymity. “The findings of the report have been noted,” said the official. But “there is no change in our assessment.”

For those of you not conversant with government-speak, “noted” is what your superiors say when you raise something and are overruled, usually with irritation.

This was compounded by overselling the report before it was released. The advance story in The Washington Post was breathless:

The Chinese have called it their “Underground Great Wall” — a vast network of tunnels designed to hide their country’s increasingly sophisticated missile and nuclear arsenal. For the past three years, a small band of obsessively dedicated students at Georgetown University has called it something else: homework.

Led by their hard-charging professor, a former top Pentagon official, they have translated hundreds of documents, combed through satellite imagery, obtained restricted Chinese military documents and waded through hundreds of gigabytes of online data.

Man, I wish I could get press like that. (Hey, I’m “hard-charging,” too, just ask around.) But no matter how anyone might spin it, it’s a bad sign when the Federation of American Scientists, the Union of Concerned Scientists, and the Defense Department all conclude that your research was, um, sub-par, to say the least.

The Chinese Tunnel Mystery is, in my view, a classic example of what happens when people get wedded to a theory and then cherry-pick the evidence until it all fits.

Full disclosure: I don’t know Professor Karber personally and I don’t have a copy of the report yet, so I can only speculate from a distance and based on the reviews by Kristensen and Kulacki, two of the top experts in the world on nuclear weapons.

Dr. No: The "Behind The Music" Documentary

From here, it looks like Karber and his students developed the really cool “extra nukes” theory while poking around into the much more substantive question of the tunnels themselves, and then got so excited about it that they went on to find plenty of evidence by taking into account anything that could possibly confirm their theory, like Chinese blogs — and we all know how reliable blogs are — and even a fictional Chinese TV series.

All of this created a serious buzz earlier on, because no one had read the actual report. And that’s the other problem with controversial research: it all sounds exciting as hell if you just skip to the executive summary.

This is a constant hazard of research in any field. But it’s especially a problem when studying closed societies like China (or during the Cold War, the USSR), for obvious reasons. Karber is a veteran of those Cold War era efforts and that may have predisposed him to worst-casing the Chinese scenario.

But this is why people submit things to editors, journals, and peer reviewers: so that someone who isn’t in love with the theory can look at it more dispassionately. You’re supposed to work just as hard — harder, even — to disprove your own theory as you did to create it. (Trust me, your peers will if you won’t.)  That’s how scholarship works.

So it’s no surprise the study got hammered by the kinds of objections that would normally have been raised in the review process; those objections probably would have prevented the study from being published at all. A solid project review early on could have saved what was clearly thousands of student hours, lots of computer time, and some serious money invested in this whole fixation. That’s too bad, because I’d still like to know why the Chinese are acting like giant earthworms.

I have a particular burr in my saddle here about the involvement of the students. Working with students on research is a practice I really enjoy and have actively supported in my career, including taking the first initiative ever to establish a successful (and still operating) civilian graduate student internship program at the Naval War College many years ago. Working with your students is crucial to their training, and it sure helps to keep the professor on his toes.

But if the scholar in charge isn’t careful, it can turn into a hothouse, because the students have a vested interest in finding what the professor wants to find. Controversial or difficult research requires a hard-assed skeptic on the team; your own students simply cannot fulfill that role, which is normally one reserved for editors and peer-reviewers.

Karber, for his part, argues he succeeded by sparking debate on the subject. I disagree: it’s one thing to shed light, it’s another thing to sow analytical chaos with sloppy research. (Look at the climate change debate. At this point, I think the earth is now heating up from sheer bloviation.)

The blogosphere at work. Pass the burritos.

You also get some hyperbolic claims as the news spreads at the speed of blogging, with triumphant messages like “Georgetown students help uncover Chinese nuclear weapons.” (No, they didn’t.)

Of course, you also get some bat-shit crazy odd blog posts, like this one about how Georgetown’s government department is “shadowy.” Oy.

Worse, after all the hullaballoo, Karber admits that he really won’t stand by those colossal numbers:

“I don’t have the slightest idea how many nuclear weapons China really has, but neither does anyone else in the arms control community,” Karber told the [Washington] Post. “That’s the problem with China — no one really knows except them.”

Gee, thanks.

Karber's chart of the possible Chinese nuclear arsenal. We were only off by about 3,000.

But okay: let’s grant some uncertainty in the nuclear count. But seriously: three thousand hidden nuclear weapons? That would qualify as one of the greatest strategic deceptions in history. Karber and his students would have been taken a lot more seriously, I think, had they been more circumspect with their numbers — but let’s face it, circumspection does not get the attention of the media.

If the Chinese really had 3000 extra weapons, it would indict the entire U.S. national security establishment, with its billions of dollars of equipment and thousands of analysts, as being functionally stupider than just one guy and a bunch of kids at Georgetown with some Macs and a Google Earth account — to say nothing of the scads of Russian, British, French, Israeli and other nuclear watchdogs who all would have to be dumber than dirt to have missed this.

I suppose you can never go wrong underestimating the competence of a big bureaucracy like the DoD or the CIA, and I could be part of a legion of people with egg-drop soup on my face when Chinese paratroopers come ashore in Seattle while we’re cowering under a massive Chinese nuclear threat we didn’t see coming. But it’s just too hard to believe that only Karber and his students managed to notice a hidden arsenal 3/5 the size of of the entire U.S. nuclear force that was created, apparently, in between shifts of the people who were taking turns watching the Chinese.

Hopefully, this will all end up (as Kulacki noted) as a cautionary tale about research in the internet age.

Case closed — for now.

Tweet about this on TwitterShare on Facebook6Share on Google+1Share on LinkedIn1Email this to someone
Print Friendly


  1. I take it that they made no effort to correlate the tunnels with fissile material, which, from having taken your nuclear weapons class, one would know was the real story.
    It is possible that one could hide a lot of HEU. Plutonium, not likely, per the MASINT signature of reactors.
    HEU, one could in theory have a lot of baby “Natanzses,” like 50 centrifuges here and there, if that would still be a lot of nuclear material to move around.
    I don’t think though, no offense, that you are right about one thing; the logic of why China would do that.
    The logic would be to acquire a real MAD force, instead of a “Muss their Hair” force.
    Granted, like you said in class, an American simulation in which there are 40 million casualties isn’t a “Go” sign. However, if there are 30 Chinese missiles at large megatonages targeted on basically the same 5-10 cities, which one could infer from Chinese Second Artillery Doctrine, “they won’t trade Los Angeles for Taipei,” then in a crisis, that might draw U.S. preemtption on missile defense and a Trident and especially B-2 strike from China’s point of view. In that simulation, it might be that at that .99 level, no zeros here, and at the .999 level, maximum of three zeros here, and the Chinese are gone. That would make me nervous if I were them, and so, to build out of that, I wouldn’t want to do something that some “hard charging Brigadier at STRATCOM” might not like; hence the secrecy. All of this is a very serious argument for China being part of START-like arrangements, from their point of view and ours, as the secrecy level and intensive burrowing activities lends itself to like you say, “cherry picking.” Of course, one lesson of this, independently, is that if we in the West like plutonium for devices, that isn’t actually necessarily what other people would do, since the path’s constraints are different. I would still like to have weapons inspectors and a head count so to speak with Chinese nuclear weapons. Also, the other reason it actually does matter as to numbers, is that if there was a successful attack on Command, Control, Communications, Computers, and Intelligence, leave out the RS part of C4RSI, that made it impossible for us to fire anything, assuming Tridents were taken out with tactical nuclear weapons per the Triomphe “collision tracking hypothesis,” then you have an issue. You make by the way a very good point as to the risks of student research, as they are too committed in some circumstances to assuming the Professor is a petty god, and so won’t see how there is groupthink.

  2. Don – I think the fissile material issue was Kristensen’s point, in that we may not be able to figure out exactly how many nuclear weapons there are, but that we certainly can say that the Chinese did not squirrel away fissile material for 3000 extra bombs. With that said, I have not read the Karber report, so I don’t know if they tried to correlate fissile material to their estimate — but it sure sounds, from the reviews, like they didn’t.

  3. Thanks for updating this story, Tom. Time to revisit the terra cotta army theory? Joking … seriously the real shame would be that the overheated media reporting of questionable analysis leads some to assert (again) that there’s no value to open source analysis, and that the IC has the monopoly on all the worthy (classified) info. In the case of closed societies (USSR in Cold War), combing through open source literature such as military and scientific journals gives valuable insight into strategic thought and direction and technical innovation. I know the NWC does the same today with Chinese military literature. Google Earth has made imagery analysts of us all, but using publicly available information must be done with great care and checked against a multitude of theories and sources, like all good all-source analysis should.

  4. Moneypenny: I think that’s exactly where this report does a lot of damage. It really makes the people working with open sources look kind of nutty, which is unfortunate. I know from my days as a working Sovietologist that we academics weren’t all that often dealing with anything different from what the CIA was reading — like the daily issue of Pravda — but we at least did talk to each other and try to triangulate a bit from more sources than just stuff we got our hands on. I don’t think that’s a good research model for the students, either. The news stories report that one kid spent years translating Chinese documents, but I immediately wondered: okay, so who’s teaching him how to read Chinese sources in their context? I learned that one-on-one with my mentor at Georgetown (Thane Gustafson, PBUH), and no one would have thought to hand me a bunch of Soviet documents and say “okay, what do you think?” This was not a good outcome, and it could have been avoided — and worse, we still don’t know what those damn tunnels are for.

  5. So you’re saying … we’re quite far removed from the days of RAND experts like Nathan Leites working on operational code and the analysis of the esoteric communications of the Soviet leaders? As you note, there’s a world of difference between translating material and providing meaning and context to the motivations and intentions behind it all. Who says there’s nothing to learn from Cold War history?

  6. I’m saying even worse than that. (And Leites…ack. That wasn’t exactly the way to go either.) We’re far removed from the days of any kind of cultural or political context for analysis. How many of those kids reading Chinese journals have read Marx, or Lenin, or Mao? Can they spot boilerplate when they see it? Do they know when a term is being used differently than in the past? There’s no substitute for deep regional knowledge coupled to experience. Cherry-picking data and then making assumptions based on the putative mileage of tunnels is the kind of thing that it took us years to stop doing in the analysis of ChiCom and Sov intentions.

  7. That would be such a basic error if they didn’t think about fissile material, although, one can wonder about the prominence of HEU is bomb designs in the literature, where Is aw a physicist once with the Norks, and he didn’t get why they would do that, and I was like “So we don’t bomb them, silly goose. You are thinking about a different set of technical issues, not like them.” That “not thinking like them” is the case for “Thick description” as having a place like you say, and, is one of the potential fallacies of nuclear deterrence optimism as to ” What we have here is a failure to communicate.” It still seems thought that maybe there is an opportunity all this to bring China into the START mechanism, as the Russians wouldn’t like it if China had a lot more nukes one would think, and the cascade effects are large; India, Japan, ROK at first order, and then with India, you have the Pakis at the second order, and then Iran has its argument, “I see 44 Magnums all over this room,” and then the Arabs. They are interlinked somewhat, if you can get the Chinese to accept the full duopoly on pure Assured Destruction capability, if, that is playing somewhat into Ivan’s view of how the world should work too.

  8. The “published” fissile material that China discloses does not have any credibility. Just as the Soviet Union’s didn’t by the end of the Cold War, which of course the Federation of American PseudoScientists also bought into…and made the exact same mistake on their calculations which the DOD bought into. When the Soviet Union collapsed and they were needing help rounding up all their warheads to secure them…the U.S. DOD offered them what they were told by the FAS was more than adequate… 20,000 mobile storage units. The Soviet General told them they needed at least 40,000….
    Has the FAS ever apologized for being wishful thinkers …instead of suspicious of the foreign threats? Not to my knowledge.
    Hypothesizing that the Chinese have already accumulated 3,000 nuclear weapons is not unrealistic or undocumentable. The existing “public disclosed” amount of fissile material does not disclose what has already been deployed. That is highly secretive….and was not revealed by China. And the only legitimate basis for the secret underground tunnel systems…exceeding 3,000 miles…which has not and cannot be discredited… is obviously not for deterrence. You are only deterred by a credibile number of weapons. Ambiguity can have some effect, but not much. And the Chinese only revealed the tunnels AFTER they were ratted out by the earthquake. The realistic assumption is that this was a surprise first strike force. Pearl Harbor style. Catching the gullible and complacently smug westerners, particularly the FAS, off-guard. Again.

    • First, I think it’s absurd to worry about a “nuclear Pearl Harbor” from the Chinese or anyone else. The Soviet Union couldn’t have pulled it off at the height if its power, and I don’t think we need to start worrying about it now. Second, short of generically bashing the FAS, what evidence is there that Karber is even close to correct? It’s one thing to say we’re not sure how many nukes China has; it’s another to say they might have three thousand hiding in tunnels.

  9. There are a lot of reasons for China to expand its nuclear force, us missile defense is progressing and 2 of its programs (GMD and sm-3 based) show a lot of promise. Tunnel entrance points can be found on radar and hit so as to close them and stop launches. The us has a policy of preemptive strike which leads me more to believe we have entered an era of the us will strike when and where and expect no return. If the use used just its ICBM forces in China it would more then likely destroy the vast majority if Chinese ICBMs, whatever survives can be hit by missile defense. I offer you the idea that missile defense was never planned for mad defense but as part of a total preemptive strike. If you were China and you saw this happening how would you respond. 2 ballistic missile subs ( which do not patrol the pacific) are out of range and lacking in testing. Chinas response to this would be expanding its sub force( its doing this quickly) and by expanding ICBM forces. This is the only way for them to avoid a preemptive attack.

    • The efforts, programs, and strategy of the dod would be very threatening to China. A nuclear pearl harbor is unlikely but in chinas eyes a Us strike with impunity would not be.

    • If you genuinely believe that GMD will ever stop a Chinese ICBM attack, you’ve been reading too many Chinese (and Russian) newspapers. The United States does not have a policy of “preemptive strike,” although I think you mean “preventive” — and it doesn’t have that policy, either. The idea that missile defenses could be a shield under which the U.S. would then obliterate the Chinese deterrent is something the Soviets used to say about SDI back in the 1980s; it was propaganda nonsense then and it’s nonsense today.

      The only thing that would placate China is complete U.S. unilateral disarmament and removing all U.S. forces from the Pacific. That’s not going to happen, so I don’t think any of us should lose too much sleep over what the Chinese want, since — at least in military affairs — they’re a regime detached from reality and still living somewhere in the Cold War. Of course, since they’re still a brutal one-party dictatorship, it’s completely understandable that they fear the intentions of the United States, but there’s not much we can do about that.

  10. I didn’t say that us missile defense was designed to stop an all out nuclear attack. I referenced what the missile defense agency commonly says it is designed for, a limited strike or rogue missile.
    Now if this was after a major us strike then the counter attack would be very limited. Here missile defense could play a serious role. Ballistic stubs were created to guarentee a successful counter-strike. This would remove the idea from the soviets that a first strike would totally disarm the us. Like you yourself said China is living in the cold war. The Us for is obliged to follow. Now I enjoy playing chess and the strategy involved in it. If I was China and I saw the united states with a large ICBM and SLBM force it would not be unreasonable to think that IF the us were to strike first they could destroy most of chinas nuclear forces. Now maybe this isn’t a huge problem, if even 20 missiles survive that would cause severe damage. Then the variable of missile defense is added, and might I remind you that actual capabilities are classified. My response would be to design a more resilient and robust nuclear force. How would you respond.

    I will copy and paste to save you time.

    Arguments for preemptive war made during the Bush administration

    Sofaer’s four elements

    The scholar Abraham D. Sofaer identified four key elements for justification of preemption:[28]

    The nature and magnitude of the threat involved;
    The likelihood that the threat will be realized unless preemptive action is taken;
    The availability and exhaustion of alternatives to using force; and
    Whether using preemptive force is consistent with the terms and purposes of the U.N. Charter and other applicable international agreements.

    I am confused as to why you don’t believe the us has a policy of preemptive war considering we just finished one. The Iraq war was started over the idea that an enemy had wmd, and was attacked due to the alleged threat to national security. the world and those who seek to attack the united states are put on notice that if it is thought that you could cause serious harm to national security the us will negate that

    • Thanks for the time-saving cut and paste. Just as a helpful suggestion, Wikipedia usually isn’t the last word on matters of international law or U.S. policy.

      But since you seem confused, allow me to help. I think I have a pretty good grasp on this, since I wrote an entire book on this subject which you can find here:

      Nichols Eve of Destruction, Univ. of Pennsylvania Press, 2008

      The war in Iraq was a preventive war, which is even worse than preemptive war, but that’s a different question than whether the United States has a policy of nuclear preventive attack against China. The fact of the matter is that the Chinese have claimed that the United States seeks to harm the PRC’s interests when it builds up nuclear weapons, and also when the United States reduces its nuclear arsenal (which Beijing has claimed is an attempt to embarrass China into following suit). There is no real possibility for a rational discussion, especially where the Chinese military are concerned, because in international affairs the PRC has the temperament of an irrational teenager who takes offense at everything. The regime is pathologically insecure, and until it mellows — or, like the USSR, collapses — it will view everything as a threat, as authoritarian governments always do.

  12. My apologies for posting a wikipedia link, I was under the impression that since you said there was no us policy on preventive strikes that they didn’t exist. What you must have meant was even though there isn’t a policy they occur anyways. Which is odd because that would have no effect on my argument because the idea is whether or not China fears them, not if the us has no policy but does it anyways. So now that we’ve wasted sometime and returned to my original point what would your response be if you were in chinas shoes.

  13. So now following this we have, like you said, and irrational China. A united states with massive nuclear arsenal. A congress that very recently ordered the us to make plans to destroy these tunnels If need be. A progressing missile defense system. How naive would it be to believe that the Chinese are not worried about maintaining parity. A parity which like I said could be disarmed by a first strike combined with a limited missile defense system. Yet again I have to reference missile submarines as being introduced to create a successful counter strike therefor maintaining MAD. Something which China barely has as it is, and certainly won’t if missile defense is actually capable of handling a limited strike.

  14. I appreciate your earlier reference to the soviet union saying that SDI would allow the Us to strike with impunity. You will note that they made efforts to combat this, much as I would expect China too.

    • I believe what I said, exactly, was that those Soviet fears were “nonsense,” as they are in China’s case. If you’d like to make excuses for the Chinese nuclear program, that is your prerogative, but your reasoning makes even less sense than theirs. You need to do a little homework before wading any deeper into this. (And by homework I don’t mean Googling “nuclear weapons.”)

  15. It doesn’t really matter if the fears are ground less does it, it matters how they respond to these fears.

  16. I appreciate you not being rude in your responses and carefully sidestepping the actual point of this exercise, which was whether or not they would build more robust nuclear weapons, not if you felt they needed to.