Jeffrey LewisSo, like, why didn't it work?

One possible explanation for a failed test is that the North Koreans attempted to skip directly to an operational weapons design (say 1000-2000 lbs like the US Mk 7). Indeed, this was precisely what was reported in the press in 2003 based on changes in the nature of North Korea’s implosion testing at Yongdok—something I blogged about at the time:

So, could North Korea build a Mk 7-like warhead?

According to Chuck Hansen’s Swords of Armageddon, Los Alamos dramatically improved implosion techniques, permitting a six-fold weight reduction from the aptly named Fat Man to the Mk-7 –which Hansen estimates as 1,600 lbs with a 30” diameter.

Expertise in implosion techniques is clearly driving CIA concern about smaller North Korea warheads. David Sanger’s July 2003 article was prompted by the detection of an “advanced nuclear testing site in an area called Youngdoktong” in North Korea. What was special about that testing site? The revelation that “equipment has been set up to test conventional explosives that, when detonated, could compress a plutonium core and set off a compact nuclear explosion.” It stands to reason that the CIA saw the North Koreans working on implosion and inferred that they were trying to reduce the weight of their warhead designs.

So, if (and it is a big if) a couple of shacks at a test site prove that North Korea has become very proficient at implosion, then Pyongyang might be able to manufacture a warhead small enough to fit on a Taepo Dong 2—assuming the untested Taepo Dong 2 performs as advertised.

The warhead, however, would not be very reliable without nuclear testing. I doubt the North Koreans can run the table, building an ICBM and miniaturized warhead without testing either. At the very least, Pyongyang has to worry that doubts like mine will be widespread, severely undermining whatever existential deterrence Pyongyang enjoys.

Wow, I guess I was right. I had no idea.

There may be a parable here about authoritarian societies and proliferation. Kim Jong Il probably believed the weapon would work because, as a colleague suggested, “doubts about a system don’t always go up easily in the command chain” of such countries.

Indeed, what David Kay called a “vortex of corruption” was a persistent drag on Iraqi WMD programs, particularly before 1991. The Iraq Survey Group suggested that Iraqi WMD efforts were “largely subsumed into corrupt money-raising schemes by scientists skilled in the arts of lying and surviving in a fevered police state.”

That would rather economically explain the both the Taepodong’s dismal record, as well as the nuclear dud.

There is another school of thought that North Korea was attempting to stretch fissile material—at least one defector report a year or so ago suggested that North Korea built a 1 ton bomb with only 4 kg of plutonium and had little confidence in the weapon. At the time, I figured there was no reason for North Korea to stretch, but maybe they had problems with reprocessing or the quality of the plutonium in the spent fuel.

The possibilities that Kim Jong Il either approved a low confidence deisgn or didn’t know the bomb wouldn’t work raise interesting questions about the much (and I think unfairly) maligned intelligence community.

I mean, the IC can tell you that North Korea’s missile can fly a certain distance or be miniaturized a certain amount beyond which either will fail, but how can the IC expect to anticipate the decisions of foreign leaders who might have less technical information about their programs and no independent advice? Or who just do plain dumb stuff?

This can get wierd: Assuming North Korea was working on an operational design, had false but high confidence in the design and we knew it would not work—how could you communicate that to US policymakers who I believe would feel compelled to assume the device WOULD work?


So why, exactly, did Russian Defense Minister Sergei Lavrov announced that the North Korean nuclear test had a yield between 5 and 15 kilotons?

Here is the data from the Geophysical Survey of the Russian Academy of Sciences. The Mb estimate is 4.0, smaller than the US estimate.

A colleague of mine offers speculates the Russian figure—which came out really early—might have been what the North Koreans told the Russians, either before the test or immediately after.

If that is what happened, then we have some reason to think they were shooting for 5-15 kt and just blew it.

One last odd fact, 5-15 kt is about what NRDC expects a low or medium skilled nuclear power to get out of 4 kg of plutonium.

Do do do do … do do do do … do-dooooooooo!


I close this discourse about operational confidence by noting that the United States has built a missile defense that does not work, to defend against a North Korean missile that does not work, that would carry a nuclear warhead that does not work.

This is all very postmodern.


  1. Earl Kirkman (History)

    How very twilight-zonish, and you even left off the part about the president that was not elected.

    I would suspect that the Iranians are quite irked as well, since they seem to be the silent partner in all of this. Silly nork’s can’t fly a rocket or pop a bomb, what’s a theocrat gotta do to get some help around here???

  2. Andy (History)

    Good analysis, especially in regard to what you call the “vortex of corruption” in Iraq.

    I think there are several parallels here to Iraq. I think I’ve mentioned it here before, but the Iraqi Perspectives Project report published by the US Joint Forces Command shows the extent of the “vortex of corruption” in Iraq. The report demonstrates a common problem with closed and oppressive regimes: A Leadership that becomes increasingly insulated from negative information and news. Saddam’s reaction to anything that didn’t fit his worldview, or even remotely called into question a decision he made, had the effect of ensuring he received only positive reports from his advisors and the entire government apparatus (as an example, an Iraqi General was imprisoned for a year for suggesting US Tanks were superior to Iraqi tanks). Saddam’s lack of truthful information on the state of Iraq’s military and its capabilities led him, in part, to make terrible decisions. Meanwhile, he diverted scarce resources from the Iraqi military to support hopeless projects and wonder-weapons that never materialized. The bigger the promise, the more money the “project” received.

    I’ve frequently wondered if the North Korean leadership is similarly insulated affected by this phenomenon. You’re correct to consider it as a possible factor for repeated North Korean technical failures, including this latest one. It would not surprise me in the least if Kim’s underlings overstated the capabilities and chance for success. It’s even possible the failure was reported to the leadership as a success – this frequently happened in Iraq – and how would the leadership know the difference between success and failure in an underground test anyway?

    As far as the IC goes, you’re correct. In a regime like North Korea’s, it’s difficult at best to anticipate leadership decision-making. Iraq is another perfect example. Additionally, as was noted recently in this blog, determining timelines and capabilities for science and engineering programs in development is largely an educated guess, even with quality intelligence (which we don’t have on North Korean capabilities).

  3. J (History)

    I can’t find the exact reference, but I thought that the North Koreans not only gave the Chinese an sixty minute heads up that the test was coming, but also said the explosion would be in the four kiloton range. This was not an intentionally small blast, that’s for sure.

  4. AHM (History)

    Well, we used ~6 kg in the Trinity and Nagasaki devices; Albright, Berkhout, and Walker 1997 suggest that a first weapon by a new proliferator might be up to 8kg (p.306).

    My speculation (to add to the others) is that they may have skimped on the Pu due to their already limited supply; as I’ve argued before, they may have a lot less Pu than most calculations suggest, perhaps only enough for 3-4 devices. Moreover, if they used the Pu from the old 8K fuel rods stored from 1994-2002, these were relatively high burnup (although my BOTE calcs suggest that it’s still about 90% Pu-239), which may have contributed to the “fizzle.” Alternatively, if they used their new, lower-burnup batch they apparently recently (2005) extracted, then they just used up their best Pu on a relative fizzle.

    Finally, another explanation for the Russian estimate: As Geoff noted, the coupling between the explosion and the terrain can vary quite a lot; the Russians may know the terrain there better than any other non-North Korean country (I haven’t read any Chinese estimates yet, those might be interesting), and so may be using a radically different equation.

  5. John Field (History)

    I think I agree with this assessment.

    It seems to me that 1 kt is a rather large fizzle. You see, the neutron dynamics are 3-4 orders of magnitude faster than the compressional dynamics. Consequently, if you’re in a preignition situation, the temperature rises almost in concert with the rise past criticality. The yield is some multiple of the kinetic energy of the pit -e.g. weight of the bomb times some reasonable factor, not 1000 times.

    The only way out of that is if the first neutron happens just somewhat after criticality, but before final compression. Possible, but odds are lower.

    On the other hand, if you’re going for a high compression low mass warhead, the slightest screwup and a)the impedance match to the load will be poor; kinetic energy will be low and/or spalling may occur or b)fluid instabilities will make compression non uniform.

    As long as you’re not pushing too hard in the direction of low mass/high compression, these mistakes can lead to a low criticality and therefore small yield as we see.

    Not proof, but I think it is more pursuasive than the alternate hypothesis.

  6. fouro (History)

    Very postmodern. And Hu’s on first.

    great post, thanks

  7. Derek

    The DPRK is now the third developing nation to have its tests not live up to expectations. While the Indian and Pakistani tests in 1998 may not have been “duds”, they apprear to have produced yields well below what each country claimed. This suggests that engineering these weapons is more difficult than often claimed. In addition, it suggests that any Pakistani design support may not have been that useful to North Korea. As a result, there is no real reason to believe that the DPRK was necessarily testing an “advanced” design.

  8. Dick Durata (History)

    Questions:1. What would be the approximate yield of an MK-7 type device?2. How accurate is the .5kt estimate?3. If this was an attempt at a warhead designed for the TD-2, are we sure it was a failure?4. If feasible, what sort of damage would a .5kt device do in Tokyo or Seoul, including radiation?

  9. Bill Arnold (History)

    The Russian estimate might have been an upper bound given the possibility of a test in a large cavity.

    (I haven’t seen a direct translation.)

  10. John Field (History)

    After thinking about it, I believe my earlier comment is dead wrong. I don’t see a problem with a fizzle building up to this yield level. So, I guess I have no preferred explanation and I’m going to shut up now.

  11. Nathan

    Interesting. Have the Israeli’s tested theirs, do you know? Might there be an embaressing fizzle?

  12. liberal (History)

    “There may be a parable here about authoritarian societies and proliferation. Kim Jong Il probably believed the weapon would work because, as a colleague suggested, ‘doubts about a system don’t always go up easily in the command chain’ of such countries.”

    Compare and contrast to the US anti-ballistic missile defense system.

  13. Derek

    The Israeli’s probably tested a device in 1979, although this remains open to debate. Data collected at the time indicated a low yield. The data, however, is inclomplete. The Israeli program, however, had relatively wide access to other proven nuclear programs, including test data from the French and perhaps the United States. The French are also believed to have given Israel nuclear weapon design information. It is thus difficult to know whether the 1979 was a “fizzle” or an attempt to design low-yield weapon.

  14. Dan (History)

    Yes It was a DUD .. its too low a yield for a Nuke .. Right now those concerned are sweeping out the mine shaft of scattered plutonium as punishment for FAILURE. However expect more tests and eventually the REAL THING ..

  15. Yale

    Dan wrote:“Yes It was a DUD .. its too low a yield for a Nuke .. Right now those concerned are sweeping out the mine shaft of scattered plutonium as punishment for FAILURE. However expect more tests and eventually the REAL THING ..”

    No, it WASN’T a dud. 1000’s of atomic weapons that have been in the US arsenal were capable of sub-kiloton yields. (The test seems to have been a fizzle.. not a dud)


    This is an aerial image of Manhattan. I have added a ring showing the range of the lethal 500 REM radiation pulse from a 1/2 kiloton blast.

    The arrows point to the World Trade Center towers. (90% of the people in the towers survived the 9/11 attack. A NK “dud” would simply kill everyone)


    Back in 1973, the weapons designer Theodore Taylor, visiting the new towers, described what would occur if a 1 kiloton device detonated at the WTC:

    “Through free air, a kiloton bomb will send a lethal dose of immediate radiation up to half a mile. Or, up to a thousand feet, you’d be killed by projectiles. Anyone in an office facing the Trade Center would die. People in that building over there would get it in every conceivable way. Gamma rays would get them first. Next comes visible light. Next the neutrons. Then the air shock. Then missiles. Unvaporized concrete would go out of here at the speed of a rifle shot. A steel-and-concrete missile flux would go out one mile and would include in all maybe a tenth the weight of the building, about five thousand tons.”

  16. Peter Hayes (History)

    Dear Wonkers, I thought you might be interested in the following from our analysis of the test data (at )

    “The fact is that the DPRK is now a self-declared nuclear weapon state, but not an actual or demonstrated nuclear weapons state. This is not a domestic political problem for Kim Jong Il at this time. Indeed, on October 20, 2006, the leadership staged a “mass rally” in central Pyongyang to “welcome the historical successful nuclear test” and, as one gigantic placard stated, to” ardently congratulate the scientists, technicians, and workers who succeeded in a nuclear test.”

    But for the reasons outlined above, the other nuclear weapons states know the true state of affairs. Until the test, it was possible for the DPRK to employ the “Israeli model” of nuclear opacity as the basis for nuclear threat, whatever the purpose of having such a threat capacity, and to keep everyone guessing.

    Having tested and failed, the DPRK can no longer rely on opacity as the basis for having a credible nuclear force, at least sufficiently credible to threaten its adversaries with a nuclear explosion. The DPRK might believe that a half kilotonne “mininuke” still provides it with a measure of nuclear deterrence and compellence; but it could not rely on other nuclear weapons states to perceive it to have anything more than an unusable, unreliable and relatively small nuclear explosive device.

    In short, the DPRK has now demonstrated that it does not yet have a nuclear capacity that enables it to threaten nuclear Armageddon against anyone but itself.

    Therefore, although it could exploit the residual ambiguity that still shrouds its remaining capacity to deploy nuclear weapons and not test again, we judge it to be more likely that the DPRK will test again to assert the credibility of its nuclear arsenal and thereby, to truly join the ranks of the nuclear weapon states.

    The exact timing of the next test will determined by how non-technical factors such as “managing China’s response” and “picking up food aid from South Korea for the next winter” interact with the DPRK leadership’s perception of the need to “fix” the demonstrated non-capability from the first test. (14) This latter factor is also political and will be primarily a function of the DPRK leadership’s view on how to (not) use nuclear threat to compel the United States to engage it on terms that it finds acceptable, whether bilaterally, at resumed six-party talks, or at some other venue and time. Thus, via this last factor, the United States has continuing and unique ability to influence Pyongyang’s decision on when and if the DPRK conducts more nuclear tests.”

    In short, they now lack existential compellence, not deterrence as Jeffrey suggested in his opening discourse.

    Paranthetically, it’s true that the Russians know DPRK geology better than most—they surveyed a lot of the country in a joint exploration program looking for oil (they mostly found graphite). If I am not mis-recollecting, that data was up for sale by the North Koreans in the 90s—a couple of western oil firms were hired to reanalyze it. I doubt very much that it provided the basis for a different calculation of yield from seismic magnitude. Perhaps someone in Moscow with connections with MOD can cast light on why he was saying these things? They don’t normally curry favor in Pyongyang, so it had to have some basis, in some dimension. Of course, there’s another state that is proximate to the DPRK with EXCELLENT sensor arrays (designed to track the slightest pitter patter of little feet, tanks etc) at the DMZ; that’s South Korea and its partner, the United States.

    The ROK’s have an excellent understanding of Korean geology. So did SK publish any estimates separate to the US? AFP reports: “The activity measured 3.6 on the Richter scale, which could be caused by the explosion of the equivalent of 800 tonnes of dynamite, said Chi Heon-Cheol, head of the Korea Earthquake Research Centre.”

    I haven’t heard anything from US Forces Korea about those sensor arrays along the DMZ. Anyone?

Pin It on Pinterest