Jeffrey LewisHASC Language on Stockpile Management

The House Armed Services Committee defense authorization bill contains language that would repeal language creating the RRW program in favor of a “stockpile management program.”

I notice that the term of art is management, not modernization, although I wouldn’t object to latter word if the effort was confined to the purposes established in the language:

Section 3112—Stockpile Management Program

This section would strike section 4204a of the Atomic Energy Defense Act (50 U.S.C. 2524), which codifies the Reliable Replacement Warhead program. This section would also amend section 4204, which establishes the Stockpile Life Extension Program, with a new provision establishing a Stockpile Management Program. This section would establish that the objectives of the Stockpile Management Program are to: increase the reliability, safety, and security of the United States’ nuclear weapons stockpile; further reduce the likelihood of the resumption of underground nuclear weapons testing; achieve reductions in the future size of the nuclear weapons stockpile; reduce the risk of accidental detonation; and reduce the risk that an element of the stockpile could ever be used by a person or entity hostile to the United States, its vital interests, or allies.

This section would also provide guidelines for stockpile management, requiring that changes may only be made to the stockpile in pursuit of these identified objectives. This section would further require that any changes must be consistent with basic design parameters, and must use components that are well understood or are certifiable without the need to resume underground nuclear weapons testing. Additionally, this section would provide that any such changes shall adhere to the design, certification, and production expertise resident in the nuclear security complex to fulfill current mission requirements of the existing stockpile.

The Stockpile Management Program would support and complement
the science-based Stockpile Stewardship Program, which focuses on sustaining the scientific and technical expertise and the experimental tools and capabilities needed to ensure that the nuclear stockpile is safe, secure, and reliable without nuclear testing. The Stockpile Management Program, in turn, would provide a framework for the activities associated with actual work on the weapons that comprise the stockpile, including limitations on any changes to the stockpile.

It seems to me that this is a pretty sensible approach — consistent with my preference that the Congress ought to dispose of RRW on the very narrow grounds that WR1 is not the most cost effective or technically appropriate (ie lowest risk of testing) option to maintain (or manage) the capability provided by the W76. (And not on the more sweeping grounds that it is “new” or “modernized.”)

In case you are curious, here is the actual language.


  1. anon

    So, if we can design a new replacement WH for the W76 that is safer & more secure that does not require nuclear testing, is that allowable under this language? Sounds like it is to me.

  2. yousaf

    “So, if we can design a new replacement WH for the W76 that is safer & more secure that does not require nuclear testing, is that allowable under this language?”

    I’m unsure of the legalese, but it would be a bad idea :

    “Would you fly on an airliner that had never had a test flight, even though its aerodynamics may be well understood? So why would you — or more importantly our enemies — believe untested new weapons would work better than the tested ones we have?

    On the other hand, if the proposed RRWs are eventually tested, it will be more difficult to stop other adversarial nations from doing the same. Either way, the RRW program is detrimental to U.S. security vis-à-vis proliferation and deterrence calculus.”

  3. Andrew Tubbiolo (History)

    I wonder if we really can design a new warhead? Let’s say someone was in their 30’s in the 1980’s. They were surrounded by what must have been the most experienced design, fabrication, and operations cadre we have ever seen. If that person still works in the arms industry, they’d be approaching their 60’s now. Figure in retirement, death, and moving on to other fields, the lack of testing etc. I’m betting that the words nuclear arsenal is reducing to much lower levels than we imagine. Maybe history will show that the various test ban treaties were the most effective arms control measures ever taken. If so, then THAT’s good effective arms control.

    Why, because you still have a deterrent. In peacetime all sides still maintain a proper fear of general warfare. However if deterrence should break down and an exchange commences ….. A chain of major malfunctions might come as a welcome break in the flow of hostilities. … That is if the breakdowns were not favored by one side to a huge disproportion.

  4. Anon

    “Would you fly on an airliner that had never had a test flight, even though its aerodynamics may be well understood? So why would you — or more importantly our enemies — believe untested new weapons would work better than the tested ones we have?”

    Kind of depends on who’s telling me. When it comes to nuclear weapons I believe America’s best nuclear weapon designers – not those whose agenda is to do away with them. As for our enemies, if they opt to challenge America’s best nuclear weapon designer as to whether our untested weapons will work – well as Dirty Harry says: Make my day.

  5. Anon

    “I wonder if we really can design a new warhead?”

    Why not ask those who design them for a living?

  6. Major Lemon (History)

    Terms such as “new” or “modernized” are really passé.

  7. Stephen Young (History)

    In a new twist today, the House Energy and Water appropriators denied funding for the B61-12, the proposed Modification/LEP that both DOD and the NNSA were counting on to break ground in terms of what can be done under these programs. In short, they want to use some of the features developed for the RRW in the B61.

    The House is objecting based on their call for completion of the Strategic Posture Commission (check), the NPR, and the QDR before other major nuclear decisions are made. Once those are done, I think this position will reverse itself 180 degrees.

  8. yousaf


    I think the deterrent value of tested legacy warheads and untested RRWs will be about the same given the uncertainties inherent in psychological metrics such as “deterrent value”.

    There, however, are a number of conceptual — not technical — flaws with RRW which I have outlined previously.

    It is interesting that RRW advocates still want to hold on to tested weapons for decades to hedge their (psychological) uncertainties.

    The day I see a general who will accept an operational stockpile consisting of only untested weapons, I might revisit my view. Which military man will accept untested weapons? (Incidentally, a similar scenario already occurred in France. Under President Francois Mitterrand, the French accepted untested new nuclear warheads into their submarine-based stockpile based on the results of sophisticated computer simulations. But when Jacques Chirac became the country’s president, France conducted a handful of nuclear explosions to make sure that the weapons actually worked. The same thing will likely happen with RRW—at least no one in the U.S. government is offering any guarantees that it won’t.)

    As for the weapons experts’ views, Robert Peurifoy, the former vice president of technical support at Sandia National Laboratories has stated, ‘The present nuclear weapon stockpile contains eight or so nuclear weapon types. That population has enjoyed perhaps 100 successful yield tests. These weapons have benefited from a test base of perhaps 1,000 yield tests conducted during the 40 or so years when nuclear testing was allowed. Is DoD really willing to replace tested devices with untested devices?’

    The argument that RRW is needed to recruit and hone the skills of a new cadre of nuclear weaponeers is also disingenuous. For the most part, the design work has already been completed, rendering the remaining part of the RRW program largely an engineering exercise.

    Contrary to what proponents of untested new warheads assert, the more credible deterrent in the eyes of one’s adversary will always be the tested legacy weapons. On the other hand, if the proposed new warheads are eventually tested, this will make it more difficult to stop other nations from doing the same. Either way, the proposed RRW program to develop “new” but untested nuclear warheads is detrimental to U.S. security.

  9. Anon

    “Terms such as “new” or “modernized” are really passé.”

    I don’t agree that the terms “new” or “modernized” are passé but, our nuclear weapons stockpile certainly meets the definition of passé (e.g., past ones prime, outmoded, behind the times)

  10. yousaf


    I would argue that long as the legacy weapons work they are not passé, in any sense of the word. They do not need to be entered into the Milan Nuclear Fashion Show every year — they just need to go “bang!”.

    From 1958 to 1996, the Stockpile Evaluation Program sampled nearly 14,000 weapons; of these, only about 1.3 percent were found to have failures that would have prevented them from operating as intended.

    What would be the reliability of untested RRWs — and how would we ever know that number when they cannot be tested?

    According to the National Nuclear Security Administration (NNSA), the current U.S. warhead stockpile is safe, secure, and reliable. Its reliability — dominated mostly by the non-nuclear components (cables, valves, pads, cushions, foam supports, electronics, etc.) —is about 98 percent with a high confidence level. Keep in mind, too, that an “unreliable” warhead may still completely destroy a city, as the government considers a warhead with an expected yield of 90 percent or less of its design yield as — ahem — “unreliable”.

    The most egregious conceptual mistake in pro-RRW thinking is conflating technical (numerical) warhead reliability with its deterrent value. Because of the massive destruction potential of a nuclear weapon, an adversary’s deterrence calculus could hardly be different for, say, a 96-percent reliable nuclear weapon versus a 99-percent reliable weapon. In fact, it’s questionable whether it would even be different for a 25-percent reliable weapon versus a 99-percent reliable weapon. Switching perspective to the receiving end of a possible massive nuclear retaliatory attack makes the argument more acute: If a country is facing the prospect of several 300-kiloton U.S. warheads destroying its 25 most populous cities, it hardly matters that perhaps one of the weapons will have a suboptimal yield and doesn’t completely annihilate the twenty-fifth city.

    This deterrent logic is especially pertinent to the RRW debate considering that nuclear targeting practice places more than one weapon on any high-value target, making the probability of annihilation essentially 100 percent whether a single warhead has a 96-percent or 99-percent reliability.

    Ultimately, deterrence is an exercise in psychology, and small differences in the numerical reliability of overwhelmingly destructive nuclear weapons don’t have a measurable influence on any adversary’s deterrence calculus.

    Consider Chinese or Russian nuclear weapons. Do we know their technical reliability numbers? No. Yet, we still take them very seriously.

    If it ain’t broke — don’t fix it. Especially don’t fix it with something untested.

  11. Anon

    “If it ain’t broke — don’t fix it.”

    IHE vs. HMX ain’t about being broke – it’s about safety & security. Arguing reliability is tilting at windmills.

  12. kme


    As for our enemies, if they opt to challenge America’s best nuclear weapon designer as to whether our untested weapons will work – well as Dirty Harry says: Make my day.

    You misunderstand the point of the weapons. They’re there to stop the enemy from attacking you; not to blow him up when he does. If the enemy doesn’t take the weapon seriously and attacks, the weapon has already failed.

  13. George William Herbert (History)

    Yousaf –

    The rosy reliability numbers your sources give seem to ignore some fleet flaws that put large quantities of (now) known defective or known rapidly aging/degrading weapons in the field.

    I can believe a 1.3% “random defects” rate in inspections.

    What particularly worries me today is that the assumptions sets by both the RRW proponents and opponents tend to fall into the “We know everything we need to know” fallacy.

    The RRW proponents believe “We know everything we need to know to design this new weapon which is intrinsically more reliable.”

    The RRW opponents believe “We know everything we need to know to keep maintaining older weapons with no fear that creeping changes as we maintain will render them useless”.

    Both sides need to read the Challenger and Columbia accident reports and Taleb’s “The Black Swan”.

    The point about deterrent value of weapons being unrelated to reliability is somewhat overstated. The countervalue deterrent value degrades slowly with decreasing reliability – To exaggerate, 10 successful nuclear detonations over (insert target nation) will still be hugely deterring even if 100 were fired, and 100 detonations out of 1000 is still catastrophic national destruction of a large country.

    However, deterrence doesn’t simplify down to blowing up cities. There are plenty of other scenarios of use – a threat nation uses WMD against US or ally military forces in the field, say, but scrupulously avoids civilian targets. We could nuke their cities – but a measured response against military installations and forces in the field is less horrific. If you retaliate with one ICBM shot against a single airfield, and the result is that the warhead fails to detonate and digs a hole in the dirt between the runways, that really has a significant effect on further decisions, on both sides.

  14. yousaf
  15. yousaf


    I agree that there may be fleet flaws with the “weapons”: but the weapons are more than just the physics package. The reliability of the delivery systems is in many cases the controlling factor in the weapon system. As the recent missile silo fire shows, there is plenty of unreliability in the non-nuclear components of the weapon-systems.

    The point about the “Black Swan” is well-taken, but surely deflates the pro-untested-RRWs camp’s views than the pro-tested-legacy-weapons camp’s views. This is why we test-fly airliners, even though we understand their aerodynamics very well. I would rather fly on a legacy 747 than on some new untested Airbus under development (especially if fitted with French pitot tubes ca. 2005).

    Your scenario of the tactical use of nukes is interesting but I would argue that this would not be a mission for a 400kT 2-stage weapon — even if it was, just the primary going off would do more than dig a mere hole. (Blowing up an airfield can be done with conventional weapons just fine, in any case, so I think the scenario itself is a stretch).

    More importantly, deterrence would likely be unaffected unless your enemy somehow knew you were intending to send them one 400 kT nuke and that it yielded “merely” 50 kT — more than 3 Hiroshimas. Lastly, if it ever comes down to nuking another country, it will not be done with 1 nuke missile — as you know, targeting practice puts many nukes on any high-value target.

    You say that RRW opponents say that “We know everything we need to know to keep maintaining older weapons with no fear that creeping changes as we maintain will render them useless”.

    This is not true — at least, not for me.

    My view is that the use of nuclear weapons is in preventing their use thus we do not need to be anal-retentive about the reliability numbers of the warhead being 99.9999% — reliability numbers of ~98% (or even much less) are fine, especially since “reliability” is conservatively defined in government circles to be a yield 90% or more of the design yield. (e.g. Your enemy is not going to care if the yield is 400kT or 200kT in any practical situation).

    kme makes the point perfectly: “You misunderstand the point of the weapons. They’re there to stop the enemy from attacking you; not to blow him up when he does. If the enemy doesn’t take the weapon seriously and attacks, the weapon has already failed.”

    The enemy (if he is rational and understands basic engineering and has read the Black Swan) will always hold tested weapons in higher regard than untested weapons. (Our enemies surely take the word of our best weaponeers seriously, but they take an explosion more seriously).

    Of course, and this is important: there is no need to semaphore our reliability figures publicly, or for anyone to publicly denigrate our weapons (e.g. Chilton in WSJ). This is why I have argued that we ought to classify nuclear warhead reliability figures. Now is the time to do it. Washington has basically announced to the world that current U.S. warheads are 98-percent reliable with a high-confidence level and that their plutonium pits have a lifetime of at least 85 years. For the foreseeable—and likely even distant—future, this is a sufficient deterrent to any deterrable adversary. To keep it that way, future discussions of warhead reliability should be classified with strict oversight from truly independent experts with appropriate security clearances—i.e., the JASON group.

    Do we know the reliability figures of the Chinese warheads? Are we deterred?

  16. FSB

    GWH sez, “If you retaliate with one ICBM shot against a single airfield, and the result is that the warhead fails to detonate and digs a hole in the dirt between the runways…”

    This assumes that the nation being attacked somehow knows that the failure mode of the weapon was a failure in the nuclear package, whereas it may be any number of things (fuzing mech., failure of the ICBM, failur of altitude/timing electronics, etc., etc., etc). It also assumes the the nation being attacked somehow guessed you intended to nuke them.

    Only a highly stupid undeterrable nation will say “oh ALL the American warheads don’t work fersure fersure, let’s go for it….yoohoo!”

    Further, even if the warhead partially works — there are a continuum of yields — it will accomplish its aim. And, as mentioned above, the nation being attacked has no idea what you are sending their way anyway…how many kT? a nuke? a conventional weapon?

    This is especially the case now that we want to put conventional warheads on Global Strike ICBMs. The attacked nation will never know we really wanted to nuke them or not…of course, Global Strike has the downside that they may think you want to nuke them
    if they see an incoming ICBM and launch a strike against you, but that is another blog post….

    Anyway, the example is highly unlikely in the real world: the response to a WMD (non-nuke) strike against the US would likely result in overwhelming conventional retaliation (similar to 9/11 response).

    If the choice is only between tested legacy weapons and untested new weapons, the smarter course of action is to stick with the (highly reliable) tested weapons.

    BTW, the “random defects” rate in inspections (1.3%), is the same as what would surface in random selections from the stockpile for launch.

    Many of the so-called “defective” warheads in those 1.3% could still blow up a whole city — i.e. they may be “just” 50% of design yield. No small potatoes. 100 kT makes for a fine fish-fry.

    Let’s please stop obsessing over shiny new untested warheads that will reduce the deterrent value of our stockpile….

  17. George William Herbert (History)

    I don’t disagree with concerns about other parts of the system – ICBM and SLBM and cruise missiles aren’t going to be anywhere near 98.7% reliable, and RVs can burn up due to TPS failures (or hitting hail or rain on the way down), fuzes can fail, etc.

    And yes, a lot of random failures will lead to partial yield rather than full, but nowhere near nothing.

    However, a lot of failure modes will result in militarily useless yield – failure of a detonator to fire, mechanical failure of HE or the pit disrupting symmetrical implosion, failure of the boost gas injector or of the pulse neutron tubes (though, those can easily be multi-way redundant, and apparently are, so that’s less likely to be catastrophic).

    My ultimate concern is that public discussions on reliability are not really mature or well informed by actual reliability engineering standards. Particularly from the standpoint of someone who’s seen plenty of Black Swans in aerospace and IT reliablility in my career – I know better than to simplify to the degree that everyone tends to do. That sort of self-deceptive engineering and rhetoric has caused 2 fatal Shuttle accidents (see the Columbia and Challenger accident reports and the umpteen books about them), umpteen unmanned rocket and space probe failures (see Ralph Lorenz and David Harland’s “Space Systems Failures), and more IT systems outages than I can conveniently count.

    Understanding the scope of what needs to be considered, and laying it all out, much less doing the math to try and come up with reasonable reliability numbers and uncertainty estimates, is hard. I have good reason to believe that DOE / NNSA do this work, but it seems to stay inside the classification wall.

    The public statements they make are idiotic by actual reliability engineering standards – Black Swan / unknown unknown type failure modes are far more of a risk than anyone will publicly credit. This is not uncommon among any organization – many that have perfectly competent risk and failure analysis teams on the back end, which fully understand unknown risks and so forth, have people ignorant on the subject in higher management and PR roles. This is unfortunately fairly normal.

    It would be good if the overall depth of public independent reliability / risk analysis was better. I am working on it in my copious spare time, but this is not the place for it.

  18. FSB

    Engineering risk analysis is important and that of tested weapon design will always be more well-grounded than that of untested weapon design.

    That is fundamental engineering, and your enemy knows this.

    Before indulging in heavy math, I would simply ask 25 generals whether they would accept only untested weapons into their operational stockpile.

    It’s really not too complicated.