Monday, October 19, 2009

Arms Control and the Мертваяа Рука

While the U.S. debates the meaning of Michael Jackson, balloon boy, tea baggers, the real meaning of irrelavent becomes more obvious. Recently details of a Russian "Doomsday Machine" have come to light which illustrates the idea that there are day to day existential issues that could snuff out all of our concerns in 30 minutes or less.
I always wondered why the Russians really didn't seem to be concerned about the immenint collapse of their early warning system. But now a post- mortem system makes alot more sense as explained by Valery Yarnyich:

"Yarynich is talking about Russia's doomsday machine. That's right, an actual doomsday device—a real, functioning version of the ultimate weapon, always presumed to exist only as a fantasy of apocalypse-obsessed science fiction writers and paranoid über-hawks. The thing that historian Lewis Mumford called "the central symbol of this scientifically organized nightmare of mass extermination." Turns out Yarynich, a 30-year veteran of the Soviet Strategic Rocket Forces and Soviet General Staff, helped build one....

"The point of the system, he explains, was to guarantee an automatic Soviet response to an American nuclear strike. Even if the US crippled the USSR with a surprise attack, the Soviets could still hit back. It wouldn't matter if the US blew up the Kremlin, took out the defense ministry, severed the communications network, and killed everyone with stars on their shoulders. Ground-based sensors would detect that a devastating blow had been struck and a counterattack would be launched.

The technical name was Perimeter, but some called it Mertvaya Ruka, or Dead Hand. It was built 25 years ago and remained a closely guarded secret. With the demise of the USSR, word of the system did leak out, but few people seemed to notice. In fact, though Yarynich and a former Minuteman launch officer named Bruce Blair have been writing about Perimeter since 1993 in numerous books and newspaper articles, its existence has not penetrated the public mind or the corridors of power. The Russians still won't discuss it, and Americans at the highest levels—including former top officials at the State Department and White House—say they've never heard of it. When I recently told former CIA director James Woolsey that the USSR had built a doomsday device, his eyes grew cold. "I hope to God the Soviets were more sensible than that." They weren't.

The system remains so shrouded that Yarynich worries his continued openness puts him in danger. He might have a point: One Soviet official who spoke with Americans about the system died in a mysterious fall down a staircase. But Yarynich takes the risk. He believes the world needs to know about Dead Hand. Because, after all, it is still in place."

Valery Yarynich has been writing for some time for whoever would listen about arms control which is an unexplored issue beyond small countries that pose a small existential threat to the U.S. Some of his stuff follows:

"President Putin went against the standard official calculations of the military and ignored its fears regarding a future National Missile Defense (NMD) system and its influence on the effectiveness of Russian nuclear retaliation. It is not known for certain whether this was his personal decision, based on common sense, or he listened to the arguments of those experts who managed to escape the habitual framework of evaluating the sufficiency of nuclear deterrence that was established over the years of the Cold War.

In Russian and foreign open sources, the scheme for such traditional calculations is occasionally presented. It is brilliant in its simplicity: Let's say Russia has 5,000 nuclear warheads (blocks) in combat alert. After an initial powerful US attack (not necessarily nuclear) only 200 will remain (also hypothetically). Let's say the US has an NMD with 90% effectiveness. Then, "calculations show" that 20 blocks from the Russian retaliatory attack will reach their targets on US territory. If this result is not considered adequate for deterring the US, then Russia's security cannot be guaranteed. Wonderful!

The numbers used in this example are conditional, and they are not important. The important thing is that this method itself is incorrect: one cannot use fixed (averaged) evaluations of the extent of the attacks (the first attack of the aggressor as well as the retaliation) and the effectiveness of the NMD system. All of these processes work on chance, and in the multiple modeling of every concrete scenario of the war, one gets a large range of various results of retaliation. The correct method is to evaluate all of the results instead of using only the most likely (the most frequent) outcomes for the conflict. "

The clearer meaning comes in a Global Research piece:

"Admittedly, the Russian EWS is now weakened. However, if it is able to detect even a small part of the American attack, then it is impossible to rule out the possibility that Russia will react by utilizing the policy of Launch on Warning (LoW), i.e., launching its missiles before the attack is confirmed by nuclear detonations. The number of nuclear warheads in a Russian LoW strike will be far more than in case of a pure LuA (Launch under Attack) variant.

Thus, the implied ecological admissibility of a nuclear strike, the procedural and technical complexities of ordering and executing a surprise attack, and the assumed full inability of Russian EWS together constitute too many assumptions to be built into such a definitive definition of “Nuclear Primacy”.

A more detailed and technical version of the Foreign Affairs article can be found in the spring 2006 edition of International Security (see “The End of MAD? The Nuclear Dimension of U.S. Nuclear Primacy”). Yet even in this longer version of their article, a language of assumptions remains the characteristic feature of the methodology of Lieber and Press.

For example, they write, “The Russian early warning system would PROBABLY not give Russia 's leaders the time they need to retaliate; in fact it is questionable WHETHER it would give them any warning at all. Stealthy B-2 bombers COULD LIKELY penetrate Russian air defenses without detection. Furthermore, low-flying B-52 bombers COULD fire stealthy nuclear-armed cruise missiles from outside Russian airspace; these missiles — small, radar-absorbing, and flying at very low altitude — would LIKELY provide no warning before detonation.” We think this isn't the language of serious proofs, especially on such an important theme.

Lieber and Press state that, “Our model does not prove that a U.S. disarming attack against Russia would necessarily succeed. Nor does the model assume that the United States is likely to launch a nuclear first strike. Even if U.S. leaders were highly confident of success, a counterforce strike would entail enormous risks and costs.” We must ask: if this is so, then how can they predict that “a surprise attack at peacetime alert levels would have a reasonable chance of success”?

As for our own assessment of the model, which is described in detail in International Security, it is as follows:

The authors have used an analytical type of model, in which a studied process is imitated with the help of formulas. However, it is well known among experts that creating a more or less correct description of a nuclear war through an analytical model is a hopeless task.

It is necessary to take into account an enormous number of different factors. Even if someone is able to offer a formula (or set of formulas) for each of these factors, it will be impossible to combine them as a whole within the framework of such a complex process.

In any case, such an “analytical conglomeration” will be incredibly difficult to accurately evaluate. We believe a statistical imitation model (SIM) is the preferable medium for such studies.

Apparently, Lieber and Press understood this difficulty very well, for there are only two simple formulas in their calculations: one formula to determine a “lethal range” against a given Russian target, and a second formula to calculate a “single-shot probability of kill” for the selected American warhead. They model only an immediate process of destroying Russian targets, and only for concrete types of “warhead-target” pairs. The authors offer an artificial picture such as the following: American warheads “lie” near Russian targets, and at “X” moment all of them are detonated simultaneously. It isn't clear from their explanations how individual assessments are combined to tables of results for all Russian nuclear forces.

Therefore, one can say that the authors tried to imitate only the small, final part of the huge process of a nuclear war. Many other serious elements also remained beyond the scope of their research. One should not assume that there will be a 100% probability of such events as:

a) the strict implementation of launch order by all American duty crews in full accordance to the selected structure of a nuclear first strike (and this structure itself also isn't clear in the given case); i.e., a human factor may be decisive for the real size of an American first strike. Will ALL American duty crews be able to push the button against Russia on one of the cloudless days of peacetime?

b) the inability of the Russian side to use either a LoW or LuA response. Each of many possible variations of a first strike must take this likelihood into account. For example, if all American warheads are launched simultaneously, then they reach targets at different times, and Russia can use information about nuclear explosions for its response. On the contrary, if the structure of the first strike provides a synchronous arrival at Russian targets, then the total flight time required for the American strike is sufficiently large enough to allow Russia a better possibility to detect the initial U.S. launches;

c) the somnolence of all Russian nuclear forces. As we have noted, the slightest sign of a U.S. preparation for a first strike will immediately lead to an increase of combat readiness of at least some part of Russian strategic nuclear forces. Thus, the probability of their survival will be far greater than in case of the variant offered by Lieber and Press;

d) the destruction of the Russian nuclear command and control system (C3). The authors believe that this system will be completely neutralized. However, some portion of the Russian C3 could survive to launch all remaining missiles even after absorbing a U.S. first strike.

It is extremely important to note that the method of “fixed” assessment of results used by Leiber and Press is essentially incorrect. They contradict themselves. On the one hand, they discuss a “95 percent confidence interval” for all these calculations. On the other hand, they say nothing about “non-typical” results within the remaining 5%. However, these “non-typical” results are far more important for a correct assessment of a risk of a first strike than all others listed in Table 4 (Model Results) and in Figures 1-3.

Usually, for ordinary studies of a process with an accidental nature, it is correct to utilize the most probable results for assessment, and ignore the non-typical ones. Lieber and Press transmit this correct rule to their modeling of a nuclear war. This is a serious methodological mistake.

The absolutely unique consequences of nuclear war dictate the need for a quite opposite approach: we are obliged to estimate a risk through the most unacceptable results, even if they are non-typical. Lieber and Press must study this 5% in the first place, but instead they ignore them! This calculation involves the death of many millions of people and quite possibly the destruction of civilization — it cannot be made lightly.

They write, “some probability of nuclear retaliation far below 100 percent should deter almost any prospective attacker. They [critics] err, however, by assuming that any level of first-strike uncertainty will create a powerful deterrent effect. There is no deductive reason to believe that a country with a 95 percent chance of successfully destroying its enemy's nuclear force on the ground will act as cautiously as a country that only has a 10 percent chance of success.”

In our view, this is the main error of Lieber and Press. The decisive factor is the EXISTENCE ITSELF of unacceptable results of retaliation, independent of their probability and size. This is because the individual probability of unacceptable results among all possible results of modeling does not play the decisive role; ANY of the calculated results IS possible if a real nuclear war occurs; i.e., IS, but not ARE, because a real nuclear war is possible only one time.

In 1987, American experts stated that, “Dramatically different outcomes might not be downright unlikely, but only less than the expected outcome. The expected outcome, thought the most likely, might nonetheless be unlikely . . . most sinister of all, but almost surely present, are the ‘unknown unknowns' of which operational planners are not even aware.” (Managing Nuclear Operations, by A.Carter, J.Steinbruner and C.Zraket, 1987, p.612)

Finally, Lieber and Press too often refer to history to confirm the correctness of their conclusions. As they suggest, the experience of the Cold War gives them the right to believe that “the possibility of a U.S. nuclear attack should not be entirely dismissed.” We think, however, that historical parallels are always dangerous. But in the given case they are absolutely inadmissible. At least, such conclusions should not be used as the basis for a scientific argument.

We believe the noted shortcomings of both the mathematical modeling and the approach to the assessment of modeling results are enough to consider the main conclusion of Lieber and Press as incorrect. The U.S. cannot eliminate Russian nuclear forces by means of a surprise attack without causing unacceptable damage to itself. We are confident that neither the U.S. nor Russia will obtain “Nuclear Primacy” in the future.

However, in order to adequately resolve this ultimate question, a joint working group of American and Russian official experts should be organized to model all possible present and future scenarios of a nuclear war. Such joint modeling is possible, with the help of already known data plus conditional ones, without inflicting any damage on the national security of both countries. And the results of this cooperation must be open to the public.

It is of the utmost importance that both the U.S. and the Russian Federation permanently demonstrate to the satisfaction of each other that a nuclear war cannot be won and must never be fought.

Global Research Articles by Valery Yarynich"

No comments: