Path: utzoo!utgpu!watmath!clyde!att!cbnews!GA.CJJ@Forsythe.Stanford.EDU From: GA.CJJ@Forsythe.Stanford.EDU (Clifford Johnson) Newsgroups: sci.military Subject: Combat simulations Message-ID: <5438@cbnews.ATT.COM> Date: 6 Apr 89 01:29:54 GMT Sender: military@cbnews.ATT.COM Lines: 71 Approved: military@att.att.com From: "Clifford Johnson" There's been some discussion of Dupuy's battle-prediction models. A detail or two might be illuminating. Dupuy is a prolific war game expert who founded the Historical Evaluation and Research Organization (HERO) dedicated to promoting the cause of historical analysis and improving the security of the United States. He regularly advises the Pentagon's top war gamers. His Pentagon-applied contributions to wargaming center on a "Quantified Judgment Model" (QJM), which purportedly measures the "Power Potential" of any military force. In combat, the force with the greater power potential "should" win. This potential is constructed from more basic "measures" such as the "Lethality Index" of individual weapons, computed by multiplying numerical estimates of: per hour rate of fire; targets per strike; relative incapacitating effect; range factor; accuracy; reliability. According to this formula, a sword's lethality index is 23 (= 60 x 1 x .4 x 1.03 x .95 x 1), and a one-megaton nuclear bomb exploded in the air has a lethality index of 695,385,000 (= 1 x 8,5000,000 x 1 x 101 x .9 x .9). I'd like to point out that derivative battle predictions based on such arithmetic are held in low regard by some commanders, and even he admits their application may be abusive. In 1985, Thomas Dupuy wrote in the magazine Army: "The senior decision makers of the U.S. military establishment are increasingly basing their decisions on the outputs of computer models and simulations which are widely recognized to be unreliable and unrealistic." In this article, Dupuy endorsed the views of Major General B. Atkeson, former director of the Army Concepts Agency, who had written an annonymous (he was on active duty then) article in the same magazine, stating that many simulations "were as perforated with logic holes as a sieve." Such techniques, trivial though they are on their face, provide the bottom-line justification for America's nuclear arsenal and war plans. See, e.g., The MX Missile and Associated Basing Decision, SASC Hearing, Dec 8, 1982, at 34, where the Senate is informed that PTP, the Probability To Penetrate (destroy) hardened targets is 1.0 for nuclear missiles, and "HTK = SSPK x WSR x PTP," where HTK is Hard-Target Kill potential, SSPK is Single Shot Probability of Kill, and WSR is Weapon System Reliability. According to this formula, the United States was deficient in "time-urgent hard-target kill potential", and so the MX was needed. I concur that PTP=1, but I am interested in the realistic values of SSPK and WSR. All models I have seen account for technological glitches only, and typically come out at perhaps a 90% single-shot probability of kill. I have heard rumours that the figure might be realistically much lower, especially due to corrosion of computer chip substrates. Is this credible? Also, does anyone know of any models that parameterize the percent of missiles that would not launch due to personnel "problems"? Despite the redundancy of launch crews per missile, it is not inconceivable that some crews would even use their inhibit launch commands to prevent others from completing their orders. I was informed by someone on misc.legal, who refused to divulge his source, and without elaboration, that a study was undertaken which established that a significant minority of ICBM launch crews would simply not launch when actually ordered to do so. Accordingly, I was told, personnel procedures were changed to reduce this proportion. If the proportion was found to be significant, surely it should be included in the regular calculations? To: MILITARY@ATT.ATT.COM