A story about planning for Desert Storm (1991)
In an email exchange with retired DIA (Defense Intelligence Agency) analyst, William (Chip) Sayers, he sent me this account. I asked him if we could publish it, as I think it is a wonderfully unfiltered account. He agreed, although pointed out that he would also be covering some of this in his presentation this fall. It is on Day 2 of HAAC and is on “The Combat Assessment Technique.” See: Schedule for the Historical Analysis Annual Conference (HAAC), 27-29 September 2022 – update 4 | Mystics & Statistics (dupuyinstitute.org).
In early 1989, I went to work at an office at DIA that specialized at looking at the world through the eyes of the Soviet General Staff. In particular, we used the Soviet’s Correlation of Forces Methodology. However, we only partially understood it and needed some data to close gaps for us. I cast around for something that would plug these holes and settled on the QJM as the best candidate. It was my belief that both models had approached the subject from the same standpoint and therefore the one could help inform the other.
This paid off — to an extent — in the run up to the ground campaign of Operation DESERT STORM. Gen. Schwarzkopf desperately wanted to know at what point he should let loose his ground forces and so had his staff casting about for a methodology that would give him a way to measure the success of the air campaign in softening up the Iraqis. You would think that after the years we put into WEI/WUVs and all the various models that J-8 and others used, we would have had a good basis for solving his problem. Well, you’d be wrong. Very wrong. As an illustration, Schwarzkopf claimed after the war that, in the summer of 1990, CENTCOM had gamed out the exact scenario that actually occurred and got the exact same results. Schwarzkopf was channeling his inner Nimitz, but I guarantee you that was impossible. I spent several years in the bowels of the Pentagon gaming the Soviet problem with J-8 using the same model and I can truly say that the model itself was geared toward making it absolutely impossible for the attacker to win. I promise you, there was no possible way that Schwarzkopf’s troops got the results he claimed unless he disregarded the output and simply directed the outcomes, himself.
In any event, in the weeks running up to the ground campaign, he didn’t have anything — much less a full-scale model — that could answer his question. I saw a bunch of Majors and LTCOLs running around like chickens with their heads cut off, without coming to any useful conclusions. We ran through exceptionally complex pseudo-science formulae and we saw some so simplistic, my third-grader grandson could have done better. None of it, however, satisfied Schwarzkopf because no one could agree on an approach. In the end, Schwarzkopf threw up his hands and directed that we simply let him know when the air forces had attritted the Iraqis by 50%, and then he’d send in the ground troops. My job at that moment, was to pretty much figure this out for DIA, and given my possession of the QJM and my hybrid methodology, I felt I could be very confident in making the call that CINCCENT needed. Unfortunately, we were on opposite sides of the impenetrable G2/G3 no-go zone, so they weren’t interested in listening to my opinions.
I knew from my historical studies that 50% attrition was massive overkill and that we could go long before we reached that lofty — and probably unobtainable — goal. What Schwarzkopf didn’t know, and I did, was that the agreements set out to decide who did what to whom did not allow DIA access to the data collected by our tactical recce jets. In other words, DIA was going to have to do all its BDA analysis using less useful means. There was simply no way our guys could see a small hole punched through the top armor of a tank from the means we had at hand. Thus began the great BDA war between CENTCOM and Washington. We knew that we didn’t have the proper resources to do the job right, but were told to get on with it, anyway. On the other hand, CENTCOM had a formula of how many “kills” to award according to the in-flight pilot reports given the type of airframe flown. F-16s were heavily discounted, while A-10s were believed as though their claims were coming down from Mt. Sinai on stone tablets. I’m a former USAF pilot and I know that the last guy to ask is the one who just came through the gauntlet.
I vividly remember running my final calculations on Friday night before the attack kicked off the next day (Washington time) and being very satisfied that Schwarzkopf could go at any time he wanted. Interestingly enough, none of this had any input into his decision to go. Few people remember it, but Gorbachev was negotiating with Saddam and had successfully convinced him to pull out of Kuwait. The agreement they came up with would give the Iraqis three weeks to pull out. At this point, it had become a major goal to eliminate the Republican Guard and we didn’t want them to pull their head out of the noose, so President Bush turned down the compromise and ordered the ground forces in.
Ok, so here’s the point: Despite all the big talk and incredible claims, when push came to shove, the Army had nothing/NOTHING to use as a basis for planning. Lord knows we threw enough time and money at the problem, but in the end, Schwarzkopf just had to pray that we had enough combat power when our troops rolled across the line. He would have given anything up to half his kingdom for the QJM at that moment. He had a lot of opinions to choose from, but nothing solidly based on history. And frankly, I don’t think the situation has changed in the intervening 30 years. Now that the chips are down, people aren’t likely to care WEI/WUVs were developed by the opinions of various branch influence groups. But a model with an historical basis would be worth its weight in gold.
Glossary:
QJM = Quantified Judgment Model, Trevor Dupuy’s earlier combat model. The TNDM (Tactical Numerical Deterministic Model) is Trevor Dupuy’s update of the QJM.
WEI/WUVs = a weapon scoring system developed by CAA and used by RAND.
This email exchange was part of a discussion of what TDI could be doing, if properly budgeted.
