Military Power by Stephen Biddle

Biddle uses databases of conflicts since 1900 to create several OLS regressions to estimate casualties, territorial gain, and duration of conflict. However, he includes two variables which cloud the analysis: nationality of the forces and average technological advantage in years. Nationality seems fluid (india / east pakistan / bangladesh) and the model fails when the nationality hasn't fought a war since 1900. And technology seems to create larger than linear differences in lethality ( compare single shot to automatic to automatic with gps and radio for long-range target acquisition ).

Given the results which show that the estimated effects were less than the national differences, it appears that Biddle needed the nationalities in order to make his linear model work.

As recently as 1991, a massive effort using state-of-the-art methods and the nation's best analysts radically overestimated US losses in the upcoming Gulf War. The prewar congressional debate hinged on casualty expectations; these were widely seen as the key to Congresss's vote on the use of force. With so much at stake, no effort was spared to achieve realistic estimates: prominent academics, government analysts, and senior military officials gave testimony using methods ranging from computer models to historical analogies to professional military judgment. Virtually all were way off. Even the closes estimate overshot the actual casualty count by more than a factor of two. The next best missed by a factor of six. The majority were off by more than an order of magnitude; official estimates were reportedly over by at least that much; while some official projections erred by a factor of over 200.
page 1
Military balance estimates are central to modern political science. Much of our current understanding of international politics rests on the assumption that state behavior is shaped by the threat of war and the pursuit of military capability...
Yet the standard capability measures at the heart of all this are actually no better than coin flips at predicting real military outcomes.
page 2
The garrisons of Belgium's state-of-the-art frontier fortresses discovered this (the power of new artillery) in August 1914 when th eGermans' new 420-mm siege howitzers reduced their steel-reinforced, six-foot-thick concrete bunkers to rubble in a few hours, burying the occupants inside. German trenches in 1917 suffered obliterating artillery barrages of literally atomic magnitudes: the ten-day Allied bombardment before Messines in July 1917 dropped about 1,200 tons of explosives -- in nuclear parlance, more than a kiloton, or more explosive power than the US W48 tactical nuclear warhead -- on every mile of German defensive frontage.
page 30
The tremendous, ongoing increase in lethality defined the central problem of modern tactics: how to survive the hail of metal long enough to perform meaningful military missions...
Two broad approaches emerged almost immediately. The first used cover and concealment to reduce the attackers' exposure why advancing; the second used suppressive fire to keep the defenders' heads down while the attackers were exposed.
page 30-1
At Operation Anaconda in March 2002 (Afghanistan), less than half of the al Qaeda fighting positions ultimately discovered on the battlefield were known to American forces prior to ground contact, in spite of an intensive prebattle reconnaissance effor using the latest in sophisticated surveillance technology. In fact, most fire received by US forces in Anaconda came from initially unseen, undetected al Qaeda fighting positions.
page 56-7
In November 1950, for example, Harry Truman and the Joint Chiefs of Staff thought North Korea was all but defeated and victory finally in hand. Though Chinese prisoners had been taken as early as November 13, and although intelligence reports showed indications that the Chinese were preparing to intervene in force, this evidence was discounted against the strong expectation that the war was all but over. When the Chinese then launched a massive offensive beginning on November 25, overstretched American forces were thus taken by surprise and overwhelmed, spurring the longest retreat in American military history.
page 64-5