Forecasts are useless to any actor large enough to change the outcome

Willmoore Kendall's 1949 review of Sherman Kent's Strategic Intelligence for American World Policy rejected Kent's model at its foundation: probabilistic forecasting breaks for any actor large enough to change the odds.

Kent borrowed his framework from the social sciences, which study systems from the outside. That works for a small country watching events it cannot move. It breaks for a superpower. The United States is the largest variable in the situation it is trying to predict. Tell the President a faction has a 70% chance of winning a civil war and you have told him nothing useful, because his own choices set those odds.

Kendall flipped the question. Intelligence should not predict what will happen; it should map the contingencies: where the situation is still loose, which variables move it, where US power can bend the trajectory. The goal is not foresight. It is to show a statesman the levers he was already reaching for.

Question	Kent	Kendall
What is the world?	An external system to be forecast	A system the reader is acting inside
What is the analyst's job?	Predict the most probable outcome	Find the leverage points that decide it
What is the policymaker?	A spectator placing a smart bet	An actor whose choices change the odds

Kent's prediction-centered model won institutionally and became the National Intelligence Estimate. Kendall's complaint kept returning every time an analysis was technically correct and useless to the person reading it.

Source claim: Kendall argued Kent's forecasting model treats the policymaker as a spectator of an independent future, but a superpower is the largest variable in its own forecast, so intelligence should map where US action can change the outcome rather than predict it.