Following up on Alex's post on Soviet economic growth forecasts, I was intrigued to read the 1940s estimates, emanating from the United States, about when the Soviets would obtain a nuclear weapon. Leslie Groves — who knew something about building a bomb — testified in front of Congress that it would take them twenty years. In 1948 many Kremlinologists were saying "five to ten years," when in fact the Soviets had a usable bomb in 1949. In 1948 an engineer in Look magazine predicted the Soviets would get the bomb in 1954. Many scientists predicted 1952 and some thought 1970. The Joint Chiefs of Staff were predicted the mid- to late 1950s. The Air Force was the one institution which got it right and remarks from Senator Arthur Vandenberg were close to the truth as well.
Groves was skeptical of the Soviet engineers, who did not turn out to cause delays and who regularly did very well with what they had to work with. Other commentators did not realize that 40 percent of the world's known uranium reserves were within the Soviet Union, or that the Soviets could use German uranium quite well.
All this is from the truly excellent new book Red Cloud at Dawn: Truman, Stalin, and the End of the Atomic Monopoly, by Michael D. Gordin. Here is one very accurate review of the book.
One question is what kind of ideological biases, if any, colored these forecasts. Another question is whether today's estimates of Iranian production are any better.