Technological innovations that produced certain major components of the United States military cannot be understood as resulting from a qualitative arms race. Those involved in decisions about new military technologies for the U.S. Army and Air Force simply do not appear to have had access to good intelligence about the Soviet military technological developments. How, then, were decisions made as to technologies to develop?Military research and development decisions are made amid great uncertainties. In an ideal world, such decisions would be managed by estimating the future costs of alternative programs and their prospective military values, and then pursuing the program with the best ratio of cost to value. But...there are tremendous difficulties in forecasting the real value and costs of weapons development programs. These uncertainties, combined with the empirical difficulty American technology managers had in collecting intelligence on the Soviet Union, meant that research and development strategies in the real world tended to become strategies for managing uncertainties. At least two such strategies are conceivable. One of the most politically important can be called, for want of a better phrase, let the scientists choose. [This approach should be] compared with the theoretical and practical arguments for a strategy that concentrates on low-cot hedges against various forms of uncertainty.

~ Stephen Peter Rosen