Some issues concerning the effectiveness of parallel strategies in R&D projects

When there is considerable uncertainty concerning outcomes such as in R&D it is common to identify and explore several approaches to a particular objective so that the best approach may be chosen. The outcome of any approach is uncertain; hence it is difficult to choose the best one at an early date. To deal with this uncertainty two or more approaches may be made, i.e., a strategy of parallel approaches. Such a strategy can maintain options while better information for a decision is gained, hedge against an unfortunate outcome, or stimulate competition. This strategy has been used in spectacular programs such as the Manhattan Project, Atomic Submarine Program, and SST Program, etc. It also characterizes the framework of decision making in Federal R&D procurement where large amounts of resources are consumed in developing and evaluating competing R&D programs. The rate of learning is a critical parameter that influences how a strategy of parallel approaches may be used. By rate of learning we mean the rate with which uncertainty is reduced and objectives sharpened with progress in a project. If the rate of learning is rapid, parallel approaches need not be carried far toward completion before a choice among them may be made and the cost of a strategy will be low. The converse is true with a slow rate of learning. This paper explores the effect of learning by determining the rate of learning that best explains the pattern of choice and outcome in 21 actual R&D projects. A fairly slow, rather than rapid rate of learning is found to be significantly associated with the pattern of choice and outcome realized in the cases that are analyzed. The methodology and implications of the results are discussed.