Putnam's Effort-Duration Trade-Off Law: Is the Software Estimation Problem Really Solved?

摘要

In 1978, Putnam presented his "general empirical solution" for the software estimation problem and claimed that the problem was solved. His results continue to be influential, and are sometimes treated as authoritative. Focusing on the effort-duration trade-off, we evaluate his study and its results, examining both his approach and studies that have been published since. Several serious problems are identified. Simulations show that Putnam's approach yields incorrect results and that it yields a non-existent effort-duration trade-off in a set of unrelated random numbers. Putnam's "software equation" and effort-duration trade-off law have no credibility and should no longer be used. Several related studies have similar issues, statistical relationships should be handled with more care.