There are at most a half-dozen books in computing that should be read by everyone involved with programming. Dijkstra's essay in Structured P r o g r d n g [ll, and Weinberg's The PsychoZogz of Computer PrOg&ng are two obvious examples; Frederick P.
Brooks's The NthicaZ Man-Month is a third.
Fred Brooks was project manager for the development of the IBM System/360 computer, and manager of 0 s 360 software development. 0s 360 is probably the most widely used, influential, and criticized operating system ever built. is a look back over that experience, and an attempt to record the lessons, positive and negative, for the benefit of future developers of software systems. Although the treatment is oriented toward the problems of large software projects, there is a lot here for any programmer who takes pride in his craft. people and months are not interchangeable commodities. that takes ten people one year cannot be done by twenty people in six months. The reasons for this are clearthe more people involved, the more complex their interrelationships, and accordingly the more of their time diverted from production to communication. At some point, in fact, adding more people lengthens, not shortens, the schedule. This leads to Brooks's Law: "Adding manpower to a late software project makes it later."
Given that the human wave approach to software production is doomed to failure, what then? The rest of The MythicaZ Man-Month deals with various considerations involved in successfully completing large software projects. Much of this discussion draws on examples from 0s 360, of course, but also takes in data from other large efforts like Multics and ESS.