A systematic approach to the development of system programs

The brief history of the development of large programming systems shows a persistent inability to predict cost and time associated with it. For this reason I want to discuss some of the principles and practice which experience (either good or bad) has shown can yield a higher level of predictability. I do not pretend that these principles are always easy to apply or that they do not have to be interpreted to particular situations. However, I do believe that the principles are fundamental in that predictability can be guaranteed to increase if they are followed.