Software Engineering discussion
The Mythical Man-Month
>
The MMM After 20 Years
date
newest »
newest »
Yes, this was one of the best chapters in the book. I think it would have been a nice first chapter to a follow up book. It's too bad this chapter is all the way at the end of this book.This chapter also shows humble qualities of Brooks when he points out areas he was wrong.
The topics are very broad, but there's great visionary topics discussed. Many companies struggle to predict the market trends 12 months out. I know some really great talent in Software Engineering, but I don't think I've ever met anyone with visionary skills (and track record) shown in this book.
This chapter nicely summarizes the main points of the book, yet does so in the light of "current" (1995) technology. It is sad to see that we seem to have made little progress in the last 15 years, because so much in this chapter could have been written today, and it would have been considered timely.
Brooks makes a rare misstep in his prediction that GUIs will give way to speech recognition as a UI. He also strangely advocates for a two-mouse (cursor) system, one in each hand. I think that, with iOS/Android, we are witnessing a break in the "every app works the same" GUI model that Brooks likes for the usability benefits across applications (and platforms, to a certain extent). In the mobile environments, even with strict guidelines from Apple (for example), the app interfaces vary quite a bit. It will be interesting to see if this is just a sign of immaturity and we will see convergence in the future, or not.
We also witness the beginnings of the iterative/agile era here.
I heartily agree with two main conclusions in this chapter. First, software engineering is a very human activity, which makes it difficult to apply traditional engineering approaches to software development, and it also makes the people-factor so important. Second, the big productivity gains come from either buying solutions or leveraging high level frameworks or software environments. My current project is built on top of Mathematica, and I can't fathom the additional effort it would take to code it in something like Java.