The primary purpose of software estimation is not to predict a project's outcome; it is to determine whether a project's targets are realistic enough to allow the project to be controlled to meet them.

A typical software project can present more opportunities to learn from mistakes than some people get in a lifetime.

It's hard enough to find an error in your code when you're looking for it; it's even harder when you've assumed your code is error-free.

Software projects fail for one of two general reasons: the project team lacks the knowledge to conduct a software project successfully, or the project team lacks the resolve to conduct a project effectively.

In my experience, you either have encapsulation and abstraction or you have neither. There is no middle ground.

95% on content and the computer language remains the same.

The default movement on a software project should be in the direction of taking elements of the software away to make it simpler rather than adding elements to make it more complex.