Recently I was asked why does it cost more to fix bugs later in the development life cycle? I gave the industry standard reply, that if a bug is found later it must go back through all the previous stages, i.e. if a bug is found during implementation it must go back through requirements, design, coding, testing, etc. which cost more money than if it was found at requirements or design. All sounds legit, and it's what has been pushed down our throats for years. But, what if you are doing Agile and the loop is much smaller? Or XP and we have continuous integration and deployment? Does it still hold? Did it ever?
I started to do some research and found some interesting things...
The original sentiment seems to have come from,http://faculty.ksu.edu.sa/ghazy/Cost_MSc/R6.pdf
This is a great post related to it - http://programmers.stackexchange.com/questions/133824/is-it-significantly-costlier-to-fix-a-bug-at-the-end-of-the-project