Cost of Change part II

A few weeks ago I discussed Boehm's Cost of Change Curve and how, in Balancing Agility and Discipline Boehm and Turner pointed out that 20% of changes made up 80% of the costs.

This suggests:

• 20% of changes have a very steep curve, and,
• 80% had a very shallow curve.

When I posted my thoughts to scrumdevelopment@yahoogroups.com, Mary Poppendieck replied :

"An experienced team can figure out which decisions need to be made early and which can wait until later.  Unfortunately, our development processes try to force all design decisions to be made as early as possible.  This is a bad approach, one that drives up the cost of change. 

Toyota has a product development process which is four times more productive than those employed by most companies. This process delays final design decisions to as late as possible in the process.  The result is that changes are generally not necessary – they don’t have to change decisions that haven’t yet been made".

So, this is the value of the Poppendiecks principle "Decide as late as possible" - if you don't make your decision until late, then when you do make your decision, your not changing anything.  The change is effectively free.  You can only "keep your options open" like this in an iterative environment ... the full blown waterfall makes you decide as early and quickly as possible.

In the mean time I've been writing up my dissertation and I chanced upon Boehm and Papaccio's 1988 article "Understanding and Controlling Software Costs" (from  IEEE Transactions on software engineering, volume 14, number 10, pp.1462-1477) where they say:

 “the major implication of this distribution is that software verification and validation activities should focus on identifying and eliminating the specific high-risk problems to be encountered by a software project, rather than spreading their available early problem elimination effort uniformly across trivial and severe problems.”

 Mr Boehm is a very clever chap.  We all owe him a beer. I hope I get to buy him when he's in Edinburgh for the 26th International Conference on Software Engineering, no that I'm attending).