Draft of the essay to be released with SPaMCAST Podcast

 

The I 35 bridge failure in Minneapolis illustrates a potential failure in many process improvement programs.  Listening to architectural experts suggests the possibility of a cascade failure, where one part fails leading to others failing one after another.  Most organizations take a similar approach for software process improvement (just180 degrees out of phase).  The theory is that if you improve one part of a process all others will get better. This was fixed a piece syndrome indicates a failure in systemic thinking. Optimizing a single component without truly without understand the whole business process will lead to sub optimization of the whole. 

Software process improvement must begin with a model (preferably a statistically predictive model) and a set of very focused goals for the program. Each process change must be vetted against the model.  Ask the questions:

  • Will the change improve the “whole” system?
  • Will the change lead to the behavior or outcome we “wanted” based on out goals? 

I would suggest an unmerciful vetting of each change being considered except that all approval processes need to be sensitive to the potential of a “black swan” suggestion.  A black swan is a completely unanticipated change that completely rewrites the rules.  An example would be the first time extreme programming was suggested in a waterfall organization. SEPG’s will (or any other type guidance committee) need to be specifically trained to recognize and accept potential positive outliers.  This is a difficult concept in most organizations as committees tend to be creatures of consensus thought. Accepting the constraints of committee group think, it is important to create the mechanisms incent some degree of dangerous thinking. 

One mechanism to ensure that some radical thoughts are evaluated and tried is to set a quota for failures. Management by objectives systems in use in most corporate organizations tend to frown on failure. Assuming the basic behavioral theory that a rational person will maximize their potential for success by focusing on areas that have the highest probability of a positive return will create an atmosphere of risk aversion.  Radical ideas rarely have a certainty of success (otherwise they would not be radical).  Setting a quota for failures is very different than an organization’s saying that figure is tolerated which is a negative statement.  Quotas send the clear message that failures are not only anticipated but are recognized as important to the growth of knowledge and possibilities.

Ask yourself whether you’re SPI. program has fallen prey to a purely centrist thinking. Centrist thinking will generally suppress all variability both positive and negative. Improvements may progress but will be purely incremental.  Radical ideas such as WIKIs, xP, SCRUM or AJAX to name a few would never gain a toe hold.  Radical ideas represent a step changes that challenge safe, conventional thought.  I would suggest that all process improvement programs must not only consider thinking out of the box but occasionally need to redefine the whole box in order to change the world and is that really why we are in the business?