Hello all! Sorry I have been away for several months – it has been a busy spring, summer and now fall here in northeast Ohio. I thought I’d drop a quick post today highlighting a great piece of research that I just finished reading.
Those of you who know me know what a great fan I am of Conway’s Law. Having worked for and consulted for many organizations over my career thus far, I have found Conway’s Law to be a solid, immutable truth. For those of you not familiar with Conway’s Law, it essentially states that:
…organizations which design systems ... are constrained to produce designs which are copies of the communication structures of these organizations.
Since software development is about expressing and constructing a solution to solve a problem, the pattern of communication about the problem domain and the vocabulary used to describe it becomes paramount. In addition, the manner in which this pattern of communication takes place and evolves – ultimately – influences solution design, construction and implementation. I guess I like to say that good, well-run organizations produce good software. And, that… …well, you get the picture. Note that in my experience the word “organization” here applies to both the technology department and the business people in the organization (can’t have one without the other).
Though there has been additional research on Conway’s Law since Melvin Conway first introduced it in 1968, empirical evidence has seemingly been hard to come by. Thus, often (at least from my perspective), Conway’s Law can end up relegated to the subject of discreet giggling around the water coolers and hallways of dysfunctional organizations. 
How fitting is it then that this study, in the midst of the unit test code coverage quality debate, actually ties organizational metrics to the quality of software produced!? I say, quite fitting. Take a moment and read it. I believe it will validate all that you’ve likely come to know and expected about the effects of “the organization” on producing software.
Non-softies take note: You just might even find it humorous that the guinea pig used to exercise the metrics was none other than Vista!
Cheers!
TheInfluenceOfOrgStructureOnSoftwareQuality.pdf (172.27 kb)