This was the title of Joe Schofield's excellent presentation at SEPG 2012 - the Software Engineering Institute's annual U.S. conference for practitioners and prospective practitioners of CMMI. Â Joe has a lot of CMMI experience from his many years at Sandia Labs (from which he recently retired), but he is also a CFPS and currently President of IFPUG, so he has a unique perspective on the importance of getting it right when it comes the sizing part of software measurement. You can browse Joe's presentation from SEPG at his website, but I wanted to share with you a couple of things that I thought were interesting:
- The term "Size" is found in the CMMI (Dev) reference model 55 times!
- Over slides 6-8, quoting as his source his own 2005 paper from Crosstalk magazine, "The Statistically Unreliable Nature of Lines of Code" (now archived on ResearchGate), Joe spells out the vagaries of using lines of code as a size metric in software development.Â Joe and others conducted a controlled experiment using the same sets of developers, coding the same programs, using the same LOC counting method, having been trained by the same instructor.Â Despite all these efforts to control the outcome to produce a consistent correlation between the functionality of a particular program and the number of lines of code, there was no correlation at all!
- Joe is careful to be objective about function points, but this level of detail clearly demonstrates that using lines of code as a size metric is not the way to go!
What do you think about Joe's presentation and function point analysis on the whole? Leave a comment below to share your thoughts!
Mike Harris DCG President