On the same day that the US government General Accountability Office (GAO) released its “Software Development: Effective Practices And Federal Challenges In Applying Agile Methods” report, it also released a report titled “INFORMATION TECHNOLOGY COST ESTIMATION: Agencies Need to Address Significant Weaknesses in Policies and Practices“. In this report, the GAO compared cost estimation policies and procedures to best practices at eight agencies. It also reviewed the documentation supporting cost estimates for 16 major investments at those eight agencies—representing about $51.5 billion of the planned IT spending for fiscal year 2012. The table below summarizes the GAO findings.
As you can see, only one out of sixteen programs fully met the mighty GAO criteria for “effective” cost estimation: the Navy’s “Consolidated Afloat Networks and Enterprise Services” (CANES) investment. Here’s the GAO’s glowing, bureaucratic-speak assessment of the citizen CANES cost estimation performance as of July 2012:
Out of curiosity, I googled the exemplar CANES program. Here’s what I found on the US government’s “IT Dashboard” website:
Note that after debuting with a rating of 5 (Low Risk) in 2009, CANES is currently rated at 3 (medium risk) and is being “closely monitored” by some higher-ups in the infallible chain of command.
Of course, no one, not even that omniscient and omnipresent devil BD00, can tell what will happen to CANES in the future. The point of this post is that spending lots of money and time on meticulous cost estimation to satisfy some authority’s arbitrary and subjective criteria (comprehensive, well-documented, accurate, credible) doesn’t guarantee squat about the future. It does, however, provide a temporary and comfortable illusion of control that “official watchers” crave. We can call it the linus-blanket affect. Maybe coarser and less comprehensive estimation techniques can work just as well or better?
Comprehensiveness is the enemy of comprehensibility – Martin Fowler