10 years and still Cannot Measure Productivity
One of the great frustrations of the software industry is that we don’t have properly built studies to figure out whether development tools and techniques such as object-oriented programming, test-driven development, or even high-level languages are good for us. Regularly we see improperly built studies that do so - often these are done poorly because they rely on a bogus measure of productivity (such as Lines of Code per staff day). 10 years ago today, my frustration led me to publish my bliki post CannotMeasureProductivity. I think this post is sadly just as sound today as it was a decade ago.