I’m taking a class on academic writing this semester. It’s been quite the experience. For the most part, the class is populated with social science majors. Attending class each week is like taking a dip in a refreshing cold stream after the clinical dryness of my usual physics and engineering activities. For class, we’re required to keep both a writing and a reading journal. The word journal, makes the activity sound soft – touch-feely – but it’s not. At the heart of the activity is a metric driven process like the ones that Shankar Hemmady, myself, a few industry luminaries wrote about back in 2007. In our writing class, the process is more organic than what’s used in most planning/metric driven verification processes. Rather than planning which metrics to measure before we begin, we start by measuring metrics that have served other others in the past. As we build up a set of data, we start to analyze it, and of course from there fall into the usual plan/execute/measure/analyze/plan loop. For folks who are not using a metric driven approach yet, here are the basics.
1. Measure what’s going on in the project. Ideally, you should figure out what you want to measure before the project begins. No worries though, as the project evolves, you’ll figure out what information is important to you. In our writing class, we’re measuring the time of day of each writing session, its location, how long it lasted, and where it took place. Since we are actually a little touchy-feely, we’re also measuring the writer’s emotions at the end of each session, were you tired, elated, powerful, etc…?
2. Analyze the data to determine where the project could be improved. For verification engineering, code, functional, and assertion coverage tell us which parts of he design have been thoroughly tested and which parts need more work. For writing, word count vs. location might tell an author where they’re most productive. Word count vs. time of day, might suggest the optimum time to shut the door and just sit down and write.
3. Make necessary changes, but most importantly, keep measuring results! Each round of improvements will bear fruit or it won’t. If you have metrics, you can find out the hows and whys of the improvements, or lack there of. As processes improve, you might drop some measurements and start taking others. Remember to document everything. Documentation is the history that will remind you why decisions were made and keep you from repeating mistakes.
The Metric Driven Life
If you’d like to take metrics to the extreme and make them your life, there’s an app for that.