January 12, 2006

...Learn TDD with Codemanship

Fad Diets & Old Wives Tales

After Christmas, the shelves are overflowing with new diet books and exercise videos. I can see some distinct similarities between fad diets and fad management theories.

For starters, the results are usually temporary at best.

Another close cousin of fad management theories are old wives tales. The problem is one of causality: X happened because I did Y. The car went faster because I put my foot down on the accelerator. The floor is wet because the sink overflowed. America is in deeper debt because they voted for George W. Bush again (okay, bad example). And so on.

In highly complex systems, simple cause and effect relationships disappear under a mountain of possible connections. A business is subject to an infinite number of complex stimuli, and sensitivity to initial conditions makes it impossible to predict the outcome of any action. And that works in reverse, too. Try as we might, we can't join the dots between an emergent business outcome and its root cause - the chain is just too long and too complicated.

In less enlightened times (and, as my previous post about Intelligent Design clearly demonstrates, that was obviously a very long time ago) old wives would tell us to stick potato peelings up our nose if we wanted to get rid of a cold, or keep seaweed in our socks to cure impotence (or whatever). They could get away with this because, occasionally, people would wake up with their colds gone, or their virility returned, and without simple cause-and-effect any theory about how that happened was as plausible as any other. The other week, for example, I cured my headache by walking in an Easterly direction for 20 minutes. Worked a treat...

And so it is with many management fads. Take Business Process Re-engineering, for example. Some businesses try BPR, and performance actually improves. Many, of course, try it and see no positive effects at all. The BPR success stories are quite possibly old wives tales that build on coincidences - "we just happened to try BPR, and performance just happened to improve".

The problem is that, at the edge of chaos, we must learn to live without the simple cause-and-effect relationships we see in individual data points. Instead, we must learn to keep an eye on the longer term trends and build a picture of the complex, statistical relationships between different factors. Statistically, BPR fails at least 70% of the time - that is to say, only 30% of companies experience performance improvements as a result of BPR projects. Who's to say that they wouldn't have experienced those improvements anyway? Where's the control data? I strongly suspect that if you selected 100 businesses who aren't doing BPR at random, and monitored their performance, you'd be hard-pushed to tell the difference.

So if you want to improve productivity on your development teams, I strongly recommend you stick a pine cone in your pants. It worked for me, and it'll work for you (some of the time).
Posted 1 week, 5 days ago on January 12, 2006