Dan North misses his own point?

Dan North is one of my favorite tech speakers. He is funny and entertaining as well as thoughtful and subversive. He gave a speech at NDCOslo about Embracing Uncertainty. He states that risk has two important aspects: Probability and Impact. Then he observes very astutely how most of us in IT focus only on minimizing probability and why that focus is misdirected. His reasoning is that you can never decrease the probability to zero so you should focus on minimizing the impact. It's a typical Dan North observation: almost embarrassingly simple yet very refreshing and eye-opening, which puts conventional methods into new perspective. This kind of ideas are the very reason I like him so much. Then he goes on to describe a few techniques with the help of which we are supposed to embrace the uncertainity of effort and uncertainty of technology and so on.

The first technique is called Real Option where you treat everything as a financial option, so that the later you make a commitment the better chance you have to make the best decision. You notice the word 'chance' in the last sentence. This method is very similar to the Last Responsible Moment and one that you should use, but it's optimizing for probability and not impact.

The other one is called Deliberate Learning where you try to find the areas in which you might be subject to second degree ignorance, in other words areas where you don't even know what you don't know. He emphasizes that there will always be stuff we don't even know to be unknown, yet he wants us to try to know anyway. Again, in my opinion this technique is aiming at minimizing the probability of things going awry, and not the impact of a possible mishap.

He's picking a little on eXtreme Programming in this very talk for trying to embrace change instead of uncertainty. I find it ironic because many of the XP practices and principles are about minimizing impact. The Iterations, Incremental Design, User Stories, Baby Steps, Short Release Cycles, Done Done. All of these are based on the simple idea that the smaller you can make the completed chunks of your product the smaller impact of any possible mistakes will be.