Monday, August 16, 2010

If I crash, just let me lie there

The other day I was feeling up to a brisk ride through the countryside. I wanted to feel the wind in my hair, experience the open road and open sky, and savor the thrill of facing death head on.

The original Son of Anarchy, about to my mount my Daimler crotch rocket.

Thinking that I would also enjoy sharing the adventure with a friend, I asked adventurer extraordinaire, George P. Burdell, to accompany me on the ride. He looked at me, and asked, "Why don't we just spend the day at the hospital?"

For the sheer thrill of facing risk head on, few activities beat going to the hospital, even a 420 mile ride on a motor bike.

Tuesday, August 10, 2010

Cosmology's not broken, so why try to fix it?

This is another interesting article on the difference between Bayesian and frequentist statistics.

I posted a link back in March to another related article in ScienceNews: "Odds Are, It's Wrong".

How do you recognize critical risk factors?

Whenever I conduct a business case analysis with a client, the model's assumptions are always treated as uncertainties until indicated otherwise. A typical analysis can incorporate anywhere from 10 to 50 such assumptions. The analysis rank orders the most important assumptions by how sensitive the decision metric (e.g, net present value, positive cash flow, sustainable levels of resources, etc.) is to the assumptions and by how likely each assumption might cause regret for taking one course of action over the next best course of action. This latter considerations reveals which uncertainties are critical to achieving satisfaction.

Without doing this kind of analysis, decision makers face thinking through a bewilderingly immense number of implications among the assumptions, which often hampers their ability to commit to action. Sometimes, in an effort to shortcut analysis, decision makers focus on favorite or easy to identify risk factors instead. But this is akin to looking for your keys under a street lamp even though you dropped them in the dark bushes ("But the light is better under the lamp!"). This behavior exposes organizations to undesirable outcomes that could have been anticipated, wasting time and resources fixing potentially avoidable problems.

For discussion:
1. On average, what percentage of the 10-50 assumptions do you think are actually critical to success?
2. How does your company identify critical risk factors and measure their potential impact on important decisions?
3. How does your company ensure that proper attention is placed on important risk factors and not just favorite ones or those easy to identify?