Wednesday, January 25, 2012

Not right does not mean wrong

I'm reading a really good document about risk analysis, and the author makes the point that when using probabilities to make predictions, at some point the future will unfold in a way that will make others perceive you were wrong. He emphasized "perceive" and that got me thinking.

We do that a lot. Someone does their analysis, makes a decision, and then acts on it. Like a football coach that decides to go for it on fourth down in overtime rather than punt and put the ball in the hands of the opponents' red hot quarterback. The play doesn't work and everyone says he was wrong. Oh yeah, based on what?

"Well, the play didn't work, that proves he was wrong." No it doesn't. It could merely be an instance where the future took the less probable path. It's gonna happen! Chances are good that his decision was right.

Someone's analysis and decisions should be judged only over time and against a pattern of how often the predictions come true. Additionally, we should judge them by whether their analysis has a feedback loop that learns from failure and how quickly they respond to the unexpected outcome.

Anything else is just Monday morning quarterbacking.


Andrea Wenger said...

It works the other way, too. Machiavelli is misquoted as saying that the end justifies the means. What he really said is that the end will be perceived to justify the means if the attempt is successful. But successful doesn't mean right.

Paul Ballard said...

I would argue that 'not right' does mean 'wrong': it's a binary decision. But I agree it is not an appropriate decision for the situation. Better a continuum or scenario based model of expessing a judgement.

Michael Hughes said...

You're absolutely right, Paul, it's a tautological contradiction. It was mainly a "headline gimmick." :-)

perpetualthought said...

Completely true. I agree with that theory. Most decision theory needs to be revalidated using data over a period of time. Instantly calling a decision wrong on the basis of a single result is actually wrong. That means that if a coach uses a tried and tested theory play against a team that's worked in the past and it fails one time, he was wrong all the other times too.