The Decision Strategist

I started writing this as a response to a comment from James Taylor, author of Smart Enough Systems, on Snap Decision Making in Blink, but realized that it was getting long enough to be a post in it’s own right. Here is the comment from James:

One of the interesting areas I found when I reviewed Blink was that of how to apply these conclusions to automated decisions. How can one combine rules and analytics to make decisions more effectively. The importance of continuous learning or adaptive control is key as is the right balance of analytic insight and rule-based decisioning.

The concept of automation to circumvent decision making errors is interesting, but I don’t think it has arrived yet. It is a solution to a problem (from the companies perspective) of having to make repeated decisions in a given situation. Dealing with unsatisfied customers is a typical example. Do we refund, replace, ignore or send a letter of apology? The answer can be systematically determined based on several variables to find the maximum customer satisfaction for the lowest cost.The problem is that these systems remain inflexible. While reducing the cost of having a person decide what to do, they lose out on the subtle nuances that real people, with their years of experience interacting with other humans, can capitalize on.

It’s not a processing power problem, it’s a sensory problem. Any human has years of experience reading body, voice, and other situational cues that a computer faced with the same decision won’t have. We haven’t figured out quite yet how to give that same complexity to a decision algorithm.

Worse, customers hate them.

But can decision systems be used to correct biases and adjust for our mental shortcuts? Perhaps, but only if three things are true:

1. Decision makers in the country use them.
2. They are trained to explicitly avoid the same biases we encounter.
3. They have access to the same information that the decision maker does.

2 is fairly easy. We develop an algorithm that involves explicitly calculating the value of each choice instead of taking the mental shortcuts that we use.

But 1 and 3 are much more difficult. I would wager that quite a bit of the success of a company is determined by the small and seemingly insignificant interactions and decisions. Are employees going to use the decision system for every small choice? It’s impossible and would probably be very detrimental. You can use it to make larger strategic and organizational decisions, but even then it’s more of an assistance to an executive, who takes the results of the decision system as one of many pieces of information and makes a decision from there.

Even if we could solve the problem of having a decision system process and weight all the subtle cues a human receives during the decision making process, we would still need to figure out a way to translate cues into data points and determine what each cue means. We don’t have to do this explicitly, but the decision system needs to know the cues and know what they mean. It also has to be able to pick up on new cues that develop as culture shifts. Adaptive learning may get us there eventually, but we aren’t there yet.

-zot

Popularity: 22%

Vote for this post on: These icons link to social bookmarking sites where readers can share and discover new web pages.

This entry was posted on Friday, October 5th, 2007 at 11:33 am and is filed under Decision Making, Technology.