Tuesday, September 21, 2010

Notes on Schank's Future of Decision Making

Waffler that I am, I both like and dislike Schank's new book, The Future of Decision Making:  How Revolutionary Software Can Improve the Ability to Decide.  I'm still reading and digesting and trying to decide on this take on decision-making.  I'm also thinking through case-based  reasoning (CBR) upon which the book is based.

Here's his thesis in a nutshell:

Humans decide in a way that is largely unconscious and intuitive and later consciously rationalize their decisions. Conscious logical thinking--within the context of decision making--is overrated and often leads us to veer from a more accurate gut.

A quick aside.  If you google the title of the book the first hit is

The Future of Decision Making: Less Intuition, More Evidence which argues that we should rely less on intuition.  Further down in the search results  you get the link to Schank's book.  So Google is a waffler too.

Bad decision happen because corporations place to much emphasis on defining rules, relulations, policies, best practices, and so on and constraining employees to adhere closely to them.  Bad decisions also happen because people's experience is inappropriate or too narrow or insufficient in other ways for the problem at hand.  Another way of putting this is that there are more jobs than experienced people to fill them.

How do we solve this problem?  Build software based on the principles of CBR that collects and indexes the rich experiences (stories) of the true experts--of the "old salts" as Schank et al like to call them.  Integrate this special knowledge base with the activities of employees such that the software delivers up appropriate stories to decision makers at point of decision.

I find the book both thought provoking, even insight provoking, and yet also simplistic (as one would expect from a book with the self-referential word "Revolutionary" in the subtitle).

The locus of insight and simplicity is the same thought--that good decision making is mostly non-conscious and intuitive.  I happen to believe that it is a mix and truly good software, in my opinion, would be that which helps a decider to use the right mix of mental dimensions (which might transcend the intuitive-rational dichotomy).

You can't have a purely intuitive decision.  In other words no-conscious activity, no words, non connections and transitions. "Intuitive" is an ideal type.  There are differnt types of non-intuitive mental activity:

1) Distinguishing, defining, explicating.  In a decision context, what is important, what are facts, what are the assessments?

2) Organe into a hierarchy.  In the case of criteria, this is a sub-criterion.

3) Aggregating, summing, calculating.  Adding all my judgements together, I say that X is this much better than Y.

So how do we prioritize these logical elements?  My gut tells me that distinguishing is most important, then summing, then organizing.  Summng however is less meaningful without an hierarchy.  Shallow hierarchies are better than no hierarchies at all.

No comments:

Post a Comment