A632.1.4.RB_HallMike

At the opening of the section discussing dynamic programming/optimal dynamic decision analysis, the book discusses how it turns out that we are just as likely to go with our gut instinct and get it right as we are to sit down and try to think it out.  In the Navy, we have a term for this – “nuking it”.  Essentially, nuclear propulsion operators tend to overthink the problem in front of them when often all they need to do is trust their gut.  There is a fantastic book out there called “Blink” by Malcolm Gladwell that discusses how important your subconscious is and how your “gut” feeling is just your subconscious processing information faster than you can consciously and coming to a pretty good solution.  Great book, especially for a decision making class, but I digress.

Looking through the description of dynamic programming, I am a bit skeptical at its usefulness at helping solve business solutions.  My big complaint is you are making assumptions/guesses on the probabilities of some event occurring and then using these guesses to determining the best course of action.  If I was being forced to make a decision, I’m not sure I would trust any information that results from dynamic processing as it is based on guesses/predictions – something I am skeptical of.  How do you know how accurate the assessments of the probabilities are?  I know in my experience, incomplete data/guesses are not good when trying to convince your superiors that you know what you are doing – I’d rather just make a decision with the available data vice make a decision based on data that is based on likelihood of some event occurring.  Now what I do like about the system is that large amounts of data is collected prior to plugging in info into the equations.  I know that I definitely prefer to have as much data as possible prior to making a decision, even though that can sometimes overload your thought process.  One of the most challenging aspects of OOD is your ability to filter the large amounts of data coming into you and focusing on the important aspects based on the situation (i.e. all data is relevant at one point or another just very little data is relevant all of the time).  As such, it is very important that you are able to key in on the key pieces of data in order to make your decision.  It is somewhat like having a box of crayons and being forced to decide on 1 color – you’d probably want to have the big box of 164 colors first, then narrow it down to the finalists, and then choose from that small collection.  When making decisions, I want all 164 colors worth of raw data so that I can pick out the key colors for me to make my decision.

Going back to the discussion points, the equation only works if your assessments of the probabilities are accurate.  This will force you to think out toward the future and make you create future plans (especially contingency plans).  When applying it to optimal dynamic decision analysis, it would definitely assist you in determining the likelihood of all of the possible outcomes, as well as the multitude of second and third order effects the decision might have.  With respect to its application, per the book optimal dynamic decision analysis assumes that the user is able to identify all possible outcomes.  Given that it is impossible to determine all of the possible outcomes from a decision, I think the second half of optimal dynamic decision analysis is more important – being able to learn from history.  By being a good student of the past, you are better able to determine potential outcomes based on previous situations.  Additionally, your knowledge of the past my help you better assess the probability that the outcomes might occur, thus helping you perform dynamic programming.