Decision Theory

A friend of mine and I had this rather involved discussion about decision making. In her honest opinion, I stink at it. I would agree with her not based on the success rate of my decision making, but that I do tend to press on it for long. I enjoy the decision making process and award each option its due, thereby giving the appearance of a perennial activity. But today's discussion about a certain decision I need to make in my life, got me all excited and forced me to drop everything and ponder over this.
Decision theory appears to be a well researched area. I have not read a single article/book nor have I visited any lectures/conferences on decision theory. I like taking a stab at something of which I only have an intuitive feel. Like curing AIDS (I had this wonderful idea when I was in the 8th class but a visiting doctor thought it was silly and too sci-fi. All that it involved was keeping the patient in suspended animation using cryogenic techniques and quickly draining the system of infected blood, followed by a super-quick cleaning, followed by a refill!! ). Today tickled me with decision theory.

What follows is my theory about the same. I would invite your critical review of this. Keywords/ideas are highlighted.

I think the first step of taking a decision is identifying what is desired and what is the worst consequence of not consciously measuring the decision towards realising the desired outcome. Once that is figured out, the cost of decision making can be justified.

I suppose the next step would be the enumeration of tuples of current options and their expected outcome in the context of the desirable. This would include the default null set of dropping the desire itself. The default null set will be a part of the enumeration of all options in every decision making exercise. Hence, a person always has an option. Outcomes are only probability expressions and should not be detailed at this stage, lest the human mind starts making unconscious assumptions about various parameters which are yet to be defined. Any tuple that is not in the context of the desirable has to be discarded right now. Expected outcomes are different from desirable outcomes; in other words, expected outcomes can be negative courses as well.

The next step is to identify parameters affecting the course of action and any decision that will be made. Once the parameters are listed, the next step is to categorise them into: Immediate and Unavoidable, Immediate but Avoidable, Historic and insurmountable, Historic but surmountable, Impending, Expected but Avoidable, Conditional (i.e. will come into force only along certain courses of action).

The next step is to create meta-categorisation of parameters, which, in essence, helps us understand which parameters affect each of the tuples that we have enumerated. This categorisation is at a meta-level as it is not a real categorisation of the parameters themselves in their native form but in the context of the identified tuples. One cannot and does not assume that all parameters and all tuples have been identified before this stage. This stage facilitates identifying any new parameters/tuples and in clearly understanding the relationship between them.

The next step is in attaching lucre and probability to the categories and courses of action. Lucre is the weight attached to each expected outcome, taking into consideration the cost of functioning on the associated parameters in the context of achieving the desirable. Probability is the value assigned to each tuple which takes into account the amount of certainty in adopting a particular option and the likelihood of the associated outcome being realised while the parameters are being provided in their required substance (quality and quantity). Not always will the probability be available and needs to be statistically obtained or guesstimated in order to define uncertain paths. Often, when the probability is not available, it might be wise to chart the course of action (described later) assuming the best and worst case probability and guesstimating a probable mean. Depending on the worst consequences of not making a conscious measurement towards the decision (as mentioned at the outset), uncertain paths can either be dropped or considered on the side while adopting a more certain path.

Now the decision-maker has the required variables in reducing the decision making equation to provide maximum benefit to the stakeholder (who could well be the decision-maker). Resolution of the equations provide the courses of actions available to the stakeholder and a plan/strategy can be drawn based on them. A second iteration, on the courses of action can be run based on the strategic input in enhancing the nature of the tuples defined as well as the substance of the parameters involved. The two iterations will be separated by a phase of discussion and consulting with the stakeholder and associated parties.

I am frankly not too excited about mathematical models and am too busy to consider them right now, so I will not get into how the equations can be reduced and resolved right now.

What say? :-)

Now let me go and google for "decision theory"!! :-D