Decision Analysis and Other Ways of Thinking
Blake's Books: Click on book to access for free
2012: ETA- Science Commentary and Sci-fi Stories
From the Agenda- Scifi Fantasy about Dolphs
In A Far Traveler we meet Stephanie Noble, the blonde, turquoise eyed, knock you over and stop your heart gorgeous nurse; she's on the run from a high pressure job and bad relationships. Later we'll learn that she is a math prodigy, her father's death forcing her to drop from medical school to nursing. thereby graduate more quickly and provide financial support for her siblings college educations.
Though she graduated from the UCSF Medical School nursing program, the Director of Research discovered her math propensities and she was brought onto his staff to help solve the complex problem of determining which branches of research the hospital should fund. She quickly rose to Assistant Director and was sent to Stanford, just a few miles south, to learn the emerging science of Decision Analysis directly from its pioneer, Herr Doctor Theodore Hofmann. Research shows that she is listed as a contributor on many of his early papers. With her education and native skill, she became the de facto consultant to the UCSF governing board, earning the animosity of the director ; add that to an amorous relationship gone sour, and we meet her on a road trip in Nevada, traveling with the narcissistic Janet, and seeking to escape to a simpler life.
It doesn't take us long to figure out that Herman also has a unique way of thinking, one that shockingly pervasive, eliminating his ability to have true emotions. This mathematical devise is, to the best of our knowledge, a form of Monte Carlo Simulation; the most important part of this statement being "to the best of our knowledge", since we can't get into his head and see what is going on. The fact is; however, even more chilling, his emotional response is an accumulation of Neuro Net Analysis.
Decision Analysis - A quick description and example:
Problem: You want to decide whether to buy up competitor air routes during an airline down cycle
Situation: The airline industry is down sizing and we have identified two factors which we think are driving this: One, our economy has run away inflation due to government monetary policies, enacted by the Carter Administration. Two, there is a shortage of fuel, due to the energy policies of the said administration; that policy being that fuel prices have been frozen at a level the public has grown to expect over many years of petroleum abundance; however, current fuel production requires more difficult and expensive techniques to produce marginal petroleum deposits. The price freeze has forced the closure of marginally producing units, since their production costs are higher than the government regulated sales price. As a result there is gasoline rationing, and citizens are lining up at gas stations, hoping to buy gas before the station's ration runs out. This and high home mortgage rates, in the range of 11-13%, are making the average US citizen wonder if the Democratic Administration has a clue, or if the environmentalists are right, we are actually finally running out of oil.
Decision Process: You decide what factors are important to your decision, this necessarily being done in a group session of all the important decision makers, so that there is buy in on the process and the parameters used to make the decision. Having agreed on the factors involved, you organize a branching tree that connects the factors in a chain of cause and effect, ending in a final set of possible outcomes. It then is necessary to determine an array of possible values for each parameter in the process, again group agreement is paramount. Since could be a very involved process, we'll only look at two parameters, so you can get an idea how it works.
Parameter 1: What was the petroleum production prior to the government price regulation? This is easy; the data for some prior period of time, say 3 to 10 years, can be gathered from various commercial sources. This data should cause very little debate among the decision makers since it is a matter of record.
Parameter 2: How likely is it that the current administration will be voted out in the next political cycle, two years from now, and that the opposite party will remove fuel price control? This cannot be determined from hard historical fact, but requires guessing about the future. You can bring in experts, like Rasmussen or Gallop poll experts, who can tell you what the current public opinion is and then give their expert prediction, Delphi (from the gods), about what the future holds. Your decision makers will have to agree to accept or modify these Delphi predictions. Let's say that Rasmussen tells us that 70% of the people are very upset and will vote Carter out, unless some international event transpires during the next two years, possibly the fall of the Shah of Iran, in which case the number may drop if Carter handles it well, or it may get worse, if he does not.
Once all of these factors are in place, it becomes a matter of multiplying the stream of probabilities and determining the likeliness and value of each final outcome. The beauty of this process is that each incremental factor is debated and decided upon by the decision makers, making the final decision clear and well understood, and therefore well supported by the team.
Monte Carlo Simulation is more obscure, in that a computer runs a statistical model, based on the probabilities of the parameter inputs, these being achieved in the same fashion as in Decision Analysis, or not. The statistical model can be run any number of times, but 1000 is a small number, and 10,000, not an unreasonable number. The results come out as a continuous profile of possible outcomes, and the user has to decide which parts of that are important and which might be ignored.
Neuron net models the world, or any aspect of it, and changes its model as it gets more information, i.e. the computer learns from input.
There is software available for all of these processes and more; do a Google or Yahoo search to track it down. Professional versions are expensive, but in most cases, cheap student friendly versions are available. If you like to play with this kind of math, I strongly suggest that you buy one or more of these modeling tools and play with it. I prefer Monte Carlo, and no, I'm not a Seriphrim model one. If you come up with a good stock investment modeling program, please email me a copy.
Reading sources in the reading list.
|Related Authors Links: Baigent, Butler, Eisenman, Hancock, Knight, Leigh, Lomas, Van Dšniken|
This site is under construction ( and always will be ), however we are currently just Starting. links will be added soon.