Saif

One of my colleagues, let's call her M, has a rather deep dislike of toes. At least that is the conclusion I would draw from the frequency with which she steps on them. It is not that she is malicious, but as she strives towards an unachievable level of perfection casualties are inevitable. Her frustration with one less enthusiastic or less intelligent (I like to think I am just "more chilled") is also just as inevitable. Suffice to say, I have to wear emotional boots with steel toe caps to meetings where she attends wearing her ideological stilettos.

While I am suitably armoured, other colleagues are less protected. Ash, (the poor fellow) providing a verbose and detailed outline of a problem he was facing, stood no chance.

Data Structure

"If you can't decide," M said helpfully, "you have either got insufficient data, or insufficient competence...which is it?"

"My problem is that I know a number of possible actions I could take, there may be others I haven't considered, and I am really not sure which way to proceed," said the chap, not knowing the danger his pedal digits were in.

"Perfect," said M to me, "lack of both data and competence."

"Come come, Ash," I intervened, "you can only do one thing at a time. Go with your gut. You will either get it right or learn for next time",

"'Go with your gut?' what kind of advice that?" M turned on me, "Surely at least some effort at objective analysis is better than an irrational decision based on what you ate last night?"

"M, the poor guy is just trying to apologise to his girlfriend. Whatever he does, chances are he is going to get it wrong, trust me." Ash's tactics typically reflect his passion for Liverpool Football Club, which usually dampens any kind of passion in his better half.

"Worst of all, I am not quite sure what I am apologising for, and I am too scared to ask." Ash would have wrung his hands is despair, if he knew what that meant, but his pain was obvious. "But, I have made a start. A list of options, to be analysed with estimates of cost, time, cheesiness, potential to backfire, humour etc...most of which remain as yet unknown"

"Your strategy should start with a tree said our resident know-it-all. "and there are many to choose from"

"Poplar?", I suggested.

"Very," said our sage, "Rational deterministic decisions are based on them, as opposed to non-deterministic ones, which rely on 'gut' or other obscure hidden black-box calculations, or a large database search which is both inherently incomplete while at the same time has data duplication. Trees are also easier to map to a reasoning engine".

"Sounds good," I countered, "but I suspect that there are more non-measurable and variably consistent parameters that shape the potential girlfriend-response-space from any decision that is made."

M turned sharply at me, making me instantly recoil "Excuse me?".

I quickly re played my words in my head...had I accidentally suggested that there may be gender-specific interpretation of rationality? No, no, no...I better stay quiet.

"A Tree can classify the options, with each branch offering up weighted sub options. A tree that is flexible,...", M continued.

M was on a roll, and we could just stand by, admire, and tuck our feet well under our chairs. Perhaps only a woman can truly understand another woman, and here is one trying to illustrate the remote possibility that, in truth, there is a genuine reasoning model underlying female thought. Ash and I gave each bewildered glances. Easier to explain the off-side rule we both thought silently.

Choosing a next move, whether it is what you need to say, buy, do etc is directed by that neural network that resides between your ears. Often it is modulated by other human or electronic neural networks driven by desires/designs to help you, exploit you, or even change you. The inner workings of these are often difficult to explain in the context of the domain of your decision dilemmas, but in the end something is chosen. Trees can help classify the problem , rationalise and add objectivity to the otherwise obscure algorithmic mechanisms based on statistics, thresholds, and digital noise in a structured systematic way. Perl is not short of modules that deliver this kind of data modelling, but it is trivial to "roll your own" use-case-specific module with abilities to add weights, serialisation/ deserialisation, restructuring, visualisation methods etc.