A note on sternutation

Should some sort of post-mortem ever be conducted on the catastrophic failure of all computer models, it will be done with the help of a computer model, that will cost billions in whatever currency to assemble. It will show the need for more computer studies. And therefore, it will be catastrophically wrong.

But note: for 100 dollars or negotiable, I will produce a minority report that will explain everything, infallibly. I will not preview the report in this Idlepost, however, because it might be worth money to me.

Aw, heck. Since I am rich beyond the dreams of avarice, let me just go ahead and blow all the beans. Let me recklessly tell gentle reader why computer models are always mistaken.

It is because their makers decide the result, before they design the model.

This does not mean they are self-interested phanatics, consciously preying on the gullibility of a drooling, ignorant public; although usually it does. For even if, by disposition, they are lofty, objective types, they will need, objectively, a lofty budget to perform a “credible” study. This means they must beg huge sums of money, and this will only be available from a source with an unhealthy interest in the result.

You see, the problem has nothing to do with computers. Even among humans, the phenomenon of “garbage in, garbage out” is well attested. The intention of following the evidence where it leads, is transient. I should think only a saint could sustain it, for longer than he could hold his breath under water.

By the way, I have “seen” a computer model that incorporates all facts about Roman Palestine, in order to predict the resurrection of Jesus. I put the word “seen” in goof quotes, because, in fact, I “saw” it in the dream, from which I woke this morning. Many of my dreams are satirical; this one would take a cupcake, at least. I was turning page after page of print-out, from computer-generated numbers. Many of my satirical dreams are nightmares.

“Scientists” tell us, that in order to arrive at a secure result, we must know everything that will, or even might, contribute to it. Only then can we predict with confidence.

But supposing, for the sake of having an argument, that everything could be known, about anything, the prediction will still be wrong. That is because, “everything being equal,” everything won’t be equal. Statisticians have gone some way to proving, repeatedly, that there will be disparities between any two groups, no matter how identical they are. What they call “randomness” will sneak into some tiny movement, and tamper with everything that succeeds it, growing until the result is overawed. (Actually, it is worse than they think, but I am trying to keep this simple.)

The “butterfly sneeze” principle attempts to account for this. A butterfly sneezes in north-western Uganda, and there goes the history of the world. One thing leads to another. “For want of a nail, … the kingdom was lost,” according to a pre-scientific proverb I learnt as a child.

But here I am temporarily with the Hegelians. Butterfly sneezes do not determine the history of the world, or even the weather in Brazil. I can know this even without knowing how butterflies sneeze (presumably through their spiracles). For complex events are necessarily too complicated for human comprehension — given the time remaining in the universe. We cannot even know what we mean by “determined.”

A sneeze may however change the course of a computer model. That is why the modeller must insert however many sneezes it takes, until he gets the result he was paid for.


SEE ALSO, my Thing column for today (here). It also reflects on public statuary.