Signal-to-noise ratio
|
- Caught in a mesh of living veins,
- In cell of padded bone,
- He loneliest is when he pretends
- That he is not alone.
- We’d free the incarcerate race of man
- That such a doom endures
- Could only you unlock my skull,
- Or I creep into yours.
- Ogden Nash, Listen...
- In God we trust, all others must bring data.
At the foot of Deming’s fashionable quote, one can lay a great deal of responsibility for the dogmatic madness our age.
That it is woven into Dominic Cummings lanyard might tell you something.
For if we we take it as a given that, to the best of our knowledge, the information content of the universe, through all time and space is infinite, and if we also recognise as a truism that the data collected (or indeed generated) by homo sapiens to the point of reading is finite, then it follows that the total value of data in which Deming would have us trust is, mathematically, nil.
And that is before one considers the apparent quality of the data we have gathered — we hear 99% of it originates in the internet age, so a good portion is cat videos and hot takes on Twitter — in its own terms. But leave the the banality of our age to one side — we don't need it to make out the argument, and I'm sure a lot of those people on Twitter think they are very clever.
In any case, if we transcend our meagre hermeneutic bubbles, the signal-to-noise ratio of our data is infinitesimal.
Just on that statistic you might wonder what is so wrong with God. But human beings are pattern-seeking machines. We don’t take the data as we see it cold, and fashion objective axioms from it, carving nature at its joints: we bring our idiosyncratic prisons and pre-existing cognitive structures to it and willfully create suitable patterns from it to support our convictions. This is not a criticism, but an observation. This is the doom our incarcerate race endures.
It is not just the Twitterati. Science, too, has its confirmation bias, that subsists at a meta-level beyond control by double blind testing methodologies. Experiments which confirm there hypotheses are a lot more likely to be published than those which don’t. of those failed experiments that are published fewer are cited in other literature. Falsifications die.
This is no cause for alarm and nor is it new. It is is a reminder to of the importance in contingency, provisionality, and above all humility.
All of these are another way of attacking a familiar problem: the world is complex, not merely complicated. Complication is a function of a paradigm. It is part of the game. It is within the rules. It is soluble, by sufficiently skilled application of the rules. Complication can be beaten by algorithm.
Complexity cannot.
Complexity, by contrast, describes the limits of the paradigm. Complexity is the the wilderness beyond the rules of the game. Complexity is is the noise, not the signal. In a complex environment the rules do not work. This is why my physical sciences apparently have a greater success rate than social sciences: physical sciences generally address the Gaussian behaviour of independent events — that is to say, they explain complication; social sciences have to deal with the inherently complex, non-Gaussian interactions between human beings.[1]
- ↑ physical sciences such as physics set up hermeneutic systems within which their rules will work, and often these systems are are dramatically simplified: Newton for example assumes a frictionless, stationery, stable, neutral frame of reference: circumstances which, in in any observed environment, simply do not exist. That's variances between Newton's prediction and and observed outcome can be explained not by falsification but as “contaminations” of the ideal experimental conditions. Hence, the proverbial crisp packet blowing across St Mark’s Square.