Template:M intro design Metis
“Der Tod eines Menschen: das ist eine Katastrophe. Hunderttausend Tote: das ist eine Statistik!”[1]
- —Kurt Tucholsky, 1925
If you owe the bank $100, that’s your problem. If you owe the bank $100 million, that’s the bank’s problem.
- —John Paul Getty (attrib.)
Metis
/ˈmiːtɪs/ (n.)
Accumulated heurism. A coinage from James C. Scott’s magnificent Seeing Like A State, metis is hard to describe — the Ancient Greek concept of “Μῆτις” combined folk wisdom, knowhow, Odyssean cunning — but in the corporate world it most resembles subject matter expertise. Experience, ingenuity, problem-solving, lateral thinking; smarts for figuring out what to do on the fly if you are in a jam.
Μῆτις was also one of the sea-nymphs, who became some kind of deity associated with wisdom, deep thought, and magical cunning. Which is nice.
Allen Farrington picks up Scott’s formulation of “metis” in the early exchanges of Bitcoin is Venice and notes how it stands in distinction to — in forlorn defiance of — high modernism which solves everything at scale, by abstract model referencing homogenised generalities in preference to the intricate, and inconvenient, particular.
This is the difference between a top-down, averagarian, view of the world, where everything is rendered roughly into broad categories and the challenge is achieving scale. and the subject matter expert’s view, where difficulty presents as idiosyncrasy.
The administrator knows that the portfolio “risk” for a given period is, say, 5 percent. She succeeds if she can “manage the portfolio” over the period to suffer an overall actual loss of less than that. [2]
An individual risk manager manages a single instrument with a given risk of 5 percent and succeeds if she can avoid that risk altogether. For the individual risk manager, there is no 5 percent loss. The loss is either nil or 100%.
A portfolio with 100 managers each managing a single instrument for which they are fully responsible throughout its life will lead to different decisions throughout the life of each instrument. Including the decision to invest in the first place.
Here is the equivalent of the averagarianist’s category error. We derive a mathematical property from an observed set of events — this is simple computation — that of a group of 100 mortgages, five did default — and make two invalid extrapolations. Firstly, that in any group of 100 mortgages, therefore, five will default, and secondly, each mortgage in that portfolio has a probability of defaulting of 5 percent.
Both conclusion is obviously barmy unless you have no better information on which to judge individual loans.