seeing like a right old state

all the things left out from elsewhere

I found that yesterday’s post on James C Scott’s “Seeing Like A State” was growing out of control, so I cut a lot of material out of it, thinking “I’ll use that another day”.  Well, it is another day, plus I realised that I have a terrible track record on thinking I’m going to write things then forgetting about them, and also several readers asked questions that were relevant, so I thought I’d do “part 2” of seeing like a state.  Here goes:

Ben Recht pointed out that I had kind of brushed off Scott’s distinction between techne and metis and taken a rather quick way with asserting that of course a good bureaucracy can deal with information that isn’t quantified and tabulated.  Can I defend this?

I think so, by expanding a bit on a point that was left rather deeply implicit in the previous post but which I harp on about endlessly in my book.  Basically, good bureaucracies have plenty of metis embedded in them.  Although their main conduit of information is the tables and reports of techne, a functioning bureaucratic state will have layers of commissars, beadles, Mssrs. Les Maires, sheriffs and whatnot, who live among the subjects.  They’re consequently aware of the ground truth on which the reports are based. 

Most of the time, their presence isn’t an important part of the system, because a well-organised techne will describe the ground truth pretty closely, and so they have nothing to do beyond passing on their reports and implementing their orders.  But functional organisations work on a “management by exception” basis – a well designed bureaucracy has channels for the people who understand ground truth at the highest level of resolution to intervene when the techne starts misdescribing reality importantly, to either depart from the bureaucratic plan, or to communicate with the higher levels to bring more capability to bear. 

Consequently, I think I am, to give a straight answer to Ben’s question, committed to the view that in an important sense, there’s actually no such thing as metis.  Just as there’s no real answer to questions like “how long is the coast of Britain?” or “how much copper is in this rock?” (the answer to both is “it depends how closely you’re prepared to look”), “metis”, in my view, really just means “everything that didn’t seem worth the effort to turn into techne”.

And that brings me on to another important question asked, as to whether the anecdote about China’s one-child policy really showed anything at all.  In my view (and definitely in Stafford Beer’s too), a huge part of designing a functioning bureaucracy is the design of feedback systems that ensure that it’s going to be able to bring the right amount of information-processing capacity to bear in the right place at the right time. (And, as I always find myself emphasising largely because I forget it myself too often, communicating the results of that analysis to a place where they can be useful, in the right form and on time).

Some people went a long way with this concept in the direction of “second order cybernetics” (the application of cybernetic principles to cybernetics itself), but in my view that’s a rabbit hole.  All we’re talking about here is “the proper design of control systems, taking into account that they will have to change their behaviour in response to changing circumstances”, and we don’t say that someone’s doing “second order plumbing” when they put a safety valve on an espresso machine.

What went wrong with the one-child policy was indeed that someone had all the right information, but used it in a dumb way and made a bad decision, with disastrous consequences.  And of course, James C Scott is right to say that “someone had all the right information, but used it in a dumb way and made a bad decision, with disastrous consequences” is a large part of the history of bureaucracy.

But it’s not an inevitable part.  A much larger part of the history of bureaucracy is “someone had all the right information, but used it in a dumb way and made a bad decision, but then the mistake was noticed and corrected so nothing too bad happened”, but of course these cases are never recorded.  This is close to being a definition of what management cybernetics is all about – the attempt to achieve what Ross Ashby called “ultrastability”, the ability of a system to reorganise itself and return to stability after a shock which was not anticipated by its designers.

A further question raised was “what the hell are you on about with this Hayek stuff”, but that one’s going to have to wait until I get back from holiday and am able to retrieve my copy of “How Many Grapes Went Into The Wine”.