One of the many main topics of this letter is “regulation”, in the abstract. Not just bank regulation, environmental regulation and indeed not just “regulation” at all. Because although it usually has the meaning “secondary legislation affecting an industry”, that’s not really what regulation is.
It’s a word whose definition has historically tended to meander through the areas I’m interested in, crossing and recrossing the boundaries of social and physical sciences. Etymologically, you could argue that yes it does just mean lists of rules. But of course, etymologically a rule is a straight piece of wood; the use of the same word to mean a social framework that keeps us on the straight and narrow is itself a metaphor.
But in any case, a system is “regulated” if it’s controlled in such a way as to ensure that its behaviour stays within set limits. At a sufficient level of abstraction, you don’t necessarily care whether that regulation is happening by physically restricting outcomes to a permissible range, or by social and legal arrangements giving human beings the responsibility for ensuring that things don’t get out of control. The “regulator” could be a little metal doohickey on a steam engine, or it could be the Securities Exchange Commission – all that matters is that it’s the thing that ensures the system remains regulated.
At this level of abstraction, one nice little piece of maths you can make use of is the Ashby-Conant “Good Regulator Theorem”. It’s a bit controversial; the authors kind of overstate its generality, but what it proves is that (for some reasonable-sounding definition of optimal) the optimal regulator for any system is one which includes a model of the system it’s regulating.
This last bit is what makes it controversial and brings in the question of generality – it all depends on what you mean by “a model”. Ashby and Conant just use the term very generally to refer to “any mapping between states of the world and states of the regulatory system”, and allows the model to be “implicit” in the design of the system, which feels like kind of a cheat.
But it does highlight an interesting and not necessarily obvious connection between the two concepts. There is some sense in which a regulatory system has to work on a representation of reality, and a failure of the regulatory system is a failure of that representation – a “Black Swan” or a Damned Thing, a fact that isn’t consistent with the stylised facts.
Let’s make this more concrete. Even the simplest regulatory systems imply models. “STAFF HAVE NO ACCESS TO THE CONTENTS OF THE SAFE” suggests a specific model of who might walk in and ask the staff to open the safe. “NO REFUNDS – NO EXCEPTIONS” is a model of how refunds work, and it’s one which fails and becomes unregulated when someone walks in with an obvious or statutory right to a refund.
If there’s a correspondence between the two things, then that’s potentially something that can be used. A problem with the model – a material way in which it doesn’t correspond to reality – is going to show up at some point in the future as a problem of the regulatory system; a Damned Thing which will crop up and cause the system to become unregulated. And to an important extent, history is what happens when things become unregulated, and when higher levels of the system have to step in to bring them back under control.
— [Programming note! I have been informed that some of the people who pledged subscriptions before I started offering them are showing up as not having subscribed. There are no benefits to being a paid subscriber (yet, and tbh I’m not currently planning any but might do some if I start feeling guilty) so you haven’t been literally swindled, but even so, this is awful. If you’re in that situation drop me a line and I’ll start investigating what to do about it. Also if I promised you a free sub and haven’t given you one, it’s because it’s a very small amount of effort to do so, which means that wheels are getting oiled in order of squeakiness, sorry.