Transaction cost economics vs contracting theory, a primer
Vitalik Buterin: I’d be interested in reading this argument :)
This is a little bit of inside baseball, but…
Transaction cost economics and contracting theory are very much tilling the same ground, at the intersection between law and economics, with the key distinction being that TCE theory is almost exclusively verbal in the legal tradition, while contracting theory builds on a mathematical, game theory–driven economic framework. In fact, Hart started on the topic to make Williamson’s “informal” theory “more formal”.
The problem with using formal models in contracting theory is that you have to start with an assumption that at any time we are able to express the state of the world — including future states of the world — in an orderly parametric way. The world becomes an orderly, dimensionalizable and measurable vector space.
In any meaningful contractual situation this assumption, that we can costlessly vectorize the world, is nonsense. This is not a scientific term, but I call this the loss of ambiguity. If you remove ambiguity from a contractual situation, you can easily model it mathematically, but you lose 80% of what made the problem so relevant in the first place.
A big problem is that mathematical economists, who have been doing things this way for decades, don’t even notice this loss anymore.
In contracting theory this means you have to build models where the contracting partners engage in pre-contract negotiations based on a small number of random variables, assume orderly probability distributions, let the partners engage in a contract based on their (orderly) assumptions, and then disclose the “true values” of the random variables to the participants after contracting, upon which they might engage in some kind of reneging. This is very much the setup of Grossman & Hart 1986, Hart’s first contracting theory paper.
In terms of the The DAO storyline, Hart’s position is “Code is Law”, while Williamson’s position is “Intent is Law”. Writing mathematical (algorithmic) contracts makes sense in a world where we are sure that we can express all (past, present or future) states of the world in a mathematical fashion. If not, verbal language, which is inherently ambiguous, gives us this cushion of ambiguity that allows us to even write a contract of finite length in a world of an unforeseeable future.
It would have been perfectly fine for the folks from slock.it to couch The DAO in a verbal statement of intent, and ex post they clearly indicated that such an intent existed in their minds and overrode their faulty code. This would have been perfectly fine, the big problem they created for themselves is that ex ante they expressly made clear that Code is Law, and they reneged on that.