[FoRK] Do You Belong to The Church?

J. Andrew Rogers andrew at ceruleansystems.com
Thu May 6 09:25:15 PDT 2010


On May 6, 2010, at 3:30 AM, Bill Kearney wrote:
>> The uglier the set of equations you have to balance, the more
>> exotic the tools you have to invent to keep it manageable.
> 
> Or, more likely, the more exotic the criminal enterprise you have to operate in order to steal.


The methods criminals use don't change, they are just applied in different contexts.  Byzantine regulatory systems force legitimate businesses to operate in such a way as to create opportunities for the crooked ones. Regulations routinely create exploitable opportunities -- you have black hats and white hats studying them -- but in some cases there is relatively little that a legitimate business can do about an exploit because they are required by law to conform to the regulation.


Adding a layer of complexity on top of this, finding and using these exploits can be automated and inadvertent.  Many years ago, I worked on a "real-time" system that did dynamic topology optimization across networks of physical business resources, legal contract terms, and regulations.  It loosely built a computing model of the entire market from the business resources all the way up the stack through the multinational regulatory framework and attempted to optimize the efficiency and profitability of the market. The purpose wasn't just to maximize profitability per se but to also detect and mitigate emergent pathologies in the market (e.g. gross mis-allocation of resources due to a market modal shift).

A number of lessons emerged from that experiment. First, at a macro level, regulatory (and contractual) complexity increase the number of modes a market has that must be accounted for; computerizing it made it much clearer how many significant modes the market had.  Markets are not static things, they are dynamic multimodal systems that have to be adapted to in order to not find one's self on the wrong side of a contract. Second, at a certain level of scale and complexity it is not possible for a human operator to properly analyze the state of the market; the system in this case was the first attempt to take decisions out of human hands. Third, a computer system designed to do little more than dynamically optimize the use of resources in a regulation-aware environment can find and use most of the inconsistencies and bad designs in legal frameworks without human knowledge or intervention. If your regulations suck, you really can't blame people for noticing that fact.

The above case was not a hedge fund or even a financial company.  It was implemented for a state-owned industrial company from a social democracy trying to improve the efficiency of their business. By modern standards the capability was crude; real quant hedge funds could do much better. 


It seems that a lot of people want regulatory "security through obscurity", not because they are obscure -- they can't be -- but by forcing everyone to simultaneously blindly follow the regulations *and* without accounting for the existence of the regulations in any way.  That perspective of reality is delusional. The better alternative would be to design a set of regulations that are not trivially exploitable with simple analysis.  

Regulations are like software code to the extent they are amenable to logical analysis. It is much easier to make strict simple designs behave robustly than mountains of inconsistent spaghetti code.




More information about the FoRK mailing list