Nomic: A Self-Modifying Game Based on Reflexivity in Law
The current issue of the New Yorker magazine (Dec. 10) contains an article ("The Checklist") about Peter Pronovost's checklist, which is like a pre-flight checklist for intensive care units in hospitals. The checklist is specific to one task: inserting a line into a patient. Peter thought that the probability of infection in a patient might be diminished by following a standard five-step process for inserting a line. So he instituted a five-point checklist for inserting lines into patients at Johns Hopkins, and the infection rate in the ICU plummeted to near zero. All the five steps were already well-known for years, yet they just weren't being followed. Peter is now working to institute his checklist in inner-city hospitals across the nation, with similar results.
The article reminded me of my first job in software engineering, which was writing the graphics software for the FAA's AAS program in the 1990s. The AAS program is the biggest poster child for failed software projects; it burned through billions(!) of taxpayer dollars before being shut down with little to show for it. Management tried to save the program by doing basically what Peter did for ICUs: institute standard software engineering procedures and checklists. The goal was to achieve some sort of ISO 9000 certification for the project by following Carnegie Mellon's Capability Maturity Model.
Both intensive patient care and large software projects involve much more complexity than the average human can handle; both fields are hardly understood today. This is why both fields have specialists and even super-specialists, because the techniques employed are much more an art rather than a science. So while reading the works of Mises (Human Action, Socialism, etc.), I wonder: what chance do we have with human government? We tend to elect the most average of men and women to craft our law codes, so don't we need at least some standard of excellence for them to follow? Is there a way to institute ISO 9000-like standards for a law code? A law code that is much more complex than, say, the Linux kernel? Will we need artificial intelligence to do it?
Artificial intelligence is another topic for another blog, but in the meantime thinking about the Linux kernel suggests one idea: Open Law. Now that we have the works of Mises and the communications of the internet, maybe we just need the "Linus Torvalds of law" to step forward and start the kernel of a rational legal code. The kernel might look much like the initial rules of Nomic, with its two-tiered system (like "kernel space" and "user space"). And various specialists in various types of law would need to review changes submitted to different modules of the law before the changes were accepted for the current "release."
Of course, the initial kernel release from "law-Linus" would be an entire new constitution, so adoption (never mind enforcement!) of such a code are probably impossible in human time scales.