Hacker Newsnew | past | comments | ask | show | jobs | submitlogin

https://en.wikipedia.org/wiki/Project_Cybersyn

(Do you seriously believe that a field as nascent and murky as macroeconomics, coupled with the hectic state of public policy and political advisory -- that these same people can then turn around and write a magical AI to quick fix government planning as we know it?)

I mean, what would go a long way is manufacturing a programming language to write laws in, so we could lint and compile time warn/error laws. That would be a good place to start.

The law already is a formal language, and court procedure is also formalized and mechanical. The language of the law is such that it permits a high degree of fuzziness, so one can benefit from common law, case precedent, jury nullification and all sorts of nice judicial perks.

The "programming language for laws" that people on HN keep proposing will only fuck everything up even worse.



> benefit from common law, case precedent, jury nullification and all sorts of nice judicial perks.

for the right class of people. the same "perks" turn into something much more sinister for the "wrong" sort of people. whether those be blacks or browns. the point is to have one set of laws that applies logically to all situations and to all people equally. Adding that as a constraint will either make the laws much worse or much better.


the point is to have one set of laws that applies logically to all situations and to all people equally.

This is already the case. Codifying laws into some sort of PL is a complete diversion and redundant.


Then why do different courts come to different decisions then if this is already the case. Codifying laws into some sort of PL exactly intends to prevent this.


> Then why do different courts come to different decisions then if this is already the case.

This is a quirk of the US Federal legal system -- independent circuits. Any sufficiently important matter is settled nationwide in US caselaw by the Supreme Court.

The law is incredibly consistent. You will, without recognising it, have hundreds or thousands of invisible interactions with it daily. It is incredibly robust in the face of massively variable outputs. No software product has ever come within a cooee of sustaining decades to centuries of uptime at a stretch in the face of sustained attack.


You don't know what you are talking about. The law is in no way consistent. The legal system is full of bugs and fails often. It ends peoples economic and social lives often enough. The only thing that keeps it in place is the lack of a better alternative and power.


> No software product has ever come within a cooee of sustaining decades to centuries of uptime at a stretch in the face of sustained attack.

If you consider each change of a law as a break of the uptime (I mean: there is some "bug" in the laws, so it's changed), the uptime of, say, good embedded software is far better.


Show me an embedded system with live patching and as large a problem domain and let's run that comparison again.


The AXD301 switch

> http://ll2.ai.mit.edu/talks/armstrong.pdf (from slide 27 on)

> https://en.wikipedia.org/wiki/Erlang_(programming_language)#...

"In 1998 Ericsson announced the AXD301 switch, containing over a million lines of Erlang and reported to achieve a high availability of nine "9"s."

Additionally it is well-known that the AXD301 has live patching capabilities (which is not surprising, since Erlang has).


Does the AXD301 switch handle trusts, estates, criminal punishment, taxes, court procedure, torts -- indeed, every single interaction between humans and the environment that they inhabit, including into the past, into the future, into space, under ground and sea, for concepts both physical and abstract, in a heterogenous cooperative world-spanning framework applicable to all human beings, animals, plants and objects, natural or artificial, living and dead?

Don't get me wrong: software engineering has accomplished great things. My point is that the law has to face different, very fuzzy, unconstrained problems that cannot be simplified.

(Also I would've used the Space Shuttle software, that's my favourite high water mark. Or the seL4 kernel).


Which is one of the reasons why we should not make it deterministic in the way you seem to want. We need a variety of decisions so that we can explore to find the least bad one.




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: