From: Anders Sandberg (asa@nada.kth.se)
Date: Wed Dec 20 2000 - 15:31:37 MST
GBurch1@aol.com writes:
> I love it when bright "amateurs" figure things out from first principles and
> come to the same conclusion that "experts" do. In this case, Anders, you'd
> done something I've seen happen a number of times since becoming involved in
> the transhumanist community: You've invented the Anglo-American common law of
> "products liability" and identified one of its key challenges. Here, in two
> paragraphs, you've recapitulated 200 years of the development of this
> jurisprudence and arrived at the frontier of a field that absorbs a legion of
> good minds in economics and law.
Thanks! That felt good! I vividly remember when as a child I managed
my first geometrical proof (the area of a parallelogram) and then
discovered that somebody Euclid had beaten me with a few millennia.
> The challenge is indeed how best to use the clear benefits of the fine-tuned
> feedback of the case-by-case common law liability mechanism. The superiority
> of this system to the Continental system of a-priori precise legislation and
> administrative regulation is clear, since it uses an on-going process of
> clarification and adjustment of standards to evolving circumstances in a way
> that's more insulated from temporary political pressures than the civil law
> system is.
Anglic empirism versus continental rationalism?
So the main issue is how to set up the feedback so that it adapts at
an optimal rate, and also adapt the "strength" of the corrections it
sets up to avoid too little or too much feedback. Oh, it is just a
simple optimization problem, as a friend once said about economy :-)
Actually it sounds a bit like the learning problems we are studying in
neural networks and the brain - how does a system set its learning
rate so that it is optimal for a given environment, and adapt it fast
enough (metalearning) when the environment changes. In this example
the individual cases and market provides the first level of learning,
setting the price for various risks. Then there is both some
adaptation where various cases and political decisions influence the
strength of the feedback and adaptation to changes in the amount of
super-mega-risk?
The problem seem to be that while there is no shortage of learning
material on the first level (the everyday cases and transactions),
making learning relatively efficient, you get far less learning and
more reliance on the a priori methods on the metalevel.
Maybe it could be improved if there was a better way of accumulating
and learning from such decisions on a global scale (a bit like how I
think politicians might have use for a database showing the results of
decisions similar to ones they are considering). This is of course the
job of the historians.
-- ----------------------------------------------------------------------- Anders Sandberg Towards Ascension! asa@nada.kth.se http://www.nada.kth.se/~asa/ GCS/M/S/O d++ -p+ c++++ !l u+ e++ m++ s+/+ n--- h+/* f+ g+ w++ t+ r+ !y
This archive was generated by hypermail 2.1.5 : Fri Nov 01 2002 - 15:32:29 MST