From: Anders Sandberg (asa@nada.kth.se)
Date: Sat Mar 06 1999 - 09:03:19 MST
ASpidle@aol.com writes:
> Lately I've been noticing the efficiency of the good cop - bad cop
> model in several different contexts to the extent that I'm beginning
> to think that it is necessary to implement this model in any
> organization that is serious about excellence.
>
> Has anybody else noticed this? What examples can you cite? Are a good God
> and evil Satan mandated by natural law, or just good and evil, at least?
In reinforcement learning, you can either reward good behavior or
punish bad behavior. In some tasks in machine learning this results in
identical results: punishing an agent for remaining in a maze or
rewarding it for escaping has the same effect when used in machine
reinforcement learning. In this case it doesn't matter if you use a
Good or Bad Cop. But if the situation is not symmetrical (as in
vertebrate nervous systems) then this might actually help. However, I
don't think there is a natural law saying this is better, rather it
suggests that the information received from the environment is not
just scalar but has slightly more information.
-- ----------------------------------------------------------------------- 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 - 15:03:16 MST