From: Adam Foust (agfoust@tds.net)
Date: Tue Dec 30 1997 - 22:41:44 MST
"Alexander 'Sasha' Chislenko" <sasha1@netcom.com> wrote:
> With that, I am ending the process of writing this message, and it's
> OK, because it has fulfilled its intended goal. It can die now.
> Other processes may later start their own typing threads partially
> based on this one.
Although... if I launched parallel threads to write messages, terminating
the threads on completion of each task... years and millions of messages
later, I'b be no better at writing messages. Since none of the writing
experiences would be with my core thread, I might even be progressively
worse over time.
Also, if I (as a a sequential thread) were to consider a given task fun or
interesting I would want to undertake the task myself (no forking),
experiencing and reinforcing the activity, thereby changing my internal
state. Naturally, I would want to fork parallel threads to complete less
interesting tasks, and optionally terminate the threads upon completion.
For a single long-lived thread I think over-specialization could occur as
the "interesting" tasks were pursued, experienced, and reinforced... and
other less interesting tasks were delegated. The problem is that the forked
parallel threads would be progressively specialized as well and would be
increasingly ineffective/inefficient at caring out the less interesting
tasks. Many self-directed threads would come to rely on foreign threads for
those tasks they could no longer perform efficiently (so a market for
services would have to exist... and a currencies). Many specialized
threads would frequently dead-end in their usefulness and expire. The
catch: the highly specialized threads would be the most efficient at
performing a given task. Ouch. The super-specialized threads might be
narrowly focused enough to be considered non-sentient automatons.
Launching and terminating threads to carry out certain tasks presents some
interesting possibilities. One application of this would be in generating
self-diagnostics. You could submit your forked copies to a battery of
tasks/tests and analyze the outputs to quickly identify areas that require
improvement, for example.
Other things come to mind, but I'm getting way out on a limb here.
[Adam Foust] * agfoust@farthest.com * http://farthest.com
This archive was generated by hypermail 2.1.5 : Fri Nov 01 2002 - 14:45:17 MST