[FRIAM] death

[Marcus Daniels]

Roger writes:

“It seems that this sort of dead code, undead code, zombie code problem is fairly ubiquitous in information processing systems.  No matter whose system, there are always things around that don't go away because nobody cared to do anything about them.  They always need a clean reboot eventually, or a clean reinstall, or some kind of purge to clear the inevitable cruft of just running too long.”

This is the sort of thing I’d expect to see in a naïve approach to a spatially extended UTM.  Without some process to clean it up, like dreaming, there would be more and more agents coming to contradictory conclusions.   When a call is made to vote on a decision there would much wasted motion in the cancellations where the heat of many weak learners might overwhelm the light from a few strong learners.

Marcus
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://redfish.com/pipermail/friam_redfish.com/attachments/20171030/6f58dd88/attachment.html>