firstname.lastname@example.org (Ilias Kastanas) says... > >In article <email@example.com>, >Daryl McCullough <firstname.lastname@example.org> wrote: >>email@example.com (Ilias Kastanas) says... >> >>[Deleted description of my model of robot evolution whereby those >>with inconsistent theories are "killed off" and those with more >>powerful theories can "paralyze" those with less powerful theories >>(that is, prevent them from reproducing).]
>>Your observations are perfectly correct. There will always be inconsistent >>theories floating around (just as there are disadvantageous mutations >>that develop). And there will be perfectly consistent theories paralyzed >>by inconsistent theories. However, a theory paralyzed by an inconsistent >>theory will eventually recover (when the inconsistent theory dies). >>A consistent theory can never be permanently incapacitated except by a >>more powerful consistent theory. > > > > I think it can; T1 is permanently paralyzed! T2 paralyzes T1; before > T2 is killed, a T3 paralyzes T1; before T3 is killed, a T4... and so on.
Okay, we patch it this way: if T1 is paralyzed by T2, and T2 is killed, then T1 gets a "grace period" of a couple of years, during which T1 cannot be paralyzed (although it can be killed).
> It seems some additional assumption is needed if progressively stron- > ger consistent T's are to arise.
I think the "grace period" idea would accomplish this.
>And in any case we cannot pick them out among the crowd!
No, of course not. The strong theories thrive and the weak and inconsistent ones don't, but you can never tell, given a finite viewing time, whether a theory is thriving or not.