formal models and historical lessons

[_rck_]

Some notes from a discussion with JonnyDeftone and SheepHerder.

• The key advantage of general laws is to keep the theories small, which simplifies inference for man and machine.
• Occam's razor has a metaphysical and an epistolmological interpretation: Either the universe loves simplicity or the simpler theory is easier to work with.
• Given that any

Comments

[mijopo]

simpler theory is easier to work with

I'm not quite sure what you're claiming, but the epistemological interpretation has to be more than that the simpler theory is easier to work with, that's almost by definition what a simpler theory is. This would almost reduce to O.R. to a tautology. We have to factor in getting things right or something, don't we?

please explain

[_rck_]

I guess that depends whether Occam razor is used to distinguish between theories that otherwise seem to be basically equally plausible. I mean, if it is obvious that theory A is right and theory B is wrong, then I dont need Occam's razor to toss B and stick with A.

Or take the Kennedy assassination. I know almost nothing about it beyond what they show in the Dallas museum; I have no plausible theories about the event. So a theory A involving ten conspirators seems as plausible as a theory B involving fifteen conspirators. Having no other recourse then, I use Occam's razor to consider A the more plausible one.