On 2020-02-06 10:50:AM, Matt Mahoney wrote:
On Wed, Feb 5, 2020, 7:25 PM TimTyler <[email protected] <mailto:[email protected]>> wrote:On 2020-02-05 13:22:PM, Matt Mahoney wrote or quoted:What do you think is the reason Occam's Razor works, if not math?Well, math permits worlds not bound by Occam's razor, such as the world where all sequences of a given length are equiprobable. So, the answer is probably physics. Locality and the speed of light limit, in particular.Math allows strings of equal length to be equally likely, which is what a Solomonoff distribution assumes. It does not allow a uniform distribution over an infinite set of strings of different lengths.
To be clear, I was describing a world of random noise. Occam's razor doesn't apply and Solomonoff induction is totally useless. Evolution wouldn't work, but we can fairly easily create such worlds for existing agents using technology such as virtual reality. That shows that Occam's razor is based on physics and/or the initial conditions of the universe - and doesn't follow from math alone.
Occam's Razor suggests that the origin of the laws of physics is that all possible universes exist and we necessarily observe one where intelligent life is possible.
That would be a mis-application, though. We know Occam's razor applies in our visible universe. We don't know that it applies in other larger spaces. Such use of Occam's razor would be purely hypothetical.
So Occam's Razor drives physics, not the other way around.
I snipped some of your argument for this, but the conclusion doesn't follow. We don't know that "Occam's Razor drives physics". That's a hypothesis, and while we can't get out of our local region and escape from what appear to be our physical laws, it is an untestable one, and so is of little interest. -- __________ |im |yler http://timtyler.org/ ------------------------------------------ Artificial General Intelligence List: AGI Permalink: https://agi.topicbox.com/groups/agi/T353f2000d499d93b-M7ab67b198cc607abe36632ec Delivery options: https://agi.topicbox.com/groups/agi/subscription
