I'm currently finishing a first degree in CS, and I've been reading LW for a few months now (since June). I've read through most of the Sequences and check the front page of the site for anything that looks interesting whenever I want to put off doing something, which is usually several times a day. I also need to get round to finishing Godel, Escher, Bach some time (I'm kinda slow).

I am, at the moment, a terrible rationalist - my goals aren't even clearly defined, let alone acted on, and I have a strong background in tournament debating, which allows me to argue myself into believing whatever I feel like believing at any given moment. I think I'm getting better at that, but of course my own opinion is almost worthless as evidence on the subject.

On the other hand, reading this site (especially Yudkowsky's stuff) at least made me stop being religious. I like to think I'd have got there in the end anyway, but seeing as I really didn't enjoy it, I thank everyone here for pulling me out sooner rather than later.

Quick question: Does anyone know of a formal from-first-principles justification for Occam's Razor (assigning prior probabilities in inverse proportion to the length of the model in universal description language)? Because I can't find one, and frankly, if you can't prove something, it's probably not true. I'd rather not base my entire thought process on things that probably aren't true.

Hoping to be able to contribute,
Ezekiel

PS Good grief, there's an average of one introducing-yourself post every couple of days! Why the heck are all the front-page articles written by the same handful of people?

Does anyone know of a formal from-first-principles justification for Occam's Razor (assigning prior probabilities in inverse proportion to the length of the model in universal description language)?

0lessdazed9yWhat if instead of assigning prior probabilities to rules governing the universe
in inverse proportion to the rules' length, we assigned equal prior
probabilities to rules governing the universe and assigned probabilities to
states of the world based on the sum of the probability of each universe that
could produce that state of the world times the probability that universe would
produce it (as many universes would have randomized bits in their description)?
I think the likelihood of outputting a string of a hundred ones in a row would
then be greater than that of outputting
0001010010100110100010000100100010100100110101101000000101101111110110111101001001100010001011110000.
We could then revisit our assumption that in the rules' world, all are equally
likely regardless of length. After all, if there is a meta-rule world behind the
rule world, each rule would not be equally likely as an output of the meta-rules
because simpler rules are produced by more meta-rules; their relationship is as
that of states of the world and rules above.
This would reverberate down the meta-rule chain and make simpler states of the
world even more likely.
However, this might not make any sense. There would be no meta-meta-...meta-rule
world to rule them all, and it would be turtles all the way down. It might not
make sense to integrate over an infinity of rules in which none are given
preferential weighing such that an infinite series of decreasing numbers can be
constructed, nor to have effects reverberate down an infinite chain to reach a
bottom state of the world.

0PhilosophyTutor9yI suspect you will never find one. To get the scientific process off the ground
you have to start with the linked assumptions "the universe is lawful" and
"simpler explanations are preferable to more complex ones". Those are more like
mathematical axioms than positions based on evidence.
The reason being, you can explain absolutely any observation with an unboundedly
large set of theories if you are allowed to assume that the laws of the universe
change or that complex explanations are kosher. The only way to squeeze the
search space down to a manageable size is to check the simplest theories first.
Fortunately it turns out we live in a universe where this is a very fruitful
strategy.
ETA: I'm relatively new here: Whoever downvoted this could you perhaps explain
your thinking?

Hi, everyone.

I'm currently finishing a first degree in CS, and I've been reading LW for a few months now (since June). I've read through most of the Sequences and check the front page of the site for anything that looks interesting whenever I want to put off doing something, which is usually several times a day. I also need to get round to finishing Godel, Escher, Bach some time (I'm kinda slow).

I am, at the moment, a terrible rationalist - my goals aren't even clearly defined, let alone acted on, and I have a strong background in tournament debating, which allows me to argue myself into believing whatever I feel like believing at any given moment. I

thinkI'm getting better at that, but of course my own opinion is almost worthless as evidence on the subject.On the other hand, reading this site (especially Yudkowsky's stuff) at least made me stop being religious. I like to think I'd have got there in the end anyway, but seeing as I really didn't enjoy it, I thank everyone here for pulling me out sooner rather than later.

Quick question: Does anyone know of a formal from-first-principles justification for Occam's Razor (assigning prior probabilities in inverse proportion to the length of the model in universal description language)? Because I can't find one, and frankly, if you can't prove something, it's probably not true. I'd rather not base my entire thought process on things that probably aren't true.

Hoping to be able to contribute, Ezekiel

PS Good grief, there's an average of one introducing-yourself post every couple of days! Why the heck are all the front-page articles written by the same handful of people?

Maybe Kevin T. Kelly's work will fit your bill? Also see the discussion on LW.