Msg Len

post by Zack_M_Davis · 2020-10-12T03:35:05.353Z · score: 47 (28 votes) · LW · GW · 4 comments

I'll be brief, omit needless words.
Intelligence is prediction is compression because
Compression is finding a code that makes the data shorter
And codeword lengths are probabilities
So codes are probability distributions
But probability distributions are prediction strategies.

4 comments

Comments sorted by top scores.

comment by NunoSempere (Radamantis) · 2020-10-17T10:53:41.263Z · score: 3 (2 votes) · LW(p) · GW(p)

And prediction strategies are almost optimization procedures?

comment by Tyrrell_McAllister · 2020-10-12T14:02:21.305Z · score: 2 (5 votes) · LW(p) · GW(p)

Did your really need to say that you'd be brief? Wasn't it enough to say that you'd omit needless words? :)

comment by Zian · 2020-10-12T19:38:12.973Z · score: 3 (2 votes) · LW(p) · GW(p)

But then he'd lose the Strunk and White allusion.

comment by Gurkenglas · 2020-10-12T10:40:11.452Z · score: 2 (1 votes) · LW(p) · GW(p)

I approve the haikuesque format.

Do you agree that the "bijection" Intelligence -> Prediction preserves more structure than Prediction -> Compression?