one of the paper's more startling claims is that it's "cheaper" (in terms of information, i.e., the length of the program) to compute all computable universes than just one specific, arbitrarily chosen one. this is because the program to systematically generate every possible program (and thus every computable universe, via a process called dovetailing) is itself very short. however, if you wanted to compute just one universe that happened to have a lot of randomness or "noise" in it, the program for that specific universe might have to be very long because it would need to encode all that noise. he suggests our universe appears highly regular and compressible so far, meaning it's likely run by a relatively short algorithm. this view also offers a take on the many-worlds interpretation of quantum mechanics: different "worlds" aren't splits, but just different computations being run in parallel by the great programmer.
May 26 '25, Monday
May 27 '25, Tuesday
May 28 '25, Wednesday
May 29 '25, Thursday
RL vs SL: Core Differences 💡openwebui
Schmidhuber Paper Explanationopenwebui
May 30 '25, Friday
May 31 '25, Saturday
Jun 1 '25, Sunday
Open this page on desktop.
