AI ALIGNMENT FORUM
AF

Wikitags

Occam's Razor

Edited by Ruby, et al. last updated 4th Jan 2022

Occam's razor (more formally referred to as the principle of parsimony) is a principle commonly stated as "Entities must not be multiplied beyond necessity". When several theories are able to explain the same observations, Occam's razor suggests the simpler one is preferable. It must be noted that Occam's razor is a requirement for the simplicity of theories, not for the size of the systems described by those theories. For example, the immensity of the Universe isn't at odds with the principle of Occam's razor.

Occam's razor is necessitated by the conjunction rule of probability theory: the conjunction A and B is necessarily less (or equally, in the case of logical equivalence) probable than the A alone; every detail you tack onto your story drives the probability down.

Occam's razor has been formalized as Minimum Description Length or Minimum Message Length, in which the total size of the theory is the length of the message required to describe the theory, plus the length of the message required to describe the evidence using the theory. Solomonoff induction is the ultimate case of minimum message length in which the code for messages can describe all computable hypotheses. This has jokingly been referred to as "Solomonoff's lightsaber".

Notable Posts

  • Occam's Razor
  • A Priori
  • Decoherence is Simple

See Also

  • Solomonoff induction
  • Occam's imaginary razor
  • Priors
  • Burdensome details
  • Egan's law

External Links

  • Ockham’s Razor: A New Justification
Subscribe
1
Subscribe
1
Discussion0
Discussion0
Posts tagged Occam's Razor
32A Semitechnical Introductory Dialogue on Solomonoff Induction
Eliezer Yudkowsky
4y
2
19Instrumental Occam?
Q
abramdemski
6y
Q
13
21Occam's Razor May Be Sufficient to Infer the Preferences of Irrational Agents: A reply to Armstrong & Mindermann
Daniel Kokotajlo
6y
14
Add Posts