Friday, March 27th 2020

Fri, Mar 27th 2020

No posts for March 27th 2020

Shortform

3Ben Pace2dHot take: The actual resolution to the simulation argument is that most advanced
civilizations don't make loads of simulations.
Two things make this make sense:
* Firstly, it only matters if they make unlawful simulations. If they make
lawful simulations, then it doesn't matter whether you're in a simulation or
a base reality, all of your decision theory and incentives are essentially
the same, you want to take the same decisions in all of the universes. So you
can make lots of lawful simulations, that's fine.
* Secondly, they will strategically choose to not make too many unlawful
simulations (to the level where the things inside are actually conscious).
This is because to do so would induce anthropic uncertainty over themselves.
Like, if the decision-theoretical answer is to not induce anthropic
uncertainty over yourself about whether you're in a simulation, then by TDT
everyone will choose not to make unlawful simulations.
I think this is probably wrong in lots of ways but I didn't stop to figure them
out.

Tuesday, March 24th 2020

Tue, Mar 24th 2020

No posts for March 24th 2020

Shortform

45dLearning theory distinguishes between two types of settings: realizable and
agnostic (non-realizable). In a realizable setting, we assume that there is a
hypothesis in our hypothesis class that describes the real environment
perfectly. We are then concerned with the sample complexity and computational
complexity of learning the correct hypothesis. In an agnostic setting, we make
no such assumption. We therefore consider the complexity of learning the best
approximation of the real environment. (Or, the best reward achievable by some
space of policies.)
In offline learning and certain varieties of online learning, the agnostic
setting is well-understood. However, in more general situations it is poorly
understood. The only agnostic result for long-term forecasting that I know is
Shalizi 2009 [https://projecteuclid.org/euclid.ejs/1256822130], however it
relies on ergodicity assumptions that might be too strong. I know of no agnostic
result for reinforcement learning.
Quasi-Bayesianism was invented to circumvent the problem. Instead of considering
the agnostic setting, we consider a "quasi-realizable" setting: there might be
no perfect description of the environment in the hypothesis class, but there are
some incomplete descriptions. But, so far I haven't studied quasi-Bayesian
learning algorithms much, so how do we know it is actually easier than the
agnostic setting? Here is a simple example to demonstrate that it is.
Consider a multi-armed bandit, where the arm space is [0,1]. First, consider the
follow realizable setting: the reward is a deterministic function r:[0,1]→[0,1]
which is known to be a polynomial of degree d at most. In this setting, learning
is fairly easy: it is enough to sample d+1 arms in order to recover the reward
function and find the optimal arm. It is a special case of the general
observation that learning is tractable when the hypothesis space is
low-dimensional in the appropriate sense.
Now, consider a closely related agnostic setting.

Monday, March 23rd 2020

Mon, Mar 23rd 2020

Frontpage Posts

Shortform

96dI have
[https://www.alignmentforum.org/posts/Ajcq9xWi2fmgn8RBJ/the-credit-assignment-problem#X6fFvAHkxCPmQYB6v]
repeatedly
[https://www.alignmentforum.org/posts/dPmmuaz9szk26BkmD/vanessa-kosoy-s-shortform#TzkG7veQAMMRNh3Pg]
argued
[https://www.alignmentforum.org/posts/3qXE6fK47JhSfkpnB/do-sufficiently-advanced-agents-use-logic#fEKc88NbDWZavkW9o]
for a departure from pure Bayesianism that I call "quasi-Bayesianism". But,
coming from a LessWrong-ish background, it might be hard to wrap your head
around the fact Bayesianism is somehow deficient. So, here's another way to
understand it, using Bayesianism's own favorite trick: Dutch booking!
Consider a Bayesian agent Alice. Since Alice is Bayesian, ey never randomize: ey
just follow a Bayes-optimal policy for eir prior, and such a policy can always
be chosen to be deterministic. Moreover, Alice always accepts a bet if ey can
choose which side of the bet to take: indeed, at least one side of any bet has
non-negative expected utility. Now, Alice meets Omega. Omega is very smart so ey
know more than Alice and moreover ey can predict Alice. Omega offers Alice a
series of bets. The bets are specifically chosen by Omega s.t. Alice would pick
the wrong side of each one. Alice takes the bets and loses, indefinitely. Alice
cannot escape eir predicament: ey might know, in some sense, that Omega is
cheating em, but there is no way within the Bayesian paradigm to justify turning
down the bets.
A possible counterargument is, we don't need to depart far from Bayesianism to
win here. We only need to somehow justify randomization, perhaps by something
like infinitesimal random perturbations of the belief state (like with
reflective oracles). But, in a way, this is exactly what quasi-Bayesianism does:
a quasi-Bayes-optimal policy is in particular Bayes-optimal when the prior is
taken to be in Nash equilibrium of the associated zero-sum game. However,
Bayes-optimality underspecifies the policy: not every optimal reply to a Nash
equil