Perhaps I'm missing something, but I don't get the part where Eliezer is claiming anti-reductionist worlds as inconceivable. I can conceive of a world, not our own, with entities exhibiting complex behaviors that can be described by high-level models. For example, Conway's game of life. It's a nicely reduction-compatible world. The entities that are present at a medium level (gliders, oscillators, guns, etc.) are formed from low-level cells. The entities that are present at a higher level (circuits, computing elements, complex spaceships, turing machines) are formed from the medium-level parts. When designing a high level life pattern, you think in terms of both low and medium level layers, switching readily between them.
But consider a different computer model: The Sims. Again, we have complex behaviors. There are medium-level behaviors (one Sim) and higher-level ones (two Sims interacting, a Sim and their house). But where is the low level? A Sim, a piece of furniture, a house, and a car are all fundamental. They are not made of separate things, and they do not all reduce to the same physics.
Now, I cannot imagine living in The Sims any more than I can imagine living as a rock, a grasshopper, or any other entity that has significantly less mental capacity than I do. I can almost, with effort, imagine living in Life, by pretending that things seem awfully similar except than physics experiments have different results, light is a bit different, etc. It's not a very high-fidelity imagining, but it does better than The Sims, where I can't even imagine where a brain can come from.
So, while I can't take the inside view of imaging living in a non-reductionist universe, I can describe one from the outside. I would also assert that claiming you can reduce down to the computer hardware running the simulation is an error. There is no conceivable in-universe experiment you can perform in the game of Life to tell you what kind of computer is simulating it. I can't quite make the same statement about The Sims, since I can't imagine living inside it to conduct the experiment, and I can't imagine an entity that could do that. I can make experiments from the outside, though, and they still won't tell me anything about the computer that underlies the simulation.
The nature of the reduction is not guaranteed to be epistemically accessible. It has tended to be so up to this point.
"The proper questions of epistemology are not actually concerned with sources at all; rather, we ask whether an assertion is true – that is to say, whether it agrees with the facts. In connection with this critical examination of the truth, all kinds of arguments may be brought to bear." - Sir Karl Popper, 'In Search of a Better World.' London: Routledge 1984.
Today's post, Excluding the Supernatural was originally published on 12 September 2008. A summary (taken from the LW wiki):
Discuss the post here (rather than in the comments to the original post).
This post is part of the Rerunning the Sequences series, where we'll be going through Eliezer Yudkowsky's old posts in order so that people who are interested can (re-)read and discuss them. The previous post was Points of Departure, and you can use the sequence_reruns tag or rss feed to follow the rest of the series.
Sequence reruns are a community-driven effort. You can participate by re-reading the sequence post, discussing it here, posting the next day's sequence reruns post, or summarizing forthcoming articles on the wiki. Go here for more details, or to have meta discussions about the Rerunning the Sequences series.