Jesse Mazer wrote: > But can decoherence really "forbid" macroscopic superpositions in principle, > or only in practice? To build quantum computers, people have to figure out > clever tricks to keep fairly large systems in quantum coherence, even though > under normal circumstances decoherence would prevent this. Is there a limit > to how far we could take this? If quantum computing follows something like > Moore's law, it seems concievable that we could eventually simulate > something as complex as a cat-in-a-box, which we'd have to treat as being in > a superposition of states as long as coherence was maintained among the > elements of the computer itself.
David Deutsch took this idea farther, I think in his 1985 paper. He proposed the following experiment: 1. Create a large quantum computer 2. Create a conscious AI program 3. Run the AI on the QC 4. Let the AI make a two-way quantum observation 5. Keep the AI in a superposition of the two states a la Schrodinger's cat 6. The AI announces that it has made a definite, precise observation 7. Recombine the two states and reveal interference This shows that the conscious mind of the AI was in two states, and both were equally real. Hence one of the following must be true: A. Consciousness doesn't collapse the wave function B. AI's aren't conscious C. QM is wrong and more generally we get a choice between: A. MWI is wrong B. QM is wrong People like Penrose believe that QM is wrong, that there is a yet-unknown law called "objective reduction" which makes state function collapse an objectively real phenomenon and determines when and how it happens. But if you're not willing to go that far, it is hard to deny the force of Deutsch's thought experiment. Hal Finney