Jay,
Random kinetic energy is high entropy and hence *low* complexity. A highly ordered state like a crystal is also *low* complexity. High complexity is mid-way between maximal order and maximal entropy. The complexity of the physics in a rock is far too low to support high level consciousness.
As for the heat in a rock, it
is complex, "entropy" notwithstanding. You can compress an approximation of the heat, in that you can randomly generate an "equivalent" temperature of kinetic motions. But to generate the exact same set of kinetic motions, for each and every one of the 10^25 atoms in a pebble, would require about 10^25 * N bits of data, where N depends on your level of accuracy per particle (given a 3D velocity vector, N is probably at least 2^6 bits per particle, but potentially much, much more, plus a few bits for encoding the type of particle).
It's the same concept behind why random noise is the most uncompressible form of data possible. You can compress it, in the sense that you can make a string of similar length and "apparent" randomness (via pseudorandom number generation), but it's not the same exact string. Random strings are virtually uncompressible, period. The very uncompressibility is the measure of the data's inherent "complexity", at least in a mathematical/computational sense. That's why a rock will always beat a brain for complexity, because the brain is
ordered to some degree, i.e. less complex, more compressible.
As for not being able to arbitrarily assign a finite state automaton to a rock, you're making the same classic mistake Chalmers did in
Does A Rock Implement Every Finite-State Automaton?, by assuming that the computations performed by a rock don't qualify because A) they don't really do anything useful, and B) they couldn't do anything else, hence they're not "causing" the state transitions. The same might be true of a human mind in the absense of free will: the brain doesn't cause the events computed therein, the causation lies A) with past events, preceding the formation of the brain, and B) random input by quantum mechanics. Since most respectable neurologists discount B as a mechanism of consciousness (much like most respectable scientists discount dualism as the nature of consciousness), we're left with A. But how is that different from the latent heat in a rock?
For example, it's assumed that because we humans could do a lot of things (like play chess, or ride a bike, if given the proper input data/situation), then that's what makes our complex computations different from a rock's. But under physicalism, you can't play chess
or ride a bike at any given moment, because there's one and only one input/situation. If the situation asks you to ride a bike, then all that data in your mind that theoretically "could" play chess if needed, actually
can't, and hence it's irrelevant to the situation at hand. If viewed from the right "slice", a rock could be seen computing the moves necessary to win a particular endgame in chess. Of course, the rock
can't compute the moves necessary for any other endgame, nor can it trigger the states that might correspond to the motor neuron firings needed to ride a bike, based on kinesthetic feedback. But, that's not the point. The input is fixed, so it only has to respond to the input, and generate the corresponding output.
It's funny how many people seem to miss this obvious fact. The complex sequence of all the complex "outputs" of the neurons of a brain is
not dependent on the complex dynamics of neurons in the brain. Make a recording of the input and the output, and then play the input through a "computer" that just spits out the same complex sequence from the recording, and functionally, for that exact input, it's no different than a real person's brain (the same output was generated), and hence that recording is just as conscious.
You can't say, "Hey, that's cheating, if we had given the brain a different input, then it would have generated a different output!". Nope, that argument is completely
empty! Why? Because under physicalism, in the original situation (the one that was recorded), the input that was recorded was the
only input that could possibly have happened (barring coherence of MWI splits, but saying consciousness relies on MWI coherence, while fascinating, is not compatible with deterministic computation, and hence software is disqualified again!), and the output that was generated was the
only output that could possibly have happened, and hence, that combination of input and output
functionally defines the conscious states that were had during the interval of time in question.
Reproducing both input and output, under functionally-defined physicalism, can only mean that the same conscious states
must be present when the recording is played back. Arguing anything else means that functionalism is
false. Of course, I don't necessarily equate functionalism with physicalism 100%, so if one is willing to part with functionalism (as I am), then there's room to shoot down my argument and still hold physicalism to be true. Adhere to functionalism (this is aimed more at Don than you, Marc), and you have to accept this argument.
Hence, under functionalism, a complex bit string
is consciousness. But complex bit strings are everywhere in nature, even in
rocks. To argue otherwise is to argue that there
is something special, over and above software/computation, that makes human consciousness more than just a rock.
Personally, I prefer the beauty of the MWI coherence possibility, and it leaves open the possibility of software using quantum events as random inputs to be conscious, by creating the same sort of coherence. Once functionalism is dropped (since admitting ultra-panpsychism is sillier than Cartesian dualism), there's still a whole slew of ideas under physicalism to be pursued, so the physicalists shouldn't be dismayed.
Actually, this isn't a surprising turn of events. Considering panpsychism, Cartesian dualism, etc., it makes one wonder whether consciousness is one of those things that, by it's very nature, is inherently ineffible. Approach the problem from any paradigm (dualism, functionalism, representationalism, etc.), and push the consequences of that paradigm to their limit, and you end up with a contradiction and/or an absurdity (an absurdity being to intuition what a contradiction is to logic, I suppose). A reductio ad absurdum can be made from any direction, so that in the end, we end up right back where we started: this is my opinion, that is yours, we disagree, and each can prove the other's idea is absurd.
It reminds me of the question of whether the software on a computer could ever determine the true nature of the computer it's running on. How could a piece of software determine whether it's running on a Pentium or an Analytical Engine (Babbage)? Two software programs running on the same computer might argue with each other, one claiming the world in which they live is a Pentium, the other arguing that it's an Analytical Engine. Which is right?
Sometimes, that's how these debates feel to me.