Re: cognitive != observable?

> If the machine answers "Yes", then we can examine a recording of its
> runtime execution and trace the complete chain of events that led it to
> say Yes. And none of those events will be special in the slightest
> regard. There is no qualia causing the machine to say Yes, merely a
> chain of unspecial mechanical events strung together.

That's true only if you think "qualia" have to be locally attached to some
particular event.  If qualia are "distributed," then what?  It only means
that you as observer aren't yet clever enough to spot how the qualia are
distributed over the trace you're looking at, but that doesn't mean
they're not there.

Also, we think of an airplane as a machine, but I defy you to get such a
trace out of an operating Boeing 747.  Traces of this sort are obtainable
and interpretable only from the simplest of machines.

John von Neumann tinkered with the idea of a self-replicating automaton,
and invented one.  One of the ideas he raises in his article on this is
that there's a sort of "complicatedness threshold."  Below the threshold,
machines are simple to describe.  At and above the threshold, the machine
itself is its own simplest description.  In other words, any attempt to
describe the machine would be more complicated than just observing the
machine itself.  Mind you, he was talking about machines, not organisms,
though not unconscious of the fact that we feel the same is true of
organisms.  There was no restriction in physical size, nor any need to
make appeal to the quantum.

If I understand you, you're saying that macroscopic finite state machines 
probably won't ever be smart.  I'm happy to agree with that, I don't think 
they'll ever be smart either.
 
> If we think that multiple such machines interacting together would make
> each other conscious, then no, because that merely resizes the problem
> upward and only makes the machine bigger.

I do agree there.  I'd put it that we've only chased the homunculus out of
the machine and into the environment, without getting at the heart of the 
matter.

> The only macroscopic systems that offer emergent effects beyond those of
> chemistry are living organisms.

I don't believe that:

Firstly, we don't know what "emergent" means.  It's one of those words,
like "complexity," which makes sense intuitively but defies attempts to
pin it down.

Secondly, when one does nail down an idea of emergence, there almost
always seem to be trivial consequences.  For instance, in vol 7 of the
journal Artificial Life, Gross and McMullin argue a pile of sand has
emergent qualities (according to Nils Baas' notion of emergence).
Simply, one grain of sand has certain features like position, but a pair 
of grains of sand has additional features like orientation relative to one 
another.  Gross&McMullin themselves argue that's a silly form of 
emergence; but, we have yet to see a criterion which reliably 
distinguishes these "intuitively uninteresting" forms of emergence from 
interesting ones.  That doesn't mean it can't be done, just that any 
argument predicated on "emergence" is on shaky ground at the moment.

> In fact, this is a good definition of life -- any macroscopic entity
> that, when aggregated, produces higher-order emergent effects

Well, you have to relativize this definition to your particular choice of
what you mean by "emergent."  Then, you think grains of sand are 
Baas-alive, because according to Baas' definition pairs of grains of sand
have emergent properties, and grains of sand are macroscopic.

> And tellingly enough, the one key distinction between life and machine
> is that living things are built with molecular-sized components.

Uh oh, you're shrinking the homunculus.  I don't think that will help.

My computer has very tiny transistors in it, but I'm pretty sure it's not
alive.  My hard drive is as small as it is thanks to quantum mechanics, 
but I'm pretty sure it's not alive either.

Anthony