[ExI] MBrain heat rejection

[spike]

>... On Behalf Of Rafal Smigrodzki
Subject: [ExI] MBrain heat rejection

On Tue, Sep 11, 2012 at 1:11 PM, spike <spike66 at att.net> wrote:
>
>>... Ja, but portions of the inner layers of an MBrain get warmer than 
> that.  I have tried to model this a few different ways, and I end up 
> with a surprising result: an MBrain cannot collect a large portion of 
> the energy from a star.  Otherwise the inner nodes cannot reject 
> sufficient heat to stay in the temperature range in which electronic 
> devices we know today would work long term.

>...### What if the swarm of thinkplates allowed elliptical orbits - in this
way a single plate would spend the perihelion part of its duty cycle in
direct exposure to sunlight, supporting high-speed computations, and then as
it recedes towards aphelion its functions would switch to cooling in the
shade of other thinkplates?
Coordination of orbits of trillions of such objects might be non-trivial but
it should definitely be possible to partially equalize temperatures between
the plates. If you include precession in the model, you could have many
interleaved streams of plates, rapidly passing each other and shortening the
duty cycle. You could have complex patterns of gravitational fly-by
maneuvers between streams introducing yet another way to mix things up
without large loss of angular momentum. And if you add to it using solar
photon pressure to further manipulate orbits, the possibilities would be
endless.  Rafal
_______________________________________________



Good thinking Rafal.  The MBrain model I have been thinking about for the
last several years is does something vaguely like this, but with some
important differences.  I don't have nodes crossing each other's paths, for
in my thinking we need to keep the relative velocities of any nodes anywhere
near each other to a minimum.  So here is my notion, as best I can explain
it without pictures.  I do have pictures as well: I have a set of powerpoint
slides I drew for an engineering presentation last fall.

Step 1: imagine a thinkplate (love that term, thanks, I will occasionally
use it interchangeably with the term nodes, if you are agreeable) about the
size of a DVD, mass of about a gram, small microprocessor, three LCD regions
near the periphery of the plate which can be either reflective or
absorptive.  Those are turned on or off to control the attitude with respect
to the sun.  Bluetooth transmitter and receiver.  So far so good?  That is a
rough description of one thinkplate.

Step 2: imagine away the earth and put in that 1 AU orbit a thinkplate, and
make that orbit round (imagine away the eccentricity for now).  Replicate
the plate a trillion times or have it copy itself through 36 doublings, so
you have a trillion identical plates, and space them evenly about a meter
apart, so with a trillion nodes at 1 AU, the string wraps all the way
around.

Step 3.  Leaving that trillion nodes in place, imagine a second trillion
nodes in a very slightly different orbit, one in which the orbit plane is
tilted through about 6 pico-radians, and has a perihelion 1 AU plus one
meter, and aphelion 1 AU minus one meter.

Note, I am not crazy or stoned, this is a real proposal, now follow me, for
it gets better from here.

If you have that 6 picoradian tilt, then a quarter of a year into the orbit,
the tilted-orbit node is about a meter above the orbit plane of the original
trillion nodes.  Three months later it is back in the plane and a meter
outboard, three months after that, about a meter below the plane and 1 AU,
and a quarter of a year after that, back in the plane and a meter inboard.

Step 4.  Once you realize what I am talking about here, you see that I have
suggested a configuration in which every node has nodes forward and aft in
the orbit, and it has another node which also stays about 1 meter away, but
is in a direction perpendicular to the line formed by the two nodes fore and
aft.

Step 5.  Add another trillion nodes with the plane tilted a negative 6
pico-radians, so now every node in the original trillion has four neighbors
which always stay about 1 meter away.

Step 6.  Add another trillion nodes in an orbit tilted about 12 pico-radians
with an orbit 1 AU plus or minus about two meters.  Now every node of the
trillion in the original round-orbit has exactly 8 neighbors, if a neighbor
is defined as any node which stays within about two meters .

Step 7.  Repeat with more rings of a trillion nodes until every node in the
original ring has about 7000 neighbors, a neighbor defined as any node which
stays within Bluetooth range, about 12 meters.  There is a reason why I
chose about 7000 neighbors: it is thought that a typical human brain cell's
dendrites branch out and eventually connect to somewhere between 4000 and
10,000 other cells, with most researchers agreeing that about 7000 is a
reasonable average estimate.

There we have it.  In a seven step thought experiment, we have an orbiting
torus of about 500 trillion thinkplates, each of which has kind of a
theoretical ability to simulate a neuron and a branching dendrite.  So will
this whole thing ever simulate human thought?  Can't say, but I can say for
sure that the asteroid belt in its present configuration will not sim human
thought, not now, not ever.  A half a quadrillion MBrain nodes perhaps could
simulate thought by some mysterious means, but a non-MBrain will not sim
human thought.  

All this goes to why I favor cryonics:  If we get frozen, we don't know what
will eventually happen.  But if we don't get frozen, we do know exactly what
will happen.

spike