Re: SETI/ASTRO: MBrains/JBrains & Globular Clusters

From: Robert J. Bradbury (
Date: Mon Jul 16 2001 - 16:51:49 MDT

[This is where I get to see if the Javien forum software is
clever enough to link a message to two nodes (Mike's & Robin's)
in the discussion list...]

Mike Lorrey wrote:

> Or old black holes. What is the Hawking estimates on the lifetime of
> black holes in the range of .1-1.0 jupiters?

I'm pretty sure any black holes that size evaporate over times
much greater than the current age of the universe.

The note:
seems to suggest that the evaporation time proportional
to the cube of the mass. (I'm surpsised that Google
of "Black Hole evaporation calculator" didn't turn up
a java or javascript application for this -- what do
you know, something the net doesn't know yet...)

> Well, you are assuming that they use silicon/electron based computation.

Michael Franks does, but I generally don't. If you can keep them really
cold, you would use Lihharev's RSFQ logic elements that I believe are
made out of silicon. For MBrains my calculations are based on Eric's
rod-logic computers which are built primarily out of carbon. If you
used Eric & Ralph's helical logic, you probably use sapphire so that's
aluminium & oxygen. Since you use up all the material in a solar
system, you probably have computers built out of "almost" everything

> If instead they use photon/quantum well based computation, they could
> construct on any more populous substrate, from lithium and carbon, etc.

Li is a fairly rare element (I assume you are talking about things
like lithium niobate crystals which are used for frequency doubling).
It isn't clear that you need that. If you can build nanoscale lasers
you may be able to directly get the frequencies you desire.

> I'd look for unusual concentrations of elements popular in laser
> mechanisms: fluoride, CO2, etc. which would be needed to supply
> concentred high power photons for computing nodes.

Then you probably want GaAs or InP, those are the most common
materials for solid state laser devices. But you are *never*
going to have the right ratios of any hetero-elemental material.
So you might as well dump the rest of the stuff back into your
star for recycling (if you don't feed it to a breeder reactor).

Robin Hanson wrote:

> The peak temperature prediction is pretty weak, but the no
> emission/absorption lines prediction seems stronger, though
> it will probably be quite a while before that prediction gets
> tested.

I've sent an email note into the lead author of the Nature article
as to whether existing ground based spectroscopes (say on
the Keck) could provide this at M22's distance for red/white
dwarf stars. He's on vacation for the rest of July unfortunately
so I don't expect an answer soon.

> How about a prediction about the motions of these objects?

Scot Stride from JPL pointed out to me that the unresolved
observations, interpreted as planetary mass objects, are
making the transits at extremely high speeds. I agreed with
him and suggested a couple of possible reasons:
(a) Perhaps the objects are orbiting around the outside
    of the cluster itself (maybe Spike would comment on
    whether the velocity range 0.25 M_jupiter objects orbiting
    around 10^5 - 10^6 M_sun objects is faster or slower than typical
    planetary orbital velocities).
(b) If you are orbiting through a galactic cluster you may
    want to limit the length of time you are "near" the
    stars to limit heat and/or radiation exposure.
(c) There is possibly an information transfer theoretic argument
    that says, if you are transfering densely packed information
    as internode "probes", then the communication throughput is
    not limited by the probe launch or receipt time, but instead
    by the time it takes you to reach the next launch/receipt point
    in the cluster. If that argument has some merit, then the
    objects may select orbits that have relatively high velocities.

(I'll freely admit these are weak arguments and that I need to
think about it some more.)

> Would you not expect to find them not attached to stars?

A classical Jupiter Brain, presumably has to get power
someplace, so I'd expect to find it around a star with
an array of solar collectors at its disposal. A fusion
reactor powered Matrioshka Brain doesn't have that restriction.

> How close to a star would you expect to find them?

The paper says the objects are at least several AU from the
nearest star. One could make an argument that Jupiter
Brains utilizing a solar power source could optimize
their architecture by harvesting the solar power using
collectors close to the star, but beam the energy to a remote
location (Jupiter-to-Pluto orbit say) so that it isn't
receiving a lot of IR exposure from the star and has
an unpolluted region of space in which to setup large
telescope arrays.

> If not attached, how fast would they be drifting between stars?

If the information transfer argument above doesn't hold,
then you would probably adopt a speed that minimizes
the matter & energy expense of orbital course corrections.
That means you are likely to do an N-body simulation of the
entire cluster, predict to the limits of chaos theory
what the least expensive (long term "safest") orbits are
and then put yourself into one of those. I suspect the
velocity varies significantly depending on how you got
to the cluster, how massive it is, whether you orbit
around it or through it, etc.

In short, I don't think one can make "good" predictions
other than spectral characteristics at this point.
Its likely that the variety of astroengineered supercomputers
may have characteristics as different as the variety of stars.

I'm going to resist getting dragged into the discussion between
Eugene & Daniel regarding colonization, interstellar wars, etc.
because I believe its simply rehashing old material.


This archive was generated by hypermail 2b30 : Fri Oct 12 2001 - 14:39:49 MDT