[extropy-chat] What Human Minds Will Eventually Do

[Eugen Leitl]

On Sun, Jul 02, 2006 at 01:09:17PM +0100, Russell Wallace wrote:
> On 7/2/06, Eugen Leitl <eugen at leitl.org> wrote:
>      How can you tell wanted from unwanted mutations, a priori?
> 
> If you wanted it then it's wanted, otherwise it isn't :)

You don't know this until the mutation happened. Computing
phenotype fitness from mutation is not feasible in principle,
given that you don't know what the others do. A major fraction
of the fitness is relative to others, with lots of maneuvers
in the dark. 
 
>      Do you realize that the mutation rate is a yet another fitness
>      component in the evolutionary optimization process? Static
>      critters are doomed, soon. Preemptively: I don't think Lamarck
>      can trump Darwin. At best they'd coexist.
> 
> Empirically, Lamarck has made quite nice use of Darwin over historical
> timescales - yes, memetic evolution is still differential survival of
> replicators, but the dynamics look quite different from those typical of
> biological evolution.

We're talking about measuring neck length, and producing the 
gene adaptively, while playing against Darwin.
 
>      What is your diagnostics to tell the parasites from symbionts?
> 
> I'll design diagnostics when I've systems to design them for (unless, as is
> likely, someone else has already done the job for me).

My point is that it's not feasible to diagnose parasites from 
symbionts. Especially, long-term. Look at your gut flora. 
What is safe to nuke, and what is not?
 
>      What
>      if by getting rid of the parasites you'll be outperformed by those
>      who didn't?
> 
> What if you won't?

Bzzt. Wrong answer. You have to focus on the branch with the outcome that is
bad for you. Everything else takes care of itself nicely.
 
>      My conjecture is that absolute fitness factors have a ceiling, which
>      will be reached relatively soon by a solid-state culture.
> 
> And my conjecture is that there is no absolute fitness, it remains relative to
> what everyone else is doing (as has always been the case in genetic and memetic
> evolution on Earth), so there is no ceiling to reach.

The absolute components in fitness are functionality concentration/volume
and Joules/function. They're not relative to everyone else. Once you
reach what's feasible with physics, no more advances are possible. The
relative fitness component continue to develop of course, producing
all manners of nonlinearities and oscillations.
 
>      > A gram of smarts (enough for an entire upload civilization) in a
>      hundred ton
>      > probe (you need the mass anyway for shielding and braking) is
>      negligible
>      > baggage and more than pays for itself in ability to outthink and
>      outfight a
>      > dumb probe that got there slightly before you did.
> 
>      You're describing the successor waves. The pioneers which came before
>      you
>      have moved on already.
> 
> No, I'm describing a smart probe arriving a few hours after a dumb probe. The

Where did 'hours' come from, kemo sabe? Try months. In MLYr jumps, try kiloyears.
I recommend you start forming the commitee for rogue probe recall now.
And you better infilitrate all the colonists, because they're not going
to phone home for a permit.

> dumb probe hasn't done much of anything at that stage.

Again, you're looking at the cases which are not interesting. There is
always a number of duds in a minefield. If you use your logic, a minefield
is safe to cross. Unfortunately, it takes one functional mine to prove you 
wrong. Pointing out existance of duds is worse than just wasting our time.
 
>      > Max feasible probe velocity is 0.2c, speed of Nicoll-Dyson laser
>      fire is c.
> 
>      Where did you pull that number from?
> 
> Extensive discussions on places like rec.arts.sf.science based on analysis of
> the performance characteristics of every means of transport that anyone has
> come up with.

I've seen data on microwave-irradiated carbon trus cloth sails being accelerable
at 3 g. Phased-array microwave radiators with ~lighthour to ~lightday aperture could
track this with enough luminosity (some 2 kK sail operation temperature) for
months. Don't ask me for the math, do it yourself.
 
>      The only bottleneck I see is limit to
>      rebuild rate under high interstellar hydrogen luminosity, which is
>      really
>      close to c.
> 
> What you're describing is the equivalent of using machine gun bullets as food
> while in the process of being shot with them, and hoping you can heal as fast

No, I know what I'm describing. It's not what you're describing.

> as you get shot. This strikes me as highly implausible even for full nanotech.

This does strike me as highly plausible. D. radiodurans does it for biology.
Redundant encoding, diagnostics and constant rebuild in the background plus
redudant probes results in effectively unkillable high-velocity probes.
Your worst problems are dust grains (take out one probe) and heavy ion
tracks. Notice that the microwave beam will spontaneously push a clean
tunnel through the interstellar medium clearing the path ahead of you,
and magnetic fields (perhaps frozen with a plasma shield) can deflect 
ionized particles sufficiently that it doesn't hit your payload with.
 
> Leaving that aside, what means of propulsion are you proposing to use?

The solar output, redistributed via a circumstellar cloud acting as
a phased-array radiator. 

> (Remember you also have to decelerate at the far end.)

I don't think whether the sacrificial sail approach would work. Antimatter-catalyzed
fusion drive and antimatter drive would work.
 
>      I don't see how this physics would work out. From my current data you
>      can
>      push a small probe at about 3 g for many months
> 
> How?

Gray sails, heated to white incandescense, microwave-driven.
 
>      and then it's effectively
>      out of reach.
> 
> No matter how fast the probe goes, a laser beam is faster.

Again you're assuming the probe is painted (you know where it is)
and you can illuminate it with critical luminosity, and that the
probe will make no attempts to shield. See mine field above. 

You *can't* kill all the postmen, long-term. The chain letter
will get out. Trust me. You better make your plans based
on that reality branch.
 
>      > between power blocs.
> 
>      Do you see that already the first step from earth surface to solar
>      system periphery is the first selection step?
> 
> Yes. Do you see that the only entities to have even sent unmanned probes
> anywhere near the solar system periphery are the governments of superpowers?

Do you realize that we're witnessing a breakdown in federal capabilities,
with a strong resurgence in private enterprise? Many private people could
afford a launch to LEO right now. Plasma drives are not unobtanium.
With machine-phase the payload needs to be tiny, which shrinks the rocket
size. Again, with machine-phase any person could build a rocket, making
fuel from ambient resources.
 
>      Um, there's plenty of aggression in an ecosystem. There is no earth
>      to scorch.
>      Real estate (resources) change hands with the death of the
>      individual.
> 
> You're getting confused between the ecosystems you're using for historical
> analogies and the sf scenarios we're discussing. That resources forcibly change

No, I'm not. You, however, are being condescending *and* breezy. This is not a good mix.

> hands today doesn't prove they will always do so; foxes convert rabbits into
> foxflesh, but rabbits don't have guns and bombs.

Viruses don't have bombs, yet kill people just fine. Guns and bombs don't
wipe stellar systems; nukes are largely useless in space. Stealth and backstabbing
work fine, guns and bombs, or no. Starvation works fine. Just because you're
no longer scheduled to die it doesn't mean you can't die.
 
>      I could actually run a number of simulations to prove my point, but
>      unfortunately this is about at the bottom of the priority pile for me
> 
>      right now.
> 
> Me too, but it doesn't matter: what you get out of simulations depends on the
> assumptions you put into them, and my point is that your assumptions are
> arbitrary (not to mention heavily biased by the sort of things us techies want

They're not arbitrary at all. I've found several folks who've arrived at
exactly the same conclusions in isolation. Your assumptions are not only
arbitrary, they're unrealistic. You're always looking at the possibility 
branches which are nice. Civilian planes can't be abducted, by a couple
of guys with box cutters. Civilian planes can't be flown into a building.
If they do, they don't causee buildings to collapse. Etc.

> to believe - I speak from experience, I used to envision the same scenarios you
> do until I realized they contained information only about my psychology, not
> about the actual future).

What precisely do you think is fishy with machine-phase? On the one hand,
you seem to assume it's possible. On the other hand, you have a very
selective view on the capabilities.
 
>      The destruction thing was figurative. Don't get hung up on this. The
>      point
>      is that the Amish won't build a circumstellar structure, and if they
>      tried
>      by building it they'd cease to be Amish. They'd become a Power
>      themselves.
> 
> The literal Amish won't get out there in the first place without ceasing to be

Precisely. By trying to beat the enemy, you're becoming it.

> Amish (at least not under their own power), but biological H. sapiens is no
> stranger to building big guns.

If you're trying to recall a chain letter, big guns are worse than useless.

-- 
Eugen* Leitl <a href="http://leitl.org">leitl</a> http://leitl.org
______________________________________________________________
ICBM: 48.07100, 11.36820            http://www.ativel.com
8B29F6BE: 099D 78BA 2FD3 B014 B08A  7779 75B0 2443 8B29 F6BE
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 191 bytes
Desc: Digital signature
URL: <http://lists.extropy.org/pipermail/extropy-chat/attachments/20060704/bcd887f5/attachment.bin>