From: Robert J. Bradbury (bradbury@aeiveos.com)
Date: Wed Aug 21 2002 - 08:57:16 MDT
On Tue, 20 Aug 2002, Brian Atkins wrote:
> Robert, sometimes you seem unrealistically overoptimistic IMO.
In my current situation one has to be to get out of bed in the morning.
:-)
> At any rate, you aren't responding to my point. Why will nanotech not
> also go through an evolutionary period where "accidents" can happen?
> And if "accidents" do happen, what kind of magnitude do you expect might
> occur?
Sorry. I'll try to do better. I think we now know how to deal with
most "accidents" that people have discussed. Broadcast architectures
and tamperproof architectures are two strategies that come to mind.
The lack of a general distribution of full blown self-replication
technologies is another. (One could have a garage that could
manufacture a car but couldn't manufacture another garage).
One could also program in limited self-replication capabilities.
I think Rafal suggested a telomere shortening approach to trigger
the activation of genes as an example in the biological realm.
I'm sure the physics & chemistry folks could come up with a
read N times type of ROM memory for the "blueprints" of
general manufacturing systems.
The approach I'm taking -- the wet path to nanotech -- actually
creates a firewall in that one uses biotech to manufacture
nanoscale parts and then does mechanical (or self) assembly of
those parts to get an assembler. So one never actually writes
the software or develops the tools that would allow an
assembler to manufacture itself. That isn't to say that
we will not eventually get that capability -- after all
once we have the major benefits of bio-nanotech people will
have a lot of free time on their hands. But early on in
the game the combined wet-dry path seems safer that a pure
dry path approach.
That isn't to say we won't stumble across some things that are
potentially harmful. Anders cited the example of the IL-4
experiments. You could point to Maxygen's evolution of
antibiotic resistant bacteria. Also on the list are
scientist's fears that there will be further crossovers
of HIV-like viruses from primates as the trade in bushmeat
expands.
When talking magnitudes I look at 20-40 million people probably
dying of HIV over the next 1-2 decades. Thats more than 4
U.S. cities going up in smoke. I would consider it very hard
to envision any terrorist or state, with the possible exception
of Russia (or an asteroid strike), causing that magnitude of
devastation over the next decade. In the following decade
things get a bit riskier (smaller cheaper technologies in
the hands of terrorists potentially) but our defensive
capabilities may be expanded significantly as I've discussed.
The real risks of nano don't arise in my view until ~2018+,
more likely 2025+. Before then I think we are at a much
greater risk from an amoral AI breakout (as I've also
discussed). The only thing that might change my opinion
is if Zyvex demonstrates real molecular nanoassembly
sometime in this decade.
Robert
This archive was generated by hypermail 2.1.5 : Sat Nov 02 2002 - 09:16:19 MST