First of all I'd like to point out the fact that Drexler's brand
of molecular manufacturing is not an absolute prerequisite for a
transhumanist future, though it would certainly come more than
handy. (Expect me to dance through streets, while drinking
expensive liquour, throwing (for no particular reason, allow to
me to introduce the charming neologism "bazooka barfing",
relatively recently invented on the infamous pigdog list)
confetti and hollering "Vivat Drexler!" at the top of my voice
the day a rigorous study will show that nanoagent
autoreplication by means of mechanosynthesis (aka machine-phase
chemistry) is indeed feasible <a long, hard knock on wood>). The
only really big minus to this is apparent impossibility of
gradual, nondiscontinuous uploading, or nanoresurrection
(verbatim re-incarnation) of corpsicles, nee cryosuspended
patients. Personally, I think believing into nanoresurrection in
the flesh to be quite silly, since both overestimating
operational constraints drextech is subject to, and furthermore
assuming an profoundly anthropocentric view of the future. The
desuspended ones might be in for quite a shock, far beyond of
what Alcor's mildly futuristic O'Neill space colonies featuring
brochure has to offer.
The discussions on this and several other lists had led to the
following alternative future scenarios so far (please point out
these I failed to mention):
1) The >Web, a near-future noosphere matrix which leads
to a technological Singularity, or even Transcension
(surely, these two have earned their CAPS initials)
by technologies unimaginable to us-current. This is
the most conservative route, since involving merely
enhanced cooperation of human and nonhuman agents
and intelligence augmentation by means of delegation
of duties (bureaucracy++) to extracorporal gadgets and
agentry, operated by radical GUIs (HMDs, speech and
optical input, Brain-Actuated Control (BAC) and trivial
neuro emplants for virtual motorics).
2) Creation and inexpensive duplication of infoprocessing
systems of human equivalent or better, resulting in a
progressive enhanced design runaway. Brittle, inference
engine - typed AI has shown to be useful enough to be
commercially feasible, yet is falling remarkably far of
these goals. GA-driven connectionist paradigm, and a
dedicated nonalgorithmic hardware (EHW) seem to make
up a much more explosive brew. The >H ALife Golem rears
its multidimensional, fractally rendered posthumanist head.
3) Space-located (macroscopic) autoreplicators (initially
designed and deployed for extraterrestrial industry
purposes) subjected to darwinian fitness pressure will
evolve into artificial life forms, deep space and high
luminance/hard vacuum being their natural habitat, letting
their meek engineered origins far behind (even forgetting us,
though future machines should be able of being sentimental).
The points 2) and 3) are best termed the "Mind Children"
scenario.
The turmoils accompanying above possible phase transitions might
be considerable, involving even destruction of all life on earth
(uneasily, one ponders diverse Great Filters, GRBs, and
whatnots), whether as the result of a military conflict, or the
equivalent of industrial or research accidents. (You see,
obviously my European origins demand their toll, requiring at
least a single dark apocalyptic stone in the otherwise uniformly
gaudy whirl of the transhumanist mosaic).
Probabilistically, the participants of this list are unlikely to
effect measurable impacts upon these future developments. Even
worse, I still think careful mentation, idea gestation and
long-winded, good-humoured mutual bickering on diverse mailing
lists and (goodheavensforbid) usenet newsgroups are essentially
incompatible with accomplishing any of these things we so like
to talk about. (We the cheerleaders, not the sweating crew out
in the field. I'd be delighted to be proven wrong, though.
Anyway, even spectator status has got its intrinsic benefits).
Apart from having a good time, and a faint chance of actually
contributing to the global transhumanist memefection conspiracy
(tee-hee), my strictly personal life heuristic is thus to
maneuvre oneself into a position of surviving the transition
imminent (it'd better), possibly even profiting by it. Somehow I
surmise am not alone in this ;)
But back to nano (though this is by now sure facultative, being
a rant (as if nobody already noticed)). Prior to Drexler, I
guess even before R. Feynman's famed table talk "There is plenty
of room at the bottom" (funny, remember SciAm's faux pas of
using "Cargo Cult Science", Feynman's term for pseudoscientfic
gobbledygook in their more than slightly dimwit Drexler
critique, a course they by now have retracted, due to one single
NASA Ames publication) there was already much mentation about
methods for building artificial molecular-scale objects. I seem
to recall reading several papers from start to mid-80's during
my high school time. I also remember leafing through an
assistant professor's (who has spent some of his time in St.
Anford) "Nanosystems" preprint, which we at that time both found
quite ridiculous. After having bought "Nanosystems" and reread
it severeal times I now tend it to be much less laughable (this
probably being attributed to me being out of the academy for
sufficient time to lose all remaining scientific perspective
traces).
Drexler's "Nanosystems" is a pretty unpalatable piece of
literature even for an interdisciplinary reader. Constructive
engineering approach applied to molecular-scaled machines of
complexity far transcending anything the art of molecular
synthesis may ever accomplish and what computational chemists
may ever hope to model is a hard thing to swallow.
We are in a curious position today: the inexpensive computing
power currently available to us has far outstripped our
capability to do useful things with them. Quite obviously we
haven't been able to keep up with the way our tools have grown.
Our tools are not sufficiently autonomous and smart yet to aid
us in their use. The year this changes will make us stumble a
yet another major step up to the Singularity.
The average chemist uses his computer to write papers, do his
mail, and to process spectra. Period. The awareness for the
gargantuan potential slumbering in computational chemistry seems
simply to be nonexistant in most faculties. I remeber once
shaking my head incredulously at the reactions of the
(grey-haired) academic staff at my old alma mater, when a vacant
professor slot had to be filled, several years ago. A guy with a
very good reputation (at least I have heard of him) came up,
made a firework display of sophisticated concepts (GA synthetic
route planning, combinatorial chemistry, modeling, and whatnot)
and even made a whopping hands-on demo on an SGI workstation
hooked to a beamer (he brought the machine, the beamer was
hitherto regarded as just a piece of expensive furniture). I
mean this guy was really hot, he was actually doing things I
merely used to dream about at the time.
What happened? The profs stood up to fire their verbal torpedoes
at him (to sink him, they thought). Instead, they demonstrated
their very profound ignorance. They had only understood about
10% of his lecture, and were even clueless about the future
potential of those 10%. This was conservativism running rampant,
an aged alphas' throng joined last stand. I still get upset when
I think of that day.
Outside the academy, the demand for computational chemistry
comes from basically two fields: material science, and pharma.
As antisense and receptor docking medicaments have arrived in
the arsenal of modern iatro, and pharma is a big growt business,
since the age histogram of industrialized countries is getting
progressively skewed towards higher age.
There are several companies writing computer chemistry codes,
both quantum and classical level. Since handling this software
is not trivial, and assumes quite a background in classical
chemistry, I'd think consulting in such regions could bring in
quite a wad of cash. As now supercomputer-class machines made
from off-shelf consumer components (Beowulf and SHARC-based
desktop supercomputers {{I have found out there are several of
them, but the companies making them apparently expect customers
to pay a lot for added value -- btw newest SHARC with 1 MBit
on-die core, six links 120 MFlops peak, about 80 MFlops
sustainable sells for $55 in OEM quantities, and
indepedant benchmarks of TI's TMX320C6201 seem to indicate it
really to be almost one order of magnitude better than
SHARCs, though I still find its lack of links quite appaling}})
are available, a nano startup could use proprietary code,
running on in-house brewed hardware, whether Beowulf- or DSP
box flavoured, and thus skim off a larger percentile of the
cream.
Any startup setting out to build Drexler's assemblers as their
primarily, and sole goal is going to be broke surprisingly soon.
But if this company starts doing business in normal modeling
consulting, slowly building up competence and resources to
embark on bootstrap projects towards molecular manufacturing,
this company might be a quite good investment. (The best
investment would be obviously to start up just such a company).
ciao,
'gene