Re: Michael Crichton's deadly nanotech novel Prey

From: Hal Finney (hal@finney.org)
Date: Mon Nov 25 2002 - 18:30:24 MST


Chris Phoenix, cphoenix@best.com, sent me his review of Prey, and said
I could post it to the list.

Glancing over the review, it does seem to contain serious SPOILERS
so be warned. I've ordered the book from Amazon using the link I
sent yesterday, so I'm not reading any reviews yet.

Hal

S
C
R
O
L
L

D
O
W
N

Chris Phoenix's review:

Well, I've read the whole book. There are a few science bloopers, such
as a confusion between STM (scanning tunnelling microscope) and electron
microscope (p. 21 and p. 133) Also, he's not really clear on scale and
number; the "camera" medical device is one ten-billionth of an inch in
length (about the size of an atom) but also "a tenth the size of the red
blood cell" (more plausible) (p. 23). Also, he says "a typical
manufactured molecule consisted of 10^25 parts" which is about 100
grams! And he doesn't seem to know the difference between
"piezoelectric" and "photovoltaic." But these are just quibbles--except
that they pervade his reasoning and allow him to invent impossibilities.

His military devices, while "quite a bit larger" than the medical
devices (p. 138), are "small enough to pass right through a
membrane."(p. 223) And he has a character worrying about whether they
can get into synapses (p. 256). They're several orders of magnitude too
big for that. So the whole mind-control symbiosis thread doesn't work.
Anyway, that part of the plot is simply lifted from "Blood Music" (the
short story; the novel is a bit different) and not done as well.

A deeper issue is the question of evolution. He doesn't make a clear
distinction between the evolution of individual nanobots and evolution
on the level of the "swarm". He doesn't mention recombination at all,
which is important. You need to mix&match the genes from each
generation to make the next generation more successful. But the idea
that a failing company can, in a few weeks, re-engineer their assembly
process to not only mutate but recombine... the fact that a swarm must
survive in order to reproduce, but the first swarms were not
survivable... the unanswered questions about the intelligence of the
wild-type swarms, and the communication between them, and... and... Not
to mention the co-evolution of biological organisms--which simply don't
reproduce fast enough to adapt to the swarm the way he's suggesting (p.
288).

The overall question: Could this happen the way he described?
Basically, no. It takes a fair amount of work to make evolution work;
to make chemicals and intestinal bacteria survive in a desert
environment; to design a self-assembly system that can assemble whole
computers from only 27 different chemicals. By the way, one of those
chemicals must be huge; the PREDPREY code is 8-10 kilobytes. Well
within the range of bacterial DNA, and of course viruses can
self-assemble. (I find myself wondering whether he's picked some of
these numbers so that when we say the devices are too complex for their
parts, he can say, "Well, nature does it...")

I was hoping there'd be a major blooper I could point at and say, "This
can't possibly work." I haven't found one. It's more of an engineering
impossibility than a scientific impossibility. And that's not an
argument that works with fearmongers--or their audience. All he has to
do is fudge things by an order of magnitude here and there, and gee, it
looks like it works. Or something like it might work--and that's enough
to make it scary.

In the climax of the book, being hit with bacteriophage makes the
swarm-people melt. yah, right. The bacteria are not part of the swarm
mechanism, just the assemblers--a logical flaw there. And it takes time
for the phage to act. This sort of order-of-magnitude problem is
present in all his science and engineering.

Of course, the problem arose because the researchers deliberately
released a self-replicating, evolving system. This violates the
Foresight Guidelines in several ways. This might be our best approach.
"Crichton had to stretch science and engineering to the breaking
point--and even then, he had to propose a deliberate violation of the
Foresight Guidelines in order to come up with a remotely plausible
story. Crichton manufactured one incredible coincidence after another,
threw in a dose of magic evolution, and capped it off with criminal
stupidity. The reasons why this story is nothing but fiction are too
numerous to list."

The evolution really is magic. We're supposed to believe that the
things, in 100 generations, can understand human neural vision well
enough to produce images of people as a side effect? Hah. They don't
even have any selection pressure to do so. But that is a trivial
problem compared to the idea that the things can understand human neural
coding.

Crichton has based his scenes and scenarios on preexisting
fiction--pretty transparently in some cases. Blood Music is the obvious
example. The bad guys melting when hit with magic water. A nest of
replicator structures in the bowels of a cave. Objects turning to
powder when attacked by nanobots. A black cloud coming from one
person's mouth into another's: right out of _The Green Mile_ by Stephen
King. He just threw together a grab bag of scary images, with fairly
flimsy justification for why the plot twisted the way he made it twist.
Strip that away, and there's not much left.

I mentioned the neural coding thing--apparently the symbiosis between
swarm and human was set up almost immediately, because Julia changed
over a period of weeks, and the release was only six weeks ago. So this
swarm, that can't even keep together in a stiff breeze, settled in to
humans immediately--with no wrong moves, and no opportunity for
evolution once it entered. Impossible, by Crichton's own description: a
new situation takes time--and evolution!--to adjust to. Oh, that
reminds me: in the cave, he had the swarms learning to avoid thermite
bombs--on a time scale of seconds. No time for evolution there.

So let's assume that the invasion of humans is an impossibility. That
underscores the fact that the whole book is cheap fiction. What's
left? A cloud of nanobots that can survive in the wild and eat
animals. Possible in theory--in practice, way too difficult to design.

And what about raw materials? We know these things have a lot of metal,
because of how they react to magnets. Well, maybe they can get iron
from mammal blood. But what about gallium and arsenic for the gallium
arsenide (GaAs) photosensors? It's not the kind of thing you find in
the desert. Yet another impossibility.

One thing in our favor: Crichton has spent too much time hanging out
with techies. His explanations will leave anyone who's non-technical
gasping and floundering. Too many undefined concepts. He'll probably
lose most of his readers. In fact, that might be a good line of
attack. "Did you understand the explanations? No? Well, that's
because they _didn't make sense_." They do make sense individually--if
you already know the background--but they don't support the story except
in a vague, general sense. A line of "Crichton is trying to scare you,
because that's his job, but he doesn't understand the science any better
than you do" wouldn't be too far amiss. He may understand parts of it,
but he certainly made some basic mistakes--see previous message.

Is this helpful? If you want to discuss it, please phone me at
520-751-1637. There's lots more I could say, but I don't know what kind
of questions you're getting.

21 "The STM. The electron microscope." (it's really electron, not STM;
see 23.)

23 "scientists could design molecular-scale devices, but they couldn't
manufacture them."
  "Our camera is one ten-billionth of an inch in length." <> "a tenth
the size of the red blood cell"

27 color vision? from single GaAs pixel swarms?
RBC's "plumped up" when they get oxygenated

32,33: debugging cycles require simple swarm behavior (setup?)

43 MRI uses pump? "chatter" is coils. (baby gets rapidly sick, MRI is
instant cure)

57 computer chip turned into gray dust. (low energy chemistry)

128 "a typical manufactured molecule consisted of 10^25 parts." ~100
grams!

131 "At the molecular level, glass is like Swiss cheese, full of holes.
And of course it's a liquid, so atoms just pass right through it."

133 IBM logo "The entire logo was one ten-billionth of an inch long and
could only be seen through an electron microscope."

135 "molecular design was nothing if not complicated. .... An inserted
atom was subject to powerful local forces--magnetic and chemical--with
frequently undesirable results. The atom might be kicked out of its
position. it might remain, but at an awkward angle. it might even fold
the entire molecule up in knots."

136 description of assembler tech. 27 "primary molecules" made from
bacteria. (number significant? Amino acids? so we can't say "not
complex enough") These form assembler which forms camera.

137 discussion of bacteria attached to assembler

138 "You've seen the bloodstream version. This is the Pentagon version,
quite a bit larger..."

138 "Those're motors. The machines actually maneuver by climbing the
viscosity of the air. .... Micromachine level, remember?" So Crichton
knows something about scaling laws--maybe read Nanomedicine?

140 venting unfiltered air. Hm... so they don't protect themselves?
And they go in and out of the building?

142-145 We meet the swarm. It's solar-powered and can hide and
reproduce. Confusion between piezoelectric and photovoltaic.

153 comm by electrical signal--short-range only

174 "The gamma assemblers break down carbon material in silicate
layers. They actually cut at the nano level--slicing out chunks of
carbon substrate." So they go after "memory chips" in car computers,
MP3 players, MRI machines...

175 self assembling chemicals (mentioned earlier) make assemblers that
can make cameras, PV, etc etc. That's a lot for 27 chemicals to do with
no chaperoning. "David was saying that a mixture of components had been
vented into the desert, and that these components--which were designed
to self-assemble in the fabrication structure--would also self-assemble
in the outside world. Assembly could be carried out autonomously in the
desert."

176 PREDPREY is a program *included in the bots* which means it must be
*encoded in the molecules*.

179 "David", I said. "That swarm's a distributed intelligence. It's a
god-damn net. It'll learn from whatever you do. Testing _is_
teaching."

193 "bacteria... good nutrient source... log phase growth where they're
doubling every two or three minutes."

199 swarm code size "just a few lines, maybe eight, ten kilobytes, no
more."

206-207 discussion of swarm evolution and memory. New swarm should
retain memory? If not, how does it know how to behave? Also, confusion
between swarms reproducing and particles reproducing.

210 "Their strategy was shifting progressively. It was evolving as we
watched."

223 "They're nanoparticles. They're small enough to pass right through
a membrane." (plastic sandwich baggies) but they're "much bigger" than
the medical imaging ones.

225 creating images in color

251 "And one thing you could count on was that any big, hot broth of
bacteria was likely to get contaminated by a virus, and if that virus
couldn't infect the bacteria, it would mutate to a form that could."
Growing on what???

251 self-org, emergent forms, interaction--core premise.

255-6 "Nanoparticles are small enough to get places nobody's ever had to
worry about before. They can get into the synapses between neurons.
.... So maybe we're infected, Jack."

259 Colors by PV tilting... was poo-poo'd earlier, now suggested
seriously. But they're too big for diffraction. And wouldn't they have
to know where the viewer was?

288 biological worms co-evolve with the factories (too fast!)

317 oral transmission

362-3 deliberate release; more about evolution



This archive was generated by hypermail 2.1.5 : Wed Jan 15 2003 - 17:58:23 MST