Suggestion for students
Christopher J. Phoenix
At 09:54 PM 3/24/98 -0800, @biokem...@merkurius.lu.se wrote:
>Yes, mathematics is THE science that other sciences depend upon, but you
>don't have to be a mathematician to be a successful chemist or physicist
>(you have to know some basic math, but that goes without saying). And
>the foundations of chemistry and physics really don't change over the
>five years or so that a high-school kid need to get through college and
>university...
>
>Nanotech are so many things, there is no single scientific discipline to
>encompass all the different techniques, and there is no scientist that
>knows it all! [....]
I'd have to agree with this. I know almost no math; I had serious trouble
with the frosh-level classes on integration, linear algebra, and statistics,
and I don't know anything about differential equations. Yet I'm doing
useful, fascinating nanotech research (in a couple of weeks I'll talk about
everything I can), and I don't expect to use anything more than arithmetic.
How? Because it's *lab work*. Yes, math is very useful for knowing how to
calculate whether a certain atom will be mobile at room temperature. But
there's also a place for just plain finding it out! You could say I'm only
half a scientist. Fine; a physicist who can't make an STM tip is the other
half. Put us together, and we make a very productive team. Which has more
mystique: programming a quantum mechanical electron-cloud simulator, or
writing words one-quarter micron high? Who cares? They're both pretty cool.
If you're good at math, by all means learn it. If you're good at math and
chemistry, or math and biology, or math and physics, then you can go into
theoretical nanotech. But if you aren't good at math, you can still do
amazing stuff. The science track isn't the only way to get into nanotech.
(I took a *very* narrow computer science track.)
Don't take this as encouragement to think up anything you want and blab
about it without checking whether it's physically possible. Blue-sky
"wouldn't it be great if" is *not* what I'm talking about. Learn to analyze
your ideas, even if only by rounding to the nearest power of ten. The
concept of lower and upper bounds is very important; it doesn't matter if
they're an order of magnitude apart, as long as you keep track of them. If
you're going to invent gizmos, *think* about power, cooling, communication,
scale, strength, accuracy, speed, and so on. Keep this in mind as you read
the next paragraph: these are what you should be learning to think about.
If you're going to work in a lab, learn to see what's there rather than what
you wish was there; notice the details in everything.
Develop your physical skills, intuition and creativity. Work as a hardware
tech (debug and rework) for a while if you have any skill with digital
logic, to develop a steady hand, sharp eye, and the ability to use some
modern lab instruments. When you see anything unusual, try to figure out
how it's interesting. If you made that a billion times smaller, what could
you do with it? Develop an ability to estimate. (I've done some great work
simply by adding and subtracting exponents.) Read Nanosystems quickly, and
don't worry about what you miss. Try to imagine what you could build with
tweezers and sticky rice. Pick a random number between three and five
million; go to the patent office, look up that patent, and read it; see how
much you can understand; try to picture how the device works. Subscribe to
ScienceNews; read between the lines. Read high school textbooks in
chemistry, biology, and physics; try to understand all the concepts, but
don't spend much time on the exercises. Pick an object around you.
Redesign it with material an order of magnitude stronger. Redesign it with
a computer in every cubic millimeter. Redesign it as an implant with direct
neural interface. Redesign the country's infrastructure. Learn to program
a computer. Learn what happens at every level from hitting a key to sending
data to the screen and network. (Take an embedded or device driver
programmer to lunch.) Do anything you can think of to improve your skills
at visualizing and forming conceptual frameworks. Design new classes of
materials: Foamed diamond. Reactive paper. Fractal fibers. Pick two or
three random words; invent a technology described by them. When you meet a
researcher in *any* field, learn as much as you can about their current
question. Try to come up with useful suggestions for them.
If that's too fuzzy, then I'll recommend a software engineering track (as
opposed to a math/science track). Go to work for a small company where
you'll be asked to write and debug entire programs. The debugging is very
important; that gives you practice in figuring out large systems from small
clues. This will be an absolutely crucial skill when working at the atomic
scale. The first nanotech will probably be digital in nature, for the same
reason that computers are digital: it's easier to think about, more
reliable, and more flexible than analog. If you can design, write, and
debug a packet driver for a network card, you're probably ready to interpret
a scanning probe microscope image or design simple diamondoid devices.
Yes, this is unusual advice. It's not for everyone, though it should work
well for visual thinkers and dyslexics. The point is that you don't need
any math above arithmetic (maybe, occasionally, simple algebra or trig)!
There's a place for all kinds of intelligence and skill. Personally, I'd
rather be working an a lab than an office any day; the lab is where the cool
machines are. Numbers are generic; how many people in the world can
recognize a grungy tip on an STM?
I know a man who does construction work; he can't read, but he can memorize
the blueprints for a house. That's the kind of thinking we're going to need
when it comes time to explore the design space of eutactic diamondoid! Of
course it would be better if he could read, but he doesn't really need to;
he's got the physical intuition to do amazing design work! I told him he
could do computer programming if he wanted. He was skeptical, so I
described in detail the cause of the most obscure bug I'd ever debugged. He
followed the explanation easily, and said, "Oh, it's just troubleshooting!
Nanotech isn't just about theory. It will take massive amounts of lab work
just to learn about mechanochemistry, and to develop the enabling
technologies for the first assembler. And yes, it will take lots of
software, but you can be a very good software engineer without knowing any
math at all. (Again, I'm an example of that.)
So if you don't like the idea of studying math, see whether my advice looks
fun and interesting. If neither path looks good, there's probably a third
and fourth path that can also get you into cutting-edge nanotech work.
Chris
--
Chris Phoenix cpho...@best.com
Work (Reading Research Council): 650-692-8990; Home/voicemail: 415-860-1536
Chris Worth
Chris P's two pararaphs strike me as worth adding to
the FAQ. They're true poems-in-prose. By the way, my
suggestion of learning math was for those unsure of
what to study - it may not be essential to all nanotech
areas, but it's the most useful bedrock, surely?
Ted Anderson
I also liked those two paragraphs a lot. It reminded me of something
from "The Notebooks of Lazarus Long", by Robert A. Heinlein (courtesy of
http://lemur.ecs.soton.ac.uk/Library):
A human being should be able to change a diaper, plan an
invasion, butcher a hog, conn a ship, design a building,
write a sonnet, balance accounts, build a wall, set a
bone, comfort the dying, take orders, give orders,
cooperate, act alone, solve equations, analyze a new
problem, pitch manure, program a computer, cook a tasty
meal, fight efficiently, die gallantly. Specialization
is for insects.
Ted Anderson
Christopher J. Phoenix
Great quote! Yes, it's very much the same spirit. I believe that acquiring
skills (especially physical skills) transfer ability to many other skills
and domains.
This was brought home to me today when I was assembling a piece of lab
equipment. I was trying to put a small washer and nut onto a
downward-pointing screw, that I couldn't see, using one hand, wearing thin
floppy plastic gloves. I realized that I was thinking about this problem
almost the same way as I had been thinking about STM scans of things that
might be buckytubes: there's low-bandwidth, unreliable feedback (touch
only); very limited ability to manipulate (only two fingers can fit next to
the screw); the sensor/actuator changes the picture (I kept getting the
gloves caught); and if you move wrong, you'll change the whole picture (drop
the washer). If someone practices and improves at this kind of physical
manipulation, I really think they'll be better at interpreting SPM scans!
(I got one nut on, then realized I didn't have to keep that part of it clean
and took off the gloves.)
Thanks for the library URL!