Re: DARPA, at least, has a clue (maybe, sometimes)
- From: Robert Myers <rbmyersusa@xxxxxxxxx>
- Date: Fri, 31 Jul 2009 16:22:22 -0700 (PDT)
On Jul 31, 5:23 pm, eug...@xxxxxxxxxxxx (Eugene Miya) wrote:
In article <h46hjs$cu...@xxxxxxxxxxxxxxxxxxxxxxxxx>, <n...@xxxxxxxxx> wrote:
In article <h45r5f$7b...@xxxxxxxxxxxxxxxxx>,
Paul Wallich <p...@xxxxxxxxx> wrote:
Eugene Miya wrote:
Gordon Moore believes tools are like the street light driving some
scientific areas. The alternative is to grope in the dark. The above
really isn't architecture as economics. But that's not what DARPA is about.
If the solution isn't under the streetlight (i.e. in the tractably
computable areas) then groping in the dark may be more effective. Even
if you fail to get results, you will fail to get results much more
cheaply and without expending huge amounts of human capital...
Another approach is to think where you are going in advance, and
use a compass. Such systematic forethought is rare in IT, but it
does happen.
A compass (magnetism) is an good important ordering tool.
Some mathematics (and simulation) are lucky to have order as a property.
The problem comes with combinatorics.
The problem with dark unstreet lit space is it's big. Real big.
Other fields don't have the same kind of order (e.g., complex variables).
Presumably you are referring to analysis of functions of a complex
variable. I don't think I've ever encountered a more orderly field of
study. You can start with Cauchy, Weierstrass, or Riemann. The fact
that so many conclusions about integers fall out of studying
meromorphic functions is a bit of a surprise, but it's precisely that
sort of "surprise" that makes me suspicious of your analogies.
Time to cite Eugene Wigner:
http://www.dartmouth.edu/~matc/MathDrama/reading/Wigner.html
"The Unreasonable Effectiveness of Mathematics in the Natural
Sciences"
One is tempted to write a parallel article: "The Unreasonable Failure
of Mathematics in Computer Science."
If you're doing a problem in applied mathematics and the difficulties
magically yield as you proceed, you suspect you are on the right
track. If the difficulties only multiply, you suspect that you have
either made a mistake or chosen an unproductive line of attack. Many
of the difficulties in computer architecture have yielded, sometimes
in surprising ways. In the realm of supercomputers, though, that's
because the field has simply decided not to bother with the hard
stuff. The difficulties in software seem only to multiply with the
passage of time.
When you're on the right track, things work well. When your approach
is flawed, things just become more and more difficult. Which would
more accurately describe computer systems (especially networked
computer systems) of interest to the U.S. national security
establishment?
Computers may have suffered from the fault that they simply proved to
be too flexible and too immediately useful, leaving too little time to
get the foundations in order before a leaning tower of Pisa had been
built.
You've mentioned chemisty, and it has suffered from a related
problem. Making miraculous compounds was relatively easy.
Anticipating all the side-effects, especially for human health and
welfare, was another story.
Robert.
.
- Follow-Ups:
- Re: DARPA, at least, has a clue (maybe, sometimes)
- From: Eugene Miya
- Re: DARPA, at least, has a clue (maybe, sometimes)
- Prev by Date: Re: comp.arch has made itself a sitting duck for spam
- Next by Date: Re: DARPA, at least, has a clue (maybe, sometimes)
- Previous by thread: Re: comp.arch has made itself a sitting duck for spam
- Next by thread: Re: DARPA, at least, has a clue (maybe, sometimes)
- Index(es):
Relevant Pages
|
