ARMS-D-Request@XX.LCS.MIT.EDU (Moderator) (09/24/86)
Arms-Discussion Digest Tuesday, September 23, 1986 9:19PM
Volume 7, Issue 18
Today's Topics:
Administrivia
SDI
----------------------------------------------------------------------
Date: Tue, 23 Sep 1986 21:01 EDT
From: Lin@xx.lcs.mit.edu
Subject: Administrivia
==>> Someone please help with the following:
Date: Tuesday, 23 September 1986 18:12-EDT
From: Mail Delivery Subsystem <MAILER-DAEMON at sri-spam.arpa>
To: < at SRI-NIC.ARPA:ARMS-D-Request@XX.LCS.MIT.EDU>
Re: Returned mail: Host unknown
----- Transcript of session follows -----
550 spam.istc.tcp... 550 Host unknown
550 <robert@SPAM.ISTC.SRI.COM>,<gd@SPAM.ISTC.SRI.COM>... Host unknown
------------------------------
Date: 23 Sep 1986 19:06-EDT
From: Dale.Amon at h.cs.cmu.edu
To: arms-d
re: SDI
[This originally appeared in SPACE. The author is not terribly
interested in flaming on this subject. He says "I made up my mind for
ethical reasons quite a few years ago, and I don't think I've heard a
new argument in at least two years. The names change and the same
flames repeat ad nauseam." Moderator]
======================================================================
Date: 15 Sep 1986 15:57-EDT
From: Dale.Amon@h.cs.cmu.edu
To: Space@Angband
Subject: Re: Phil and SDI
I realize that Phil has a personal dislike for SDI, and likewise there
are many who share his opinion.
There are also quite a large number of people who think it is the only
morally and ethically concievable way of defending ourselves. And who
likewise feel that the difficulties in SDI have nothing to do with
science per se, but with engineering.
I would suggest that a physicist or most other scientists are not
qualified to judge anything but most the basic physical laws
underpinning interception of objects.
Scientists have to understand why something works. Engineers don't
care, so long as it does. The Roman's, for example did quite well with
pure engineering. They had no science worth speaking of when they built
their aquaducts, used concrete caissons to build un-natural harbors
(such as Caesaria).
Thus a scientist will often say something can't work (like airplanes
and bumblebees) simply because they don't have the equations down yet.
Well we may not have the equations for anti-missile defense down yet,
but my engineering sense tells me it can be done, and that it is worth
doing.
I can't accept a world in which defense means threatening to immolate
millions of Russian men, women and children who have little say in the
aggressive policies of 'their' government. There has got to be a better
way to protect our right to be left alone, and it is worth trying to
make it real.
I really don't want to get into a long discussion of this again,
because I'm frankly pretty bored by the debate. I've made my decision
based on researching and studying the topic over several years, and it
is unlikely anyone here will come up with something I haven't heard and
discounted three years ago.
I would like to mention several hopeful signs towards feasibility of
SDI.
1) The phase conjugate laser techniques appear to have the
potential to track a target, correct for beam diffraction
and to focus the full energy of the main laser on the target
without requiring computing. The weapon need not even be
physically pointed since the phase conjugate effect can
generate an off-axis beam.
2) Similarly, the phase conjugate effect allows the
construction of a 'phaser'. This allows any number of
individual high power lasers to combine their beams IN
PHASE, such that all of the beams are in precise focus ON
THE TARGET.
3) The computing problem has been discussed by the Eastport
panel and found to be tractable if modern techniques of
modularity are applied rather than the monolithic and
certain to fail approach castigated by Parnas. The Eastport
panel had doubts that the DOD and it's contractors could
adjust to these techniques, but I would say that a great
deal of the Software Engineering ideas that have gone into
ADA are moves in the correct direction.
I feel that I can speak as an expert in the software engineering area
since I applied many if not most of the techniques being discussed in
developing a commercial turnkey automation system that I was lead
designer and manager of in the seventies. The system went from QA to
field installation with no fail/hard faults. And by field I'm talking
about systems of several hundred hierachically interconnected
microprocessors with thousands of sensors and control points, some of
which were critical life safety functions, spread over large buildings
and complexes of buildings. If I could turn out a near fault free
system of many megabytes (the final version control freeze tape
contained about 100MB including regression test files) of source of
operating systems, loaders, link handlers, compilers, interpreters,
regression and verification programs in 3 years using
wet-behind-the-ears college grads for my staff on a less than $2M total
development budget, then I have little doubt that it can be done. My
experience was that the techniques bring the complexity explosion under
control to the point that scaleup problems do not occur. It also leads to
reliable software with fail/soft characteristics.
The thing which these people need to learn though, is that none of it
works unless there is ONE SINGLE chief architect with complete design
dictatorship, even to the point of forcing changes in initial
specifications.
Blind adherence to initial specs is often the cause of unreliable and
off schedule software; a single line item can be more trouble then the
rest of the spec put together.
So my suggestion is that if we seriously want a working SDI, then we
should have a skunk works responsible for the whole thing. Get the best
hacker/manager in the country, give him or her any equipment or
resources they require and the right to handpick their staff, then
leave them alone for five years. No visiting generals, no congressional
junkets, no congressional testimony, no upper management meetings. If
you can't trust them to do the job as best they can, then you should
have picked someone who you think could do the job.
Unfortuneatly, most such people are mavericks that the military will
feel very uncomfortable and distrustful about, so they probably will
spend lots of money to get far less system. As the Westport committee
pointed out, it is feasible to build SDI, but only if there are drastic
changes in the way the DOD does things.
REFERENCES:
1,2: Applications of Optical Phase Conjugation, David M Pepper
SciAm 1/86 p74-83
Optical Phase Conjugation, Vladimir V Shkunov and
Boris Ya. Zel'dovich, SciAm 12/85 p54-59
3: Resolving the Star Wars Software Dilemma, Arthur L Robinson
Science 5/9/86 p710-713
------------------------------
Date: 1986 September 23 07:31:01 PST (=GMT-8hr)
From: Robert Elton Maas <REM%IMSSS@SU-AI.ARPA>
To: arms-d@xx.lcs.mit.edu
re: SDI
DA> There are also quite a large number of people who think it is the only
DA> morally and ethically concievable way of defending ourselves. And who
DA> likewise feel that the difficulties in SDI have nothing to do with
DA> science per se, but with engineering.
SDI can defend against only a fraction of a small attack, virtually
not at all against a large attack unless we make an incredibly large
SDI system. During the transition phase when we have only a small
portion deployed my claim would be true in any case. Thus SDI can
defend against an accidental launch, or against retaliation against a
first strike, but not against a first strike itself. Thus barring
accidental launch, SDI deployed by our side enables our first strike
to be effective (first strike knocks ot 90% of their ICBMs and SDI
knocks out most of the rest) but doesn't prevent first-strike by their
side. We therefore will go ahead and do a first strike as soon as such
SDI is deployed, but they will do a first strike earlier to prevent
our first strike. This creates an untenable unstable situation. Is
this moral? I say it is much more immoral than MAD which however
brutal actually reduces chance of thermonuclear war and stabilizes
things.
DA> I can't accept a world in which defense means threatening to immolate
DA> millions of Russian men, women and children who have little say in the
DA> aggressive policies of 'their' government. There has got to be a better
DA> way to protect our right to be left alone, and it is worth trying to
DA> make it real.
Unless you have something better, I think you'd better accept it. Have
you ever heard the phrase "from the frying pan into the fire"? Are you
aware that when a fire strikes a high-rise building people often die
unnecessarily by jumping out of windows, when they should have just
waited for rescue squad to reach them? I can't accept first-strike
capability covered by SDI protection against retaliation even for a
moment. I can't accep MAD in the long term but can accept it until we
find something better. SDI isn't better, it's worse. Arms reduction by
two orders of magnitude, followed by SDI, would be better, but that's
not the course of SDI under the Reagan/Graham administration.
DA> 1) The phase conjugate laser techniques appear to have the
DA> potential to track a target, correct for beam diffraction
DA> and to focus the full energy of the main laser on the target
DA> without requiring computing. The weapon need not even be
DA> physically pointed since the phase conjugate effect can
DA> generate an off-axis beam.
If you don't aim your beam at a particular target, then your automatic
phase-conjucation response is distributed among all possible
(reflective) targets in the wide field of view of your apparatus. In a
massive attack, you distribute your limited energy among many targets,
so instead of destroying one you simply warm many, to no useful
effect. Even a single warhead in your vicinity can evade destruction
by deploying a large surface area of reflective debris. Your
phase-conjugation system warms a lot of debris and the warhead, again
to no useful effect, the debris isn't even heated enough to make it
unreflective. -- You really have to select a target and then
concentrate on it, which means tracking it and aiming at the position
you've tracked it at. Phase conjugation, being an analog algorithm,
can't select a particular target when many are in field of view
equally.
DA> So my suggestion is that if we seriously want a working SDI, then we
DA> should have a skunk works responsible for the whole thing. Get the best
DA> hacker/manager in the country, give him or her any equipment or
DA> resources they require and the right to handpick their staff, then
DA> leave them alone for five years. No visiting generals, no congressional
DA> junkets, no congressional testimony, no upper management meetings.
With no cross-checking, it's too easy for the KGB to buy this person
off or otherwise convert this person to their side, and then design a
system with a very big flaw that only the KGB and this one manager
knows. That's a sure way to have a system nearly everybody in the USA
thinks will work but which the USSR really knows won't. I strongly
vote against that approach. -- How about this: we have one manager who
can make all the decisions, but a team of perhaps 20 others who are
required to keep up to date on the design and have the duty of telling
the manager and each-other if they find any flaw, and if the manager
doesn't make a timely fix they have the obligation of going public.
This will allow the centralized design you want, while also forcing
readable documentation and protecting against easy KGB control.
------------------------------
End of Arms-Discussion Digest
*****************************