ARMS-D-Request@MIT-MC.ARPA (Moderator) (12/13/85)
Arms-Discussion Digest Thursday, December 12, 1985 5:18PM
Volume 5, Issue 55
Today's Topics:
still more on the launch on warning debate...
submission to arms-d
----------------------------------------------------------------------
Date: Tue, 10 Dec 85 17:02:56 EST
From: Herb Lin <LIN at MIT-MC.ARPA>@MIT-MC.ARPA
Subject: Freedom of Information?
on the existence of LOWC: that is not in dispute; rather, that is a
matter of public record -- the U.S. does possess LOW capability.
Whether or not having LOWC translates into a matter of *policy* that
will be actually executed if we get warning of attack is another
question. What you want to do is eliminate that response from the
President's menu of options entirely, since you think it is
unconstitutional due to electronic unreliability.
That's a different question from the existence of LOWC.
------------------------------
Date: Tue, 10 Dec 85 17:03:00 EST
From: Clifford Johnson <GA.CJJ at Forsythe>@MIT-MC.ARPA
Subject: Freedom of Information?
REPLY TO 12/05/85 13:18 FROM LIN@MIT-MC.ARPA: Freedom of Information?
Yep, you understand the thrust of my complaint. But I'm no so
sure about LOWC being a matter of clear public record -
Weinberger's attorneys made a point of NOT LETTING ON WHETHER
LOWC EXISTED, and as you can see from the Senate record, it's
NOT QUITE admitted. Moreover, my suit stung Zeigler (and Ikle)
into firmer and firmer assertions (to reporters) that LOWC did
not exist, although again they stopped just short of definitive
pronouncement.
(When the case first hit the wires, it was headlines in
Colorado Springs; NORAD put out a wire to sqelch it.)
However, having reviewed evidence, I agree the record clearly
indicates LOWC's existence.
Cliff
To: LIN@MIT-MC.ARPA
------------------------------
Date: Tue, 10 Dec 85 17:03:04 EST
From: Clifford Johnson <GA.CJJ at Forsythe>@MIT-MC.ARPA
Subject: Freedom of Information?
Herb:
I got mailers re U.S. & Soviet misrepresentations, and have the
following reflection on freedom of information.
Re mis- and missing information, it's not exactly easy to get an
admission as to launch on warning's existence (or otherwise), which
is a non-techspec issue, and as of this year NORAD has begun
denying FOIA (Freedom of Information Act) requests. My experience
with the Freedom of Information Act is that part one says you can
see anything, and part two says except everything. Specifically,
when I once demanded a copy of a memorandum having myself as
subject from a bank regulatory agency, I was denied this not on
one, but on two overlapping exception rules. Meanwhile, I was fed
bullshit.
This is not to say that the Soviet system isn't much much worse,
but IF a U.S. mistake could cause Armageddon, I don't a think a
claim to being better has much weight either.
Re said Feb. 1981 report, its subject is PD-59, and in particular
it addresses the concept of limited nuclear war, which everybody
agrees is impossible, but the U.S. argues that it must have
limited retaliatory options in the SIOP because the Ruskies will
think it incredible that the U.S. would retaliate massively because
of a limited Soviet attack. It's a bit ambiguous, as always, re
launch on warning, but this in conjunction with other documents
make the existence of LOWC as a SIOP option clear:
"SEN. PERCY: We have had in our lifetime experience where a
nation waged war because in their perception they were acting in
anticipation of a hostile action that was going to be waged
against them. Certainly if we had intelligence that we were going
to be attacked at Pearl Harbor, we would have done something about it
then, rather then waiting to retaliate. [Ironically, the U.S. did
have this information, but no-one paid attention due to the more
engrossing intersticial defense rivalries, as described in Ford's
"The Button".] Do we have a plan? If our intelligence, for
instance, gave us 100-percent assurance that there was an intended
strike on us unless we did something about it, what would we do?
Do we have such a plan to deal with such a hypothetical situation?
SEC. BROWN: There are options that cover that situation. (Deleted.)
SEN. PERCY: Would you say then, in your judgment (deleted). Because
I can recall when Tom Watson and I, 20 years ago, were at NORAD,
that famous, or infamous, incident occurred - when all hell broke
loose; they were absolutely convinced there were missiles coming at
us.
THE CHAIRMAN: Was it a flock of ducks?
SEN. PERCY: It was radar reflections off the moon or something.
They didn't retaliate or do anything; they straightened it out.
SEC. BROWN: That matter has been a matter of discussion before me in
other committees. I think launch under attack is something that is
important to have as an option, but the time to decide, and only the
President can decide, in terms of retaliation, in terms of launching
missiles, is so short and the possibilities of mechanical malfunction,
computer malfunction, so great that I don't think that that is a
credible sole strategy. Let me put it this way: The Soviets should
not be able to count, AND I THINK AREN'T ABLE TO COUNT, on our
not doing it, but we surely should not count on being able to.
That is uncomfortable [sic; unconstitutional?] but that is the way
it is, and I think that contributes to deterrence. Ed, you have
thought about this, I know, a lot.
SEN. MUSKIE: Let me give you the specifics of it. (Deleted.)"
To: LIN@MIT-MC.ARPA
------------------------------
Date: Tue, 10 Dec 85 17:03:10 EST
From: Herb Lin <LIN at MIT-MC.ARPA>@MIT-MC.ARPA
Subject: Freedom of Information?
.... But I'm no so
sure about LOWC being a matter of clear public record -
Weinberger's attorneys made a point of NOT LETTING ON WHETHER
LOWC EXISTED, and as you can see from the Senate record, it's
NOT QUITE admitted. Moreover, my suit stung Zeigler (and Ikle)
into firmer and firmer assertions (to reporters) that LOWC did
not exist, although again they stopped just short of definitive
pronouncement.
Who's Zeigler?
I can't believe that DoD could deny it; the way you have LOWC, it is
impossible to deny it. You defined it as "any set of procedures
whereby the retaliatory launching of nuclear missiles may occur both
in response to an electronically generated warning of attacking
missiles, and prior to the conclusively confirmed commencement of
hostilities with any state presumed responsible for said attack." It
*is* a matter of public record that in 1981 LOW was an option. The
question that remains is whether or not it still is an option.
It may or may not be. (Note Weinberger and Vessey testimony to
Congress in 1982? that "The Soviets would not be sure that we would
ride out an attack." (citation unknown, try NY Times).) However, it
is clear that by your definition, LOWC does exist, because it is "a set
of *procedures* that *may* occur" regardless of whether it is
currently a formal option.
(When the case first hit the wires, it was headlines in
Colorado Springs; NORAD put out a wire to sqelch it.)
Say some more about this.
------------------------------
Date: Tue, 10 Dec 85 17:03:17 EST
From: Herb Lin <LIN at MIT-MC.ARPA>@MIT-MC.ARPA
Subject: Freedom of Information?
.... But I'm no so
sure about LOWC being a matter of clear public record -
Weinberger's attorneys made a point of NOT LETTING ON WHETHER
LOWC EXISTED, and as you can see from the Senate record, it's
NOT QUITE admitted. Moreover, my suit stung Zeigler (and Ikle)
into firmer and firmer assertions (to reporters) that LOWC did
not exist, although again they stopped just short of definitive
pronouncement.
Who's Zeigler?
I can't believe that DoD could deny it; the way you have LOWC, it is
impossible to deny it. You defined it as "any set of procedures
whereby the retaliatory launching of nuclear missiles may occur both
in response to an electronically generated warning of attacking
missiles, and prior to the conclusively confirmed commencement of
hostilities with any state presumed responsible for said attack." It
*is* a matter of public record that in 1981 LOW was an option. The
question that remains is whether or not it still is an option.
It may or may not be. (Note Weinberger and Vessey testimony to
Congress in 1982? that "The Soviets would not be sure that we would
ride out an attack." (citation unknown, try NY Times).) However, it
is clear that by your definition, LOWC does exist, because it is "a set
of *procedures* that *may* occur" regardless of whether it is
currently a formal option.
(When the case first hit the wires, it was headlines in
Colorado Springs; NORAD put out a wire to sqelch it.)
Say some more about this.
------------------------------
Date: Tue, 10 Dec 85 17:03:25 EST
From: Herb Lin <LIN at MIT-MC.ARPA>@MIT-MC.ARPA
Subject: Freedom of Information?
"It is not U.S. policy to launch on warning" said Pentagon spokesman
Fred Hoffman. "I have no comment on whether there is such a
capability." (S.F. Examiner, Mar. 9, 1984.)
That is consistent with what I have been saying.
Re the Colorado Springs report, etc., I can send you a copy.
Pls do.
Re NORAD's squelch wire, see the (International) New York Tribune's
story Mar. 14, 1985 story.
Again, if you have it, pls send it.
Several reporters called me up and
said NORAD/DOD denied the existence of launch on warning. A main
spokesperson (Ziegler) had a disclaiming denial read out as an
official announcement (denied LOW "policy"), and persistent
reporters were told that under no circumstances whatsoever would
the U.S. launch on warning because it was much too dangerous.
Ikle confirmed the latter in unmistakable terms to a European
correspondent.
I would have to see the precise text of the reply to comment; DoD
spokesmen are very cagey, and phrase their responses very carefully.
------------------------------
Date: Tue, 10 Dec 85 17:03:31 EST
From: Clifford Johnson <GA.CJJ at Forsythe>@MIT-MC.ARPA
Subject: Where we stand on the facts
REPLY TO 12/07/85 11:23 FROM LIN@MIT-MC.ARPA: Freedom of Information?
It seems we agree that:
(1) LOW capability as defined in my complaint exists re Minutemen,
and is a matter of public record.
(2) The DOD's public denials of LOW "policy" don't mean the U.S.
would not launch on warning.
Do we agree that:
(1) As it exists now, LOWC gives rise to a peacetime-nonnegligible
and crisis-substantial probability of spurious launch.
(2) As it exists now, the DOD deems radar confirmation of satellite
warning sufficient to trigger a Presidential decision whether to
launch, and this scenario is categorized "launch-under-attack"
by the DOD.
?
Cliff
To: LIN@MIT-MC.ARPA
------------------------------
Date: Tue, 10 Dec 85 17:03:35 EST
From: Herb Lin <LIN at MIT-MC.ARPA>@MIT-MC.ARPA
Subject: Where we stand on the facts
It seems we agree that:
(1) LOW capability as defined in my complaint exists re Minutemen,
and is a matter of public record.
Yes.
(2) The DOD's public denials of LOW "policy" don't mean the U.S.
would not launch on warning.
Qualified yes. It is also not true that the existence of LOWC does
not mean that the U.S. would LOW.
Do we agree that:
(1) As it exists now, LOWC gives rise to a peacetime-nonnegligible
and crisis-substantial probability of spurious launch.
Absolutely not; I disagree fundamentally with this statement. The
trick is what you mean by non-negligible and substantial. As I read
these words, I disagree. If you would like to propose a different
definition, I would reconsider.
I hope you mean the words to mean "in an absolute sense" -- certainly
what is substantially probable to a Soviet isn't necessarily that to
an American.
(2) As it exists now, the DOD deems radar confirmation of satellite
warning sufficient to trigger a Presidential decision whether to
launch, and this scenario is categorized "launch-under-attack"
by the DOD.
Qualified yes. In fact, since BMEWS would see things rather late in
the time line, I expect that were the data showing launch from just
the IR satellites, the President would be notified. He would make the
decision either then, or wait for radar data, as he decided. He would
probably received lots of advice to wait.
My uncertainty here is what you mean by "trigger Presidential
decision". Notification isn't triggering a decision.
I still don't agree that your use of the terms launch under attack and
launch on warning are accurate; I still maintain that LUA is AFTER at
least one warhead has exploded in the US and that LOW is before an
explosion. However, both are after warning of launch. (I'm not sure
our differences are terminological, however.)
Do we have agreement or disagreement on the extent to which military
commanders make decisions to launch?
BTW, I have gone through the stuff you sent by snail mail; I will have
comments very soon.
------------------------------
Date: Tue, 10 Dec 85 17:03:41 EST
From: Clifford Johnson <GA.CJJ at Forsythe>@MIT-MC.ARPA
Subject: Where we stand on the facts
REPLY TO 12/09/85 11:53 FROM LIN@MIT-MC.ARPA: Where we stand on the facts
It seems we have roughly converged on all points except the size of
the probability of spurious launch, and what the DOD understands by
"launch-under-attack".
I tend to ignore whether the military might launch without
Presidential intercession because I deny any possibility of
verifying a warning in the response time. I don't think we quite
disagree on this in the sense that a procedure would probably be
violated if the generals went ahead sans the President, at least
without a nuclear-alert DEFCON.
As regards "launch-under-attack" meaning "launch-after-impact", it
seems clear to me that launch-under-attack, while including launch
after impact, also includes some launch on warning scenarios.
Indeed, I think the evidence of this is quite strong, as in the Feb.
1981 report and as in Bob Aldridge's paper (I have mailed).
Do you know of any simulation models of the procedures, or LOW
studies, estimating accidental launch probabilities, however crude?
Regarding the difference, to test the measure of our differences,
how do you react to the following attempt at objectivizing the
probability of accidental launch?
Let's translate the probability of spurious LOW into the
mean time to wait for a spurious launch. I'd say that if
the MTTW were less than 100 years, then the probability would be
non-negligible, and if less than twenty years, substantial (and if
less than infinity, positive).
Alternatively, how about non-negligible being defined as greater
than the probability of the Titanic sinking on its maiden voyage?
To: LIN@MIT-MC.ARPA
------------------------------
Date: Tue, 10 Dec 85 17:03:47 EST
From: Herb Lin <LIN at MIT-MC.ARPA>@MIT-MC.ARPA
Subject: Where we stand on the facts
I tend to ignore whether the military might launch without
Presidential intercession because I deny any possibility of
verifying a warning in the response time.
We may disagree on this point too. What would you regard as adequate
verification independent on response time?
I don't think we quite
disagree on this in the sense that a procedure would probably be
violated if the generals went ahead sans the President, at least
without a nuclear-alert DEFCON.
I think you don't quite understand DEFCONs. DEFCON refers to a state
of readiness, and for the strategic forces it is mostly applicable to
bomber forces and to submarines, neither of which are subject to LOW.
For ICBMs, they refer to things like putting missiles back on line,
not doing certain tests normally done at the lowest DEFCON, and the
like. Most are at the ready and capable of quick firing even at
DEFCON 5 (peacetime).
As regards "launch-under-attack" meaning "launch-after-impact", it
seems clear to me that launch-under-attack, while including launch
after impact, also includes some launch on warning scenarios.
Indeed, I think the evidence of this is quite strong, as in the Feb.
1981 report and as in Bob Aldridge's paper (I have mailed).
As I said before, I think this issue is secondary; I will agree to
disagree on it, and I don't think our discussion on this particular
point serves to advance the state of your argument.
Do you know of any simulation models of the procedures, or LOW
studies, estimating accidental launch probabilities, however crude?
I have seen one, done by someone at Michigan State University, that I
threw out because it was so silly. I think he is trying to get it
published in the Journal of Conflict Resolution, but I'm not sure.
Regarding the difference, to test the measure of our differences,
how do you react to the following attempt at objectivizing the
probability of accidental launch?
Let's translate the probability of spurious LOW into the
mean time to wait for a spurious launch.
Fine. I agree with this as a reasonable metric, with the caveat that
the spread is important too.
I'd say that if
the MTTW were less than 100 years, then the probability would be
non-negligible, and if less than twenty years, substantial (and if
less than infinity, positive).
I disagree with your numbers, because systems change on a time scale
of a decade. No way will any current system be operating 100 years
from now, so the probability of LOW *per unit time* (the crucial
parameter) changes as a function of time.
The problem with your approach (inherent in the problem, not due to
any fault of yours) is that false LOW refers to a mistake that the
President makes. While it may be possible to quantify the number of
false alerts that reach the President (zero to date -- so you have no
data on frequency, except to say that it probably doesn't happen more
than once every 25 years), no President would make the decision purely
on technical input data; he would consider the state of political
relations, the state of the world, and so on. These might make him
more or less likely to launch, AND YOU CAN'T QUANTIFY THEM.
Alternatively, how about non-negligible being defined as greater
than the probability of the Titanic sinking on its maiden voyage?
Post-hoc probabilities are fun, but always misleading. I assume this
wasn't a serious proposal.
Herb
------------------------------
Date: Tue, 10 Dec 85 17:03:52 EST
From: Clifford Johnson <GA.CJJ at Forsythe>@MIT-MC.ARPA
Subject: The nitty-gritty
REPLY TO 12/09/85 16:54 FROM LIN@MIT-MC.ARPA: Where we stand on the facts
The adequacy of verification and the size of probability of
accidental launch are one and the same thing.
You're right, I know nothing of DEFCONS, thanks for the education.
So they go from 1 to 5. At what level would missile crews insert
launch keys? Helen Caldicott told me they did this on the Nov.
1979 false alert, but I'm skeptical myself. Is DEFCON 1 the max.,
what DEFCON did Nixon set when he went on nuclear alert in the oil
crisis?
Re "launch-under-attack" ... I wish I thought the DOD did mean after
impact; better still, I wish I knew what they meant instead of
having to double-infer. I agree there's no more to discuss, only
clear DOD statements to resolve. One point re authorization and
Presidential "mistake" is that Braken and others seem to make clear
that NORAD determines the veracity of warning, and the President
is simply informed "we are under attack", taking this as a given
if it gets that far. Of course, he could say "I don't believe you"
or "wait for a bang" or "ride it out", but my understanding is he's
told the U.S. IS under attack, that's NORAD's function to
determine. I'm also afraid that in such circumstances, the military
would be in DE FACTO control of a response decision. The President
is a figurehead according to most analysts.
If there is any study that, however crude, gives any
order-of-magnitude estimate of the probability, it might be refined,
or might be usable as a best-guess in the dark. (But I expect it
was a silly study.)
I understand the "changing probability" problem, but am setting
forth the MTTW figures on the basis of the LOWC that now exists.
We do seem to be approaching the nitty-gritty of the problem. Would
you at least agree to a positive probability of accident at DEFCON
5, and a non-negligible probability in a crisis?
Much of what I say might ultimately be litigatable only on the basis
of a requirement of CERTAINTY of an attack warning. At this level,
the post-hoc argument is not invalid - even remotely improbable
accidents can and do happen. Post-hoc comparisons with such systems
as Three Mile Island are standard in the literature.
Consider two crisis scenarios.
The President judges there to be a sudden risk of a massive attack
(intended to be a surprise). To forestall this, and having
information that Washington D.C might go first, or to be sure that
launch on warning will work, he authorizes the military to launch
missiles if and as soon as there is a radar confirmation after a
satellite warning of attack (there being no tactical warning at all
the time of authorization). Is this legal? Is it provided for in
the SIOP? What is the MTTW in a crisis? I've given my subjective
guess, what's yours?
To: LIN@MIT-MC.ARPA
------------------------------
Date: Tue, 10 Dec 85 17:03:57 EST
From: Herb Lin <LIN at MIT-MC.ARPA>@MIT-MC.ARPA
Subject: The nitty-gritty
The adequacy of verification and the size of probability of
accidental launch are one and the same thing.
No, at least not by your definition of probability. You proposed and
I accepted the measure of MTTW as your probability of accidental
launch.
The adequacy of verification is fundamentally different, since it
refers to the STANDARD OF EVIDENCE, which becomes relevant BEFORE a
launch decision (bad or good) is made. The decision DEPENDS on
verification, but is NOT synonymous with it.
You're right, I know nothing of DEFCONS, thanks for the education.
So they go from 1 to 5. At what level would missile crews insert
launch keys? Helen Caldicott told me they did this on the Nov.
1979 false alert, but I'm skeptical myself. Is DEFCON 1 the max.,
what DEFCON did Nixon set when he went on nuclear alert in the oil
crisis?
DEFCON 5 is peace, DEFCON 1 is "war is imminent". Nixon went to
DEFCON 3. The US has never been at DEFCON 1 or 2.
I believe Caldicott is wrong, but I am not certain. (She is wrong
about most matters of fact.)
One point re authorization and
Presidential "mistake" is that Braken and others seem to make clear
that NORAD determines the veracity of warning, and the President
is simply informed "we are under attack", taking this as a given
if it gets that far. Of course, he could say "I don't believe you"
or "wait for a bang" or "ride it out", but my understanding is he's
told the U.S. IS under attack, that's NORAD's function to
determine. I'm also afraid that in such circumstances, the military
would be in DE FACTO control of a response decision. The President
is a figurehead according to most analysts.
I think you are probably right about this, in the sense that the Pres
must rely on NORAD. But you should know that under Kennedy and
Johnson, McNamara told them thay they should never believe a report
even of impact until he (McN) had flown over the detonation and spoken
personally with them. In this sense, Presidents *can* wait.
For Presidents who won't wait, we can't do much about them; after all,
safety procedures of any sort can be circumvented, and if the
President wants to keep his head in the sand, we can't stop him. In
this sense, you are right; the Pres is a figurehead.
If there is any study that, however crude, gives any
order-of-magnitude estimate of the probability, it might be refined,
or might be usable as a best-guess in the dark. (But I expect it
was a silly study.)
If I could help, I would. I've tossed it.
We do seem to be approaching the nitty-gritty of the problem. Would
you at least agree to a positive probability of accident at DEFCON
5, and a non-negligible probability in a crisis?
I don't know how to put *any* numbers on them, since it depends on who
the President is, who the staff are, and what the national and
international conditions are (and have been), and what US/Soviet
capabilities are, etc.
Much of what I say might ultimately be litigatable only on the basis
of a requirement of CERTAINTY of an attack warning.
What is certainty? When I asked about the adequacy of verification,
this is what I meant. What evidence would you consider adequate to
support a launch decision by the President? Additionally, whatever
your answer is, what makes you believe that that information could not
be faked.
At this level,
the post-hoc argument is not invalid - even remotely improbable
accidents can and do happen. Post-hoc comparisons with such systems
as Three Mile Island are standard in the literature.
But lots of unlikely things don't happen too. I don't understnad how
to analyze on the basis of retrospective knowledge.
The President judges there to be a sudden risk of a massive attack
(intended to be a surprise). To forestall this, and having
information that Washington D.C might go first, or to be sure that
launch on warning will work, he authorizes the military to launch
missiles if and as soon as there is a radar confirmation after a
satellite warning of attack (there being no tactical warning at all
the time of authorization).
Is this legal?
I believe so.
Is it provided for in
the SIOP?
Absolutely not. The SIOP is a menu of options, from which the
President (or whoever) can choose.
You mentioned two crisis situations? Where's the second?
------------------------------
End of Arms-Discussion Digest
*****************************