Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP Posting-Version: version B 2.10.2 9/18/84; site utcsri.UUCP Path: utzoo!utcsri!hogg From: hogg@utcsri.UUCP (John Hogg) Newsgroups: can.politics Subject: Some problems with Star Wars (LONG) Message-ID: <1186@utcsri.UUCP> Date: Fri, 14-Jun-85 15:30:49 EDT Article-I.D.: utcsri.1186 Posted: Fri Jun 14 15:30:49 1985 Date-Received: Fri, 14-Jun-85 15:43:22 EDT Reply-To: hogg@utcsri.UUCP (John Hogg) Distribution: can Organization: CSRI, University of Toronto Lines: 161 A few days ago I said that I'd post the papers submitted to the Liberal Task Force on Peace, Security and World Disarmament that were presented by members of our department. Below is the SECOND of three. (I have received permission from the author to post this; I haven't talked to the others yet.) He attacks SDI on the basis that it will *increase* the threat of nuclear destruction, not *decrease* it; due to time limitations, he barely scratches the surface, and I may bring up more points later. In the other two papers, Kelly Gotlieb shows that the economic benefit to Canada of SDI will be negative, while Ric Hehner suggests that, from a computer scientist's point of view, the system has no hope of working correctly. If you agree, I suggest that you write to the Prime Minister, with a copy to your MP. SUBMISSION TO THE LIBERAL TASK FORCE ON PEACE, SECURITY AND WORLD DISARMAMENT Andrew P. Gullen 1985 May 30 First, I would like to clarify part of our position which has been misunderstood by some proponents of the Strategic Defense Initiative. Much has been made of the predicted impossibility of things which have turned out to be possible after all. Passing over the fact that most such predictions are correct, we would like to emphasize that we are not predicting that ballistic mis- siles cannot be knocked out. Indeed, some techniques for this have been successfully demonstrated, and undoubtedly others will be. What we would claim, however, is that it is not practicable to build a system which can deal with 5000-plus missiles, and stop enough of them to be worth the financial cost and the in- creased risk of nuclear war. The worst of all possibles outcomes would be a system which is only partially effective - and this is what we predict. While exact estimates of the cost of the SDI system vary, all agree that it would be very expensive, both to build and to main- tain. While virtually any price would be worth paying for freedom from the threat of nuclear war, we will argue that not only would the system fail to provide such protection, but that it would in- crease the risk of the war it was supposed to protect us from. Complex systems, especially computer systems that must in- teract with real world, can fail in more than one way. The fami- liar way is failure to perform when action is needed - probably everyone here has had a bank machine fail to operate. But anoth- er, opposite class of failures exists : a system can take action when none is called for. Most of us are less familiar with this class of failure, but common-life examples are available. Consid- er the controversy over the safety of air bags in automobiles. The question was not whether air bags saved lives in collisions, for the evidence clearly showed that they did. Instead, the con- cern was whether the bags could be prevented from deploying dur- ing normal driving, as this would itself cause an accident. Exactly the same concerns apply to the strategic defense sys- tem, and in this case the second class of failure is the more serious of the two, for it risks precipitating a course of events which would likely otherwise not happen - an accidental nuclear war. The actions undertaken by a ballistic missile defense system would not be innocuous and purely defensive. Some proposals have involved: nuclear explosions in space over the Soviet Union, with the unfortunate side-effect of electromagnetic pulse; massive rocket launches from submarines off the Soviet coast; huge swarms of small interceptor vehicles being launched over the Soviet Un- ion; and massive microwave irradiation of large parts of the So- viet Union. All approaches, to attack the missiles in boost phase, will require offensive action in Soviet airspace, because this is where the missiles will be. Furthermore, the ballistic missile defense system would be tied into the larger American system, and its false alerts would add to the nearly one-per-day rate of false alerts that system suffers at present. Even worse, the activity and false alerts of the American system cannot be hidden from the Soviets, who must take precautionary steps of their own, thus reinforcing the Amer- ican alert. As Paul Bracken writes in "Command and Control of Nu- clear Forces", "... a threatening Soviet military action or alert can be detected almost immediately by American warning and intel- ligence systems and conveyed to force commanders. The detected action may not have clear meaning, but because of its possible consequences protective measures must be taken against it. The action-reaction process does not necessari- ly stop after two moves, however. It can proceed to many moves and can, and often does, extend from sea-based forces to air- and land-based forces because of the effect of tight coupling. In certain political and military situa- tions, this action-reaction process can be described as a cat-and-mouse game of maneuvering for geographical and tac- tical position.... The possibility exists that each side's warning and intelligence systems could interact with the other's in unusual or complicated ways that are unantici- pated, to produce a mutually reinforcing alert. Unfor- tunately, this is not a new phenomenon; it is precisely what happened in Europe in 1914. What *is* new is the tech- nology, and the speed with which it could happen. The ballistic missile defense system forces a drastic reduction in the decision times available to the two sides. All parties agree that effective ballistic missile defense rests on effective interception of missiles during boost phase, when the rocket boosters may be attacked and before the number of targets expands due to MIRVing and release of decoys. Unfortunately, boost phase lasts at most 300 seconds, and for proposed missiles such as the U.S. Midgetman will only last 40-50 seconds. We have already seen reductions in decision time: from several hours for manned bomber attack, to 30 minutes for ICBM attack, and to 5-10 minutes given forward-based weapons such as submarine-launched missiles and the Pershing. With SDI, we are looking at decision times measured in seconds. This is far too fast for consideration, too fast for consultation, and in fact too fast for human judgement. Computer decision making has been proposed as a replacement, but this is unsatisfactory, as Professor Hehner will show. The Soviets must react quickly to the American alert. If the possibility exists that the American SDI system would be able to stop a ragged, greatly weakened Soviet retaliatory attack, and if the Americans have deployed accurate, silo-killing missiles such as the Trident D-II, the Pershing II and the MX, the Soviets must take into account that any activation of the American system may presage a pre-emptive strike. They are then forced to take their forces to a high-alert, launch-on-warning status, for once an at- tack strikes, they may be powerless to retaliate. This speedup in events is intrinsic to any forward-based force, whether defensive, offensive or both. Yet more problems: while missiles are many and unpredictable, satellites are few, fragile and move in predictable orbits. They are thus easy targets for the SDI weapons, which would serve as effective and very fast anti-satellite weapons. With their sens- ing, communications and ballistic missile defense satellites under the threat of sudden destruction in a crisis, the forces of both sides would be under more pressure to take precautionary moves, even to the extent of pre-emption. The SDI system would not be the first such strategic blunder. The MIRVing of missiles in the 1970's made it possible for an enemy to knock out a number of one's own warheads with just one of theirs. In fact, with the MIRV ratios in use, an enemy can as- sign two or three warheads to each silo, raising the kill proba- bility, while reserving most of his force for ensuing blackmail (at least this is what goes on in the minds of strategic planners). Thus MIRVing has led directly to today's first-strike worries, for with rough parity in single-warhead missiles, one cannot reliably knock out one's opponent's forces and have any- thing left over. In conclusion: the SDI systems, by wrongly taking dangerous actions, by reducing decision times and by forcing certain courses of action, will increase the danger of nuclear war without sufficiently protecting us from it. -- John Hogg Computer Systems Research Institute, UofT {allegra,cornell,decvax,ihnp4,linus,utzoo}!utcsri!hogg