Relay-Version: version B 2.10 5/3/83; site utzoo.UUCP Posting-Version: version B 2.10.1 6/24/83; site ubc-vision.CDN Path: utzoo!utcsri!ubc-vision!havens From: havens@ubc-vision.CDN (Bill Havens) Newsgroups: can.general,can.ai Subject: Star Wars Message-ID: <897@ubc-vision.CDN> Date: Mon, 1-Apr-85 14:41:43 EST Article-I.D.: ubc-visi.897 Posted: Mon Apr 1 14:41:43 1985 Date-Received: Mon, 1-Apr-85 15:41:48 EST Organization: UBC Computational Vision Lab, Vancouver, B.C., Canada Lines: 86 We as Computer Scientists must accept a new responsibility of our technology. We are apparently at a critical decision point. Having created weapons that can literally destroy life on the planet, we are now considering entrusting the use of these terrible weapons to computers and computer software. To the lay public (and unfortunately to our leaders as well) the allure of computers makes it seem like an appropriate technology to apply to our strategic defense against a very real Russian threat. Reagan has painted a picture of a "defensive umbrella" that would forever shield us from Russian hegemony. He has even offered to share the technology with the "enemy". The dream is wonderful but the reality is quite the opposite. Unfortunately, the "arms race" greatly increases the likelihood of nuclear annihilation. Every weapon improvement which makes our forces more accurate, less visible, more numerous, and now more automatic increases the vulnerability of our enemy. Their forces become less effective unless they increase their number, power, accuracy, etc. More ominously, when faced with inherently overwhelming "first strike" weapons (such as the MX-missile, the Pershing missile, and now laser battle stations), their only effective strategic response is "launch on warning" (sometimes referred to as "use 'em or lose 'em"). The Russians have been put in exactly this situation by America's superior technology. And they have stated that, given the 5 to 6-minute flight time between West Germany and Moscow, they must adopt this strategy. How does this argument relate to Computer Science and, in particular, current events here in peaceful Canada? We are the EXPERTS that our government is asking (or will soon ask) to develop automatic systems to protect us from incoming Russian missiles. The decision times are too short for our leaders to make strategic decisions. Laser battle stations will have at MOST 60-seconds of "boost phase" in which to shoot down the Russian missiles. Who will make these decisions? Suppose our leaders are indisposed (or can't be found as happened in the US recently)? Even if available and awake, what kind of decision about the fate of the Earth can anyone make in 60-seconds? The US Congress asked the same questions in hearings and were told by the "Strategic Defense Initiative" representatives from the Pentagon that maybe the President won't be in the decision loop. In other words, the decisions about our planet will be made by automatic systems, that is, by complex AI programs communicating over a vast satellite computer network. The implicit assumption in this technology is that computer systems can be devised to be perfectly safe, absolutely reliable, and employ correct, completely debugged algorithms. Alan Borning has recently written an article called "Computer Reliability and Nuclear War" in which he dispels this myth. No computer system ever will have these properties. At best we can expect machinery and algorithms which are constructed to exacting standards and tested extensively. Unfortunately, automatic strategic defense can never be tested under real conditions (except once!) and subtle bugs will always remain. As Borning points out, it is exactly those subtle unexpected situations when software fails most spectacularly. Yet the Pentagon wants to use our technology for exactly these complex, confusing, split-second battlefield decisions. The idea is insane! Every system will eventually fail. The domestic Nuclear Energy industry has relied on ever increasing safety precautions, redundant systems, and reliability testing to develop a "safe" energy source. But the Three Mile Island (TMI) reactor was within 30-minutes of a real meltdown. All the automatic systems had failed or been disabled. The technology was out of control and no HUMAN understood what was going on. Indeed, President Carter visited the reactor at this time unaware of the real threat of the accident. In short, technology is only "safe" when you are willing to accept the consequences of infrequent but inevitable failures. If TMI had "gone critical", a large part of Pennsylvania would have been made uninhabitable for years and thousands of people would have been killed or contracted cancer. BUT life on the Earth would have continued. We would have survived (and perhaps have been made wiser). This is not the case with "Star Wars". An eventual catastrophic failure will mean a holocaust that no one can even imagine. To relinquish our destiny to our own imperfect technology is not a sane decision. Either our leaders are afflicted by this insanity or they are unaware of the real dangers involved. They do not realize that we as a nation are much less safe with this technology than without it. I choose to believe the latter and have faith that we can use our expertise and reputations to modify our national direction. But common sense will not necessarily prevail. We must make a public stand now if it is to have any real effect. I urge you to voice your concern loudly to colleagues, your MP, anyone in the Press who will listen, and to sign the declaration circulated by Ray Reiter over this network. Bill Havens... havens@ubc.CSNET ..!ubc-vision!havens