Before deciding that I enjoyed software development more then engineering, I was involved in nondestructive testing (NDT). One of the major functions of NDT is to determine the adherence of the item under test to standards. Generally these standards are related to the safety of the larger system in which the the tested item is a component. My experience directly contradicts that expounded above. Anyone who orders and uses a product simply based on its "meeting" standards is being extremely foolish. For example, an NDT lab that I worked for was owned by a steel warehouse firm. The parent had a contract to supply structural steel for a major office complex. Twice the firm purchased steel only to discover that the mill certificates were faked. In the first case, this was discovered only when welders reported problems during fabrication. This kind of problem is more widespread then most people would like to admit. Consider the recent spate of reports on substandard areospace fitting being sold with false documentation. Further, I was several times asked to falsify NDT results to certify that items met standards. In one case, I failed a number of welders taking a certification examination (on the same building project noted above). The fabricator simply took the SAME weld coupons to another lab and EVERY ONE of the welders passed. In another case, at nuclear weapons plant I worked at, a team of EE's was prevented from inserting additonal circuitry into a test system that would falsify test results only when one threatened to go public with the information. While the above examples are concerned with outright fraud, many things involved in applying standards are open to interpretation. Consider the RS-232 standard. How does an inspector of power plant welds determine if an ultrasonic echo means the weld is substandard, when it is the "gray" zone? Many standard compliant items are not in compliance. The point of all this is that standards guarantee little or nothing. Questions of liability are meaningless. If the profit to be made is high enough and the risk of detection small enough, many firms will falsify certification. Worse, the falsification may be impossible to trace. A part fails and loss occurrs, but often the damage is such that no reconstruction of the exact cause can be made. Since the part was certified, the search is likely to turn elsewhere (assembly, operation, etc.). I am tired of "real engineers", who are no more exact, informed or methodicai then programmers pretending that engineering is somehow less prone to exactly the same problems in project management and control as programming. It would a trival exercise to compile a list of engineering failures, just as it would be for programming failures. The real issue is how to design and manufacture anything correctly.
From: Assoc. Press article in the May 1, 1989 Raleigh(NC) Times: The TCASSII system consists of a sophisticated transponder ... antennae, and a computer that analyzes and displays the movement of nearby planes. ... "The system has no bugs," said Don Dodgen of Honeywell. If two computers meet, he said, orders to the pilots will be reconciled automatically: if one plane is told to climb, the other will be advised to descend or to say on course. No comment can do justice to this. --henry schaffer n c state univ
(condensed from Albany NY Times Union Wed April 26, 1989, page B-17) The up-again down-again Nine Mile Point 2 nuclear power plant near Oswego was back on line Tuesday, following a weekend shutdown that "shouldn't have happened," according to a federal official. An employee accidently keyed a hand-held two-way radio near circuitry for the turbine generator monitoring system Saturday night. The transmission shut down the system, which in turn triggered an automatic shutdown of the entire facility. A section chief of the NRC region 1 office said that he has never heard of a similar accident but that most plants are sensitive and there are strict rules to prevent this. Replacement fuel costs $350K per day when the 1080 MW plant is down. The plant had been up less than a week after a shutdown caused by corrosion and loose wiring in a meter.
"Reprinted with permission from The Philadelphia Inquirer, April 24, 1989. Further reproduction of this article without the written permission of The Philadelphia Inquirer is strictly prohibited." B-2 BUILDERS: PROTOTYPE NOT NEEDED By Mark Thompson, Inquirer Washington Bureau WASHINGTON - The builders of the Pentagon's B-2 Stealth bomber are boasting that their computer-aided design for the revolutionary boomerang-shaped aircraft is so good that the $500 million plane will leap from the computer screen into the air by July without benefit of a prototype model to test the blueprints. "The first B-2 is a production aircraft," the Northrop Corp. said in its just-released annual report. "There are none of the prototypes that have been required in previous generations of aircraft." But critics warn that the Air Force decision to begin building the $68 billion fleet of 132 sinister-looking planes before flight testing has even started could prove disastrous. "I think the B-2 will crash the first time it flies," said Kosta Tsipis, director of the Program in Science and Technology for International Security at the Massachusetts Institute of Technology. "I wouldn't be a passenger aboard it for anything in the world." The lack of a prototype will make the planes' first flight "pretty exciting," agreed John Pike, associate director of the Federation of American Scientists in Washington. "I'm perfectly prepared to see the airplane fly more or less as advertised," he said. "At the same time, I'm equally prepared to see the airplane crash more or less immediately." But Capt. Jay DeFrank, an Air Force spokesman, said, "We're confident that it will make a successful first flight." The plane's two seats will be occupied by pilots from Northrop and the Air Force for the inaugural flight, which may occur secretly, he said. The top-secret B-2, successor to the troubled B-1B, has been designed to fly into the Soviet Union undetected by radar. The not-ready-to-fly B-2, unveiled in November, is scheduled to be operational within the next several years, but Defense Secretary Dick Cheney said yesterday on NBC-TV's _Meet the Press_ that full production would not start in the 1990 fiscal year as planned. Asked whether he would consider killing the program, Cheney replied, "We're going to postpone actually going into full procurement because I'm not comfortable with the program yet, there are a lot of technical problems with it, and it is extremely expensive. And until I have time to review it, which I've not yet had, I'm not prepared to make that judgement." The B-2's flying-wing design is an updating of Northrop's YB-49 aircraft, a 1940s-era prototype bomber that the Air Force killed before production began. The B-2's shape is naturally unstable, and the lack of a tail means it will be much harder to control than a conventional airplane. "It is essentially a boomerang," said James W. Kelley, a former Northrop aerodynamicist. "Once it goes into a spin, it cannot recover." B-2 skeptics question both the plane's radical flying-wing design, first revealed a year ago, and the Air force's decision to save money by going straight from the drawing board to the production aircraft. Historically, new aircraft designs are tested with a series of custom-built planes, each flown and modified until all major problems have been eliminated. Only then does production begin. But in the case of the B-2, about a dozen planes are under construction, although not a single one has flown, several sources said. In recent years, experts have urged the military to build prototypes to let them "fly-before-buy," confirming the designs before committing billions of dollars to production. Prototyping should be done "to uncover operational as well as technical deficiencies before a decision is made to proceed with full-scale development," the presidentially appointed Packard commission said in its 1986 study critical of Pentagon purchasing. But while the Air Force is requiring prototypes for its fledgling and highly secret Advanced Tactical Fighter, it does not believe the B-2 needs them. "It was determined because of its revolutionary technology and the highly sensitive nature of the program that prototyping was not the best way to go," DeFrank said. The secret nature of the program prevented further elaboration, he said. Others contend the plane's radical flying-wing design and high price tag demand prototyping. "A $70 billion program with no prototypes?" asked an incredulous Thomas S. Amlie, an Air Force engineer at the Pentagon, who said computers and models could not replicate the rigors of flight. "Of course we should prototype. We ought to fly one, and wring the hell out of it, with zero-zero ejection seats so the pilots can eject at zero altitude and zero air speed and live through it." Amlie dismissed Air Force arguments that there were classified reasons why prototyping the B-2 makes no sense. "They always say there are classified things that we can't know about because we don't have the clearance," Amlie said. "Well, I've been in the business for 37 years, and every time someone has told me that it turns out they were lying." But Northrop says its battery of high-powered computers, whose data base contains drawings of all of the B-2's parts down to the smalles rivet, has "systematically eliminated" most of the risk inherent in a new aircraft design. With the computers, design changes can be made before production begins. Such changes are particularly painstaking aboard the B-2, where the plane's radar-evading design requires a frozen exterior shape into which all of the plane's systems and weapons must be crammed. "Given all the aerodynamic and performance compromises they've had to make to reduce the radar cross-section of the B-2, you're just flying much closer to the margin," said Pike of the Federation of American Scientists. "That's precisely why you need to do prototyping." "It's very strange that they're not being required to prototype," added Joseph V. Foa, an aeronautical engineer at George Washington University who first studied flying wings 40 years ago. "When you have an aircraft that's going to cost a half-billion dollars apiece, it's a good idea to prototype. Pike said recurring delays -- the plane's first flight originally was set for 1987 -- showed that Northrop's computers had not eliminated the B-2's problems. "That tells me this thing is no different from anything else," he said. "Just because it looks right on the computer screen doesn't mean that it's going to work in the real world." Without prototyping, the Air Force -- if it discovers problems -- will argue that the $20 billion investment it already has made in the program requires repairs instead of cancellation, Pike said. "They're basically front-loading the program so that regardless of what the test results are, they'll already have spent so much money on it that it will be difficult to cancel," Pike said. "You're paying to have the work done twice -- first time to do it wrong, and then the second time to do it right." Stephen W. Thompson, (215) 898-4585 [no relation to Mark], Institute for Research on Higher Education, U. Pennsylvania, Philadelphia, PA 19104
Here is another addition to the list of risks of information age. There is an article in Thursday morning edition [May 4] of San Jose Mercury News titled "Member learns the hard way: American Express is watching". It described how American Express called a member to voice their concern that he might not be able to pay their recent bill. American Express was able to access his checking account and find that he had less than what was owed to them. His card was temporarily "deactivated" after the member refused to give any financial information except that he would pay up the bill with cash when it came in. Apparently, the card application, in finer print, declares that "[American Express reserves] the right to access accounts to ascertain whether you are able to pay the balance". After some arguments with the company, the member comments that "I learned a lesson: My life is not as private as I thought". First, this is news to me. I hold an AmExp card, and I wasn't even aware that my accounts are constantly being checked. Second, how could the banks dish out information on the account holders to third parties without proper authorization? Sundar Iyengar, Microprocessor Design, Intel, Santa Clara, CA 95051
I was tracing the phone wires in my house yesterday afternoon trying to find out why my phone was "off-hook" when all of the phones were actually hung up. Just before the lines enter my house I found a gray box labelled "Telephone Network Interface". Curious, I opened the box to find two RJ-11 modular phone jacks with black connectors in them that were held in by clips. I popped the clip, unplugged the plugs and plugged in a normal phone. Lo and behold, a dial tone! I wandered around the neighborhood a bit and found a few more of these boxes. Looks like you can wander around Boston with a phone, plug into someone's circuit, and make as many phone calls as you like. Who needs lineman's equipment? -David C. Kovar, oOffice of Information Technology, Harvard University
* COMPASS '89 * * JUNE 20th - June 22nd, 1989 * * * * NATIONAL INSTITUTE OF STANDARDS * * AND TECHNOLOGY (formerly NBS) * * Gaithersburg, MD * * * * PROGRAM * * MONDAY, 19 JUNE 1989 * Meeting of the Tri-services Software Safety Working Group * TUESDAY, 20 JUNE 1989 * 0730 REGISTRATION 0900 CALL TO ORDER, General Chair---Dario DeAngelis, Logicon 0910 OPENING REMARKS Honorary Chair---The Honorable Tim Valentine, Chairman for the House Subcommittee on Transportation 0930 PROGRAM OVERVIEW Program Chair---John C. Cherniavsky, Georgetown University 0940 INTRODUCTION OF KEYNOTE SPEAKER AND PANEL Chair, COMPASS Board---H.O. Lubbes, Space and Naval Warfare Systems Command 0950 KEYNOTE ADDRESS "Computer Assurance: Safety, Security, Economics" Allen Hankinson, National Institute of Standards and Technology 1130 KEYNOTE DISCUSSION PANEL: Peter Neumann, SRI International Nancy Leveson, UC Irvine and MIT Allen Hankinson, NIST Michael Brown, Naval Surface Warfare Center 1400 Special Presentation - Computer Related Risk of the Year "Misplaced Trust in Computer Systems" Peter Neumann, SRI International 1430 Minitutorial "Formal Analysis of Safety" Nancy Leveson, UC Irvine and MIT 1600 Software System Safety in the Military Chair---Michael Brown, Naval Surface Warfare Center * Software Safety Handbook Archibald McKinlay VI, McDonnell Aircraft Corporation * Role of the System Safety Manager in Software Safety, Bruce Hill, Consultant 1730 ADJOURN 1900 BANQUET * "It is June 1989. Do you know what your computers are doing?" Peter Neumann, SRI International * WEDNESDAY, 21 JUNE 1989 * 0900 SOFTWARE SYSTEMS SAFETY: Chair---Nancy Leveson, MIT and UC Irvine * Software Safety Goal Verification Using Fault Tree Techniques: A Critically Ill Patient Monitor Example Brian Connolly, Hewlett Packard * Using Petri Net Theory to Analyze Software Safety Case Studies Wade Smith and Paul Jorgensen, Consultants * VMM Concepts Revisited Marvin Schaeffer, Trusted Information Systems 1100 VERIFICATION, VALIDATION, AND TESTING Chair --- Dolores Wallace, NIST * RM 2000 Approach to Software Major Sue Hermanson, USAF * Condition Testing for Software Quality Assurance K.C.Tai, North Carolina State University * Helping the Army Succeed Through Software V&V Richard O'Reagan and Michael Edwards, Teledyne Brown Engineering * Experimental Evaluation of Six Test Techniques Linda Lauterbach and B. Randall, Research Triangle Institute 1430 NEW DIRECTIONS Chair---Richard Hamlet, Portland State University * Access Control and Verification in Petri-Net Based Hyperdocuments P. David Stotts and Richard Furuta, University of Maryland * Unit Testing for Software Assurance Richard Hamlet, Portland State University * Validation Through Exclusion: Techniques for Ensuring Software Safety John C. Cherniavsky, Georgetown University * A Simple Way of Improving the Quality of Login Security Khosrow Dehnad, AT&T Bell Laboratories 1700 RISK ASSESSMENT Chair---Janet Dunham, Research Triangle Institute *Risk Analysis: Case Studies of Two Approaches with an Expert System Based Tool Jane Radatz, Logicon 1800 ADJOURN * THURSDAY, 22 JUNE 1989 * 0830 SYSTEM VALIDATION Chair----Martha Branstad, Trusted Information Systems * Techniques for Data and Rule Validation in Knowledge Based Systems Jong P. Yoon, University of Florida * How to Qualify Knowledge Based Systems Claude Vogel, Cisi Ingenierie * Description of a Formal Verification and Validation Kenneth Lindsay, Magnavox Electronic Systems * Taxonomy of the Cause of Proof Failure in Applications Using the HDM Methodology Kenneth Lindsay, Magnavox Electronic Systems 1100 HARDWARE AND REALTIME VALIDATION Chair----Thomas F. Buckley, University of Leeds * Programming a Viper Thomas F. Buckley, University of Leeds * Formal Verification of Microprocessor Systems Mandayam Srivas, Odyssey Research Associates * Prospects for Verifying the PSN Code Stephen Crocker, Trusted Information Systems * Requirements for Process Control Protection John McDermott, Naval Research Laboratory 1430 VERIFICATION OF SYSTEM FEATURES Chair---H.O.Lubbes, Naval Research Laboratories * Assurance for the Trusted Mach Operating System Martha Branstad, Trusted Information Systems * Verifying Asymptotic Correctness Mark Howard and Ian Sutherland, Odyssey Research Associates * Security Analysis of a Token Ring Using Ulysses Daryl McCullough, Odyssey Research Associates * Penolope: An Ada Software Assurance Editor Carla Marceau, Odyssey Research Associates 1700 PANEL ---- ADVANCES IN FORMAL SOFTWARE ASSURANCE TECHNIQUES Chair---John C. Cherniavsky, Georgetown University Panel Thomas Buckley, Leeds University Steven Crocker, Trusted Information Systems Darryl McCullough, Odyssey Research Associates Mandayam Srivas, Odyssey Research Associates 1800 ADJOURN * FRIDAY, 23 June 1989 TUTORIALS* 0900 TUTORIAL * A Guide to VIPER, A Verifiable Integrated Processor for Enhanced Reliability - or - Why, How, and Wherefore of Using a Formally Proved Microprocessor for High Integrity Control Systems Thomas F. Buckley, University of Leeds Jon Wise, Charter Technologies 0900 TUTORIAL * Formal Specification and Verification of Ada Programs David Guaspari, Odyssey Research Associates Carla Marceau, Odyssey Research Associates 1200 ADJOURN FOR FULL REGISTRATION INFORMATION, SEND NETMAIL TO firstname.lastname@example.org, or FTP KL.SRI.COM, get stripe:<risks>COMPASS.INFO. [I edited out the coffee and lunch breaks for brevity and nonredundancy. PGN]
Please report problems with the web pages to the maintainer