[From the [Toronto] Globe and Mail, 10 April 1986] BLUNDER BY ALBERTA COMPUTER LEADS TO $36 MILLION MISTAKE A botched computer operation jeopardized the [Canadian province of] Alberta Government's ability to keep track of vehicle licence revenue, causing $36 million too much to be reported in a bank balance, the province's Auditor-General reported yesterday. The Solicitor-General Department's new motor vehicles computer system was designed with little help from department accounting staff, an omission which ``undoubtedly'' led to many of its weaknesses, said Auditor-General Donald Salmon. The division's bank balance was shown at $48 million on March 31, 1985, when it was actually $12 million. In addition, the vehicles division lost track of accounts which could not be immediately processed, and unearned revenues were misstated by $2 million in March of 1985. ``These and other ancillary problems were caused largely by insufficent direction and control by senior financial management,'' the report said. The Auditor-General picked up similar problems in 1981-82 in a massive new computer system developed to keep track of about $2 billion a year in natural gas royalties. Oil revenues were miscalculated in a confused federal-provincial transfer of information involving three different price categories under the old regulated pricing system. The governments later agreed to forget it rather than try to sort out the mess. ``The province didn't lose money,'' Mr Salmon said. ``You could probably say the producers lost some . . . but we did not quantify.''
This arises out of a discussion in mod.legal over the meaning of UNIX as a trademark, and how it (the name) might/might not pass into the public domain by becoming a generic descriptive term for a type of operating system rather than refering to a specific product of AT&T. One of the postings which I quote below raised the broader question of the use of postings to a computer network as evidence. In a recent posting (Message-ID: <8604011618.AA15083@bu-cs.ARPA>), Barry Shein said the following: "What immediately occurs to me is that if I were an ATT lawyer I would squirrel away the note imploring people not to attribute UNIX as a (whatever) of (whomever.) It could prove very useful to open an argument that any appearance of it coming into common use was in fact a conspiracy on the part of the technological community." I have no idea of the likelihood of the "conspiracy" defense working to hold onto AT&T's trademark, however the part about holding onto the note got me to thinking. Does anyone out there know if any precedents have been set for the admissability/inadmissability of computer files as evidence in court? I, for one, find the thought that some court of law might, in ignorance, accept computer files as evidence frightening. Certainly on UNIX if you can get access to a privileged account, whether legally or illegally, you can change anything on the system, including editing i-node entries to alter creation dates, etc., with no way I can think of of proving that alterations were made unless the hacker does something extra-ordinarily stupid. I suspect that the same is true of most other systems. No matter how good system security is, given sufficient knowledge of how it works, it is breakable. Coupled with the unfortunate tendency of the layman to accept whatever comes out of a computer as gospel, this provides some very strong reasons for not trusting computer files as evidence, but considering the growing number of transactions being performed by/on computers, there are, or soon will be, a great number of areas where the computer's audit trail may be the only evidence of a transaction. Have any precedents been set already, and if not, what do people think the solution is? Kathryn Smith (...decvax!gsg!kathy) General Systems Group Salem, NH [This is a very valid question. The crypto community has all sorts of techniques for crypto sealing for integrity and crypto authentication. Reasonable techniques exist to give some better assurance, but there are always going to be some internal vulnerabilities. However, since most legal and administrative people do not yet recognize the ease with which on-line evidence — including audit trails — can be altered, and for other reasons as well, these techniques are not yet in widespread use. PGN]
In an article in the Sunday San Diego Union, Gregory Fossedal (Copley News Service) discusses the "rapid advance of SDI." He indicates that progress is good enough that a "decision to deploy a Star Wars defense ... could be made before Ronald Reagan leaves office." He describes some progress made in lasers and other hardware areas. He then goes on to discuss progress by software engineers, and says that "concepts in computer software ... have leaped ahead." He indicates that critical arguments "...that 'a single error' could cripple the whole shield apply only to outmoded types of unwieldy, highly centralized software. Thanks to new software ideas, Star Wars defenses need not be run by a grand central brain." --Walt Thode (thode@nprdc) [Announcements of great BREAKTHROUGHS often coincide with great BREAKDOWNS — in communication and common sense. This one is being hyped like a great BREAKFAST cereal — distributed Wheaties are better than old Wheaties, the breakfast of chumpions. Don't put all your eggs in one basket — just use thousands of baskets instead, and train the hens to BREAKDANCE in space. But don't forget to distribute the roosters as well. Walt, thanks for the enlightenment. I note that in principle there are indeed some software engineering advances, but nothing that GUARANTEES that distributed systems are sound — especially in their operating environments. The tradeoffs are very complex, and thus this is not a simple discussion. Many problems of centralized systems reappear in other guises in distributed systems, and wonderful new problems arise. Perhaps some day we will have a dispassionate, technically motivated analysis — although many of the arguments are nontechnical. PGN]
One of my colleagues, a visiting prof. from the UK, bought a new Ford Escort in mid-February and at the same time purchased the "Extended Warranty" package. Following a trip to Florida for Spring break, the vehicle broke down outside Daytona (that may suggest this is a put-up job!!) on Saturday afternoon March 29th (also Easter Weekend). Calling the 800 number he was referred to a specific repair shop. On arriving there the owner called the 800 number to confirm the warranty and was told that there was no record of a warranty "in the computer" and that any additional enquiries would have to wait until Monday. They stayed in a hotel over the weekend (at a high rate since they had no reservations and limited means of transportation) and on Monday were again informed that there was no record of their warranty. It took most of the rest of that day to have the dealer from whom they purchased the car to confirm that ARTh a warrenty did exist and to have the repair shop agree to START the repairs. It turns out that the dealer doesn't send in the warranties until the end of each month, and the backlog doesn't allow the warrantor to get them in the computer for perhaps another month. This is probably based on the probability that a new car won't need repairs in the first two months and in any case the owner would probably be close to home still! Here is a typical case of having a computer in the system and thus being able to "hide" behind it. By the way, check you own extended warranty to see if it covers the cost of hotel accomodations! Also, I am still researching the Melbourne Bridge Failure for you — I have got the sequence of events and a precis of the findings of the Royal Commission which blamed the failure on a computer program, but I am waiting for a copy of the actual report before I send you more. The sequence of events is well documented in the London Times but I am not sure I want ot trust their reporting on this about the program use until I see the report. JAN PS. Did you see the Hackers Report in CACM this month? [Yup. Arrived today.]
According to a NYT article reprinted in the Daily Illini, a local student newspaper, the phone system in Hackensack N.J. experienced a problem with billing long-distance phone calls from pay-phones. I quote: Technology in an electronic switching center here failed New Jersey Bell, and for nearly two months perhaps half the international calls placed from 400 pay phones around town went through without charge, according to Ted Spencer, a spokesman for the company. ``Apparently a problem developed in a computer program - in the software,'' Spencer said. ``We don't have a record of the calls that got through. They bypassed the billing system.'' Does anyone have anymore in-depth information concerning this? Several people who used the loop-hole were arrested and charge with theft of services. Dirk Grunwald, Univ. of Illinois
Please report problems with the web pages to the maintainer