Subject: RISKS DIGEST 17.56 RISKS-LIST: Risks-Forum Digest Tuesday 19 December 1995 Volume 17 : Issue 56 FORUM ON RISKS TO THE PUBLIC IN COMPUTERS AND RELATED SYSTEMS (comp.risks) ACM Committee on Computers and Public Policy, Peter G. Neumann, moderator ***** See last item for further information, disclaimers, etc. ***** Contents: Navy hacked by Air Force (Darrell D. E. Long) Japanese breeder-reactor incident (Chiaki Ishikawa) Re: Montgomery County, PA, voting machines (Gary Greenhalgh) Definitions for hardware/software relibility engineering (Meine van der Meulen via K. van de Wetering) Medical Diagnosis by Computer (amplification) (Gretchen Herbkersman) Pay online, release your SSN (Robert Mayo) Indelible words (Brian Hawthorne) Re: Another Sign Spoof (Don Root, Coleman) Re: a well-managed risk (Andrew Koenig) ABRIDGED info on RISKS (comp.risks) ---------------------------------------------------------------------- Date: Mon, 18 Dec 95 13:54:26 -0800 From: "Darrell D. E. Long" Subject: Navy hacked by Air Force http://www.telegraph.co.uk/et/ A few clicks and then the e-mail message entered the ship's control system... War of the microchips: the day a hacker seized control of a US battleship BY SIMPLY dialing the Internet and entering some well-judged keystrokes, a young US air force captain opened a potentially devastating new era in warfare in a secret experiment conducted late last September. His target was no less than gaining unauthorised control of the US Navy's Atlantic Fleet. Watching Pentagon VIPs were sceptical as the young officer attempted to do something that the old Soviet Union had long tried to do and failed. He was going to enter the very heart of the United States Navy's warships - their command and control systems. He was armed with nothing other than a shop-bought computer and modem. He had no special insider knowledge but was known to be a computer whizzkid, just like the people the Pentagon most want to keep out. As he connected with the local node of the Internet provider, the silence was tangible. The next few seconds would be vital. Would the world's most powerful navy be in a position to stop him? A few clicks and whirrs were the only signs of activity. And then a seemingly simple e-mail message entered the target ship's computer system. First there was jubilation, then horror, back on dry land in the control room at the Electronic Systems Centre at Hanscom Air Force Base in Massachusetts. Within a few seconds the computer screen announced "Control is complete." Out at sea, the Captain had no idea that command of his multi-million-dollar warship had passed to another. One by one, more targeted ships surrendered control as the codes buried in the e-mail message multiplied inside the ships' computers. A whole naval battle group was, in effect, being run down a phone-line. Fortunately, this invader was benevolent. But if he could do it ... Only very senior naval commanders were in the know as the "Joint Warrior" exercise, a number of experiments to test defence systems, unfolded between September 18-25. Taking over the warships was the swiftest and most alarming of the electronic "raids" - and a true shock for US military leaders. "This shows we have a long way to go in protecting our information systems," said a senior executive at the airbase where the experiment was conducted. The exact method of entry remains a classified secret. But the Pentagon wanted to the first to test the extent of their vulnerability to the new "cyberwarriors" - and had the confidence to admit it. Now they believe they know what they are dealing with and the defences are going up. ET | Front | News | World | Features | Sport | City | What's new | Help | ET search | Gazette | Back Reply to Electronic Telegraph - et@telegraph.co.uk Electronic Telegraph is a Registered Service Mark of The Telegraph plc ------------------------------ Date: Tue, 19 Dec 1995 19:47:58 +0900 From: Chiaki Ishikawa Subject: Japanese breeder-reactor incident Japanese breeder reactor, Monju had an accident in which the coolant Natrium (Na, Sodium) spews out of yet to identified rupture in the secondary cooling loop. It occurred early this month. What surprised me was this: The operator has been prepared for a major accident in which 150 cubic meters of Natrium (Sodium) is breached and had some large-scale major accident simulation and prepared the operator manual. One of the accident simulation used a real building and from what I read was a very solid experiment. So they were quite confident that they are ready for mishap. On this particular accident earlier this month, the amount of sodium spewed out was about 1 cubic meters. Much smaller than the operator expected for a major accident, and as a matter of fact, the sensors were slow to catch the escaping sodium that filled the small room where the rupture occurred. Because of this the operators on site were slow in finding that the sodium indeed esacaped from the closed loop. Thus the shutdown of the reactor occurred about 1hour 40 minutes after the smoke-detecting fire alarm sounded. (The operator's manual stated if the sodium escapes, shutdown procedures are to be taken. But it doesn't say exactly what to do if fire alarm sounded in the house that houses part of the secondary loop.) By reading today's newspaper Asahi Shimbun, I came to the conclusion that the operating company was not prepared for the accident of this minor scale, so to speak. When we hear that someone was preparing for and ready to cope with an accident of certain quantitative index, say, X, we naturally assume that all smaller accidents with index x (< X) will be dealt with by that someone without problems. I.e., All accidents such that the index of severeness x is 0 <= x < X they are ready! (or are they?) This is NOT true. Come to think of it, this is obvious. I wonder why I and many others was led to believe that the breeder reactor operating company was ready for smaller accident at all. This seems to me a lapse of judgement when we hear quantitative index for accident severeness. We should probably never trust qunatitative index when we talk about risks. This is because we can't tell if the all the accident upto certain severeness is considered and/or if only certain sampled accidents of certain severeness are considered. (My English is not as good as to convey the nueance, but I hope the readers of RISKS notice this (almost) blind-faith or lapse of judgement. At least, I didn't realize they didn't pay attention to smaller accident!?) BTW, spewed sodium (500 degrees C when they escaped the cooling pipe) reacted with oxygen and water vapor in the air and punched holes in the air duct and MELTED(!) catwalks. From the video taken by alarmed local city officials who investigated the site past midnight after the accident, it looks a real mess although the amount is indeed small and no major radio active material escaped. Chiaki Ishikawa Personal Media Corp. Shinagawa, Tokyo, Japan 141 ishikawa@personal-media.co.jp ------------------------------ Date: Mon, 18 Dec 1995 15:07:39 -0800 From: ggreenha@ix.netcom.com (Gary Greenhalgh, Microvote) Subject: Re: Montgomery County, PA, voting machines (Finegold, RISKS-17.50) On November 7, 1995, we experienced several problems in Montgomery County with both the electronic voting machines and software. However, none of these problems affected the basic integrity of the system AS MONTGOMERY COUNTY HAS NOW OFFICIALLY CERTIFIED ALL OF THE ELECTION RESULTS. IN ADDITION, THE COUNTY HAS RECEIVED NO CHALLENGES TO THE ELECTRONIC VOTING MACHINE RESULTS! First of all, it should be noted that the County, against our advice, kept some 60 machines (Of a total of 900 purchased by the County) as spare machines. These machines could have, easily, been deployed to high turnout precincts which would have reduced voting lines considerably. Second, there were NOT "Massive Voting Machine" problems as reported by the press. Of the 822 machines used on election day, we had 48 DOCUMENTED IN WRITING MACHINE service calls, most of which were handled easily. In fact, our voting system EXCEEDED THE FEDERAL ELECTION COMMISSION STANDARDS FOR AVAILABILITY ON ELECTION DAY! The "150 service calls" reported by the press dealt with ALL questions received at the County's voting machine warehouse regarding calls from precinct workers including ballot questions, deployment of machines, etc. In researching the above machine calls on election day, we have now TENTATIVELY concluded that many of the service calls were "Power failure" calls that involved an anti-moisture spray (Called a Conformal Coating) which may have leaked into the contact area between the main controller chip and the power board in certain machines. This spray is absolutely essential to insuring equipment integrity, especially in areas of the country with high humidity, so the spray or coating (The power boards are dipped in the coating) has to be done. However, THIS IN NO WAY AFFECTED THE INTEGRITY OF THE VOTE COUNT. Indeed, there is a very simple procedure for precinct workers getting the machine up and running again on election day. If this is discovered as the main culprit, we will resolve this problem by "Cleaning" all these contacts in each of the 8,070 machines now in use throughout the United States. As to the election night software results, please note, first of all, that our unofficial election night results were being "Dumped" into the Microvote Media Package that displayed the results, electronically, for the press. We have now discovered that a software operator, in adding 9 machines to the inventory at about l0:00 a.m on election morning, inadvertently added the machines without assigning them to proper precincts and without "Reinitializing" the system. THus, when we read some of the machine cartridges from the machines, the first unofficial totals were posted incorrectly. When we discovered the problem, we reinitialized the system and the results were correct. Unfortunately, by the time we discovered the problem, certain press reporters had already ran with the original incomplete, unofficial, and inaccurate election results. Microvote takes the blame in that our software should not allow an operator to add voting machines in this fashion and we also take the blame for not providing Microvote staff to monitor the election day process. Neither of these will happen again. ------------------------------ Date: Tue, 19 Dec 1995 12:05:22 +0100 (MET) From: K.van.de.Wetering@simtech.nl (K. van de Wetering) Subject: Definitions for hardware/software relibility engineering >FROM Meine van der Meulen SIMTECH kvdweter@simtech I would like to draw your attention to a dictionary of terms I wrote: Meulen, M.J.P. van der, Definitions for Hardware/Software Reliability Engineers, ISBN 90-9008437-1, June 1995, 137 pages. The book is a glossary of terms in the field of hardware/software reliability engineering. It lists existing definitions from standards and other sources, ordered in alphabetical order. Every definition is accompanied by its source. The list of sources is in the back of the book. The structure of the book enables various uses. When writing new standards and reports it enables finding of existing definitions. Many terms come with several definitions, and then comparison deepens insight. Of course, it can also be used as a dictionary where unknown terms can be found. The book has over 1700 entries, lots of them containing more than one definition. As an example I give you the first entries for the letter F: Fail-Safe The built-in capability of a system such that predictable (or specified) equipment (or service) failure modes only cause system failure modes in which the system reaches and remains in a safe fall-back state (IEC65A 122) (IEC65A 94). This is the capacity of the system to remain in a safe condition when a failure occurs or to skip directly into another safe condition (VDE3542) (VDE0801). A concept that defines the failure direction of a component/system as a result of specific malfunctions. The failure direction is toward a safer or less hazardous condition (CCPS). Design features which provide for the maintenance of safe operating conditions in the event of a malfunction of control devices or an interruption of an energy source (). The capability to go to a predetermined safe state in the event of a specific malfunction (ISA-dS84/16N). Pertaining to a system or component that automatically places itself in a safe operating mode in the event of a failure (IEEE Std 610.12-1990). A design property of an item in which the specified failure mode is predominantly in a safe direction (IEC1508). Fail-Safe Shutdown The ability of a PC-system to have its outputs assume a predefined state within a specified delay after detecting the occurrence of a power supply voltage drop or an internal failure (IEC1131-1). Fail Soft Pertaining to a system or component that continues to provide partial operational capability in the event of certain failures (IEEE Std 610.12-1990). Fail-to-Danger An equipment fault which inhibits or delays actions to achieve a safe operational state should a demand occur. The fail-to-danger fault has a direct and detrimental effect on safety (CCPS). Design features which inhibit or delay automatic shut-down of a process on failure of a critical control device or on interruption of energy source. The fail-to-danger fault has a direct and detrimental effect on safety (). Failure A system failure occurs when the delivered service deviates from the specified service, where the service specification is an agreed description of the expected service. A failure, in short, is the manifestation of an error in the system or software (IEC65A 122) (IEC65A 94). The termination of the execution of an established task from a unit as a result of a cause which is located in the unit itself and within the framework of the permitted working conditions (VDE3542) (VDE0801). A system failure occurs when the delivered service deviates from the specified service. The specified service is stated in the service specification and is an agreed description of the expected service (IEC65A 96). Termination of the ability of an item to perform its specified function. OR, Non-conformance to some defined performance criteria (Smith81). Behaviour of a unit which is carrying out some task which is not in accord with its intended or specified function (NTG3004) (TV86). The termination of the ability of an item to perform a required function (BS4778) (O'Connor81). The inability of a system or system component to perform a required function within specified limits. A failure may be produced when a fault is encountered (DO178b). A failure is the manifestation of an error in the system or software (). The delivered service deviates from the specified service (Shell94). The termination of the ability of an item or equipment to perform its required function. Failures may be unannounced and not detected until the next test or demand (unannounced failure), or they may be announced and detected by any number of methods at the instant of occurrence (announced failure) (IEEE Std 500-1984) (OREDA84). The termination of the ability of an item to perform a required function or its inability to perform within previously specified limits (ISO7.30). The event, or inoperable state, in which any item or part of an item does not, or would not, perform as previously specified (MIL721). The inability of a system or component to perform its required functions within specified performance requirements (IEEE Std 610.12-1990). (1) The termination of the ability of a functional unit to perform its required function. (2) An event in which a system or system component does not perform a required function within specified limits. A failure may be produced when a fault is encountered (IEEE Std 729-1983) (IEEE Std 982.1- 1988). A system failure occurs when the delivered service deviates from the intended service. A failure is the effect of an error on the intended service (IEC1508). A failure of any technical unit under consideration occurs if the permissible deviation from the performance target for this unit is exceeded (DIN25424). Failure Analysis Subsequent to a failure, the logical systematic examination of an item, its construction, application, and documentation to identify the failure made and determine the failure mechanism and its basic course (MIL721). Failure, Catastrophic See Catastrophic Failure. Failure Cause The identified original cause of the failure; the circumstances during design, manufacturing, assembly, installation or use that have led to failure (OREDA84). Failure, Common Cause See Common Cause Failure. Failure, Common Mode See Common Mode Failure. Cost of the book is NLG 150, excluding VAT and postage. The book can only be obtained through my office. ir. M.J.P. van der Meulen, n Simtech b.v. Oostmaaslaan 71 3063 AN Rotterdam The Netherlands Phone: +31-10-4244386 Fax: +31-10-4244253 Email: info@simtech.nl ------------------------------ Date: Tue, 19 Dec 95 09:01:28 PST From: odinba!odin!gretchen@uunet.uu.net (Gretchen Herbkersman Dept 5428) Subject: Medical Diagnosis by Computer (amplification) Sorry, I thought I'd sent more info. Here's a quote from the Wall Street Journal, Monday, Dec. 18, 1995, page B1: "The next time you visit the doctor with an aching back, it may be a computer that decides you don't need tests and sends you home with painkillers and a list of exercises. "So-called smart software programs, packed with data on the treatment of maladies from heart disease to depression, are the newest tools of managed-care companies in their constant fight to control expenses. Eventually there may be no arguing with them -- a doctor will follow the computer's advice or find another health plan to work for. "The programs are operated either by an HMO clerk who answers an 800 number when your doctor calls to authorize treatments and tests, or, increasingly, directly by doctors in their offices. They are a powerful new tool for deciding whether your case justifies what you're claiming." As a sometime programmer and dabbler in databases, I find this news exceedingly disconcerting. It is the best reason I can think of for staying well. At any rate, please read the rest of the article. ------------------------------ Date: Mon, 18 Dec 95 17:10:26 -0800 From: Robert Mayo Subject: Pay online, release your SSN If you pay a bill on-line using Quicken's On-Line bill payment, you wouldn't suspect that you are giving your SSN out. But you are! If the merchant receives one check on a given day from Quicken users, your SSN is safe. But if they receive more than one check on a given day, they get a listing of all the checks, each sender's bank account number, name, and Social Security Number. Having this information all in one place is especially conducive to fraud. Information is available at the URL: http://www.mc4.com/mayo/quick.html --Bob [usual disclaimers omitted] ------------------------------ Date: Tue, 19 Dec 95 21:49:22 0500 From: Brian Hawthorne Subject: Indelible words On Wed, 18 Sep 91 11:57:20 EDT I wrote to RISKS (now archived in http://catless.ncl.ac.uk/Risks/12.36.html) a brief note on: "The risks of a computer-based forum" I said: >Many people seem to approach e-mail and submissions to forums like RISKS as >informal conversation. Given the persistence of the typed word, however, it >may often be more appropriate to consider these forums as un-refereed >journals. Today, I did a search for myself on Digital's new search site (http://www.altavista.digital.com). Much to my surprise, one of the first items was a link to a posting from 1991 where I mused on the persistence of the typed word in on-line forums. If you search for a person's name on the Alta Vista site, you can find all the newsgroups they post to, as well as many past postings that have made it to archive sites on the Web. The risk? The same as it was in 1991, but magnified many fold. ------------------------------ Date: Tue, 19 Dec 95 01:23:45 PST From: der@oes.ca.gov (Don Root) Subject: Re: Another Sign Spoof (RISKS 17.54) I drive the subject segment of I-80 in Richmond, CA on a regular basis. I suspect the Changeable Message Sign in question belongs to a contractor involved with the I-80 reconstruction project. If so, this sign is trailer mounted, with provisions for a local key pad. Other versions of this trailer-mounted sign type have cellular phone modems for remote message configuration. The RISKs should be obvious... Don Root Calif. Office of Emergency Services Telecommunications Branch ------------------------------ Date: Tue, 19 Dec 1995 12:12:00 +0000 From: USBTISEA!MSUSTSEA!Coleman@usbti1.attmail.com Subject: Re: Another sign spoof (Levy, RISKS-17.55) It's been my understanding that a lot of these sign actually have some sort of a modem connected to a radio. If you know the frequency and can transmit data, you can control the sign. Apparently, because most people would not have this capability, no security was built into these type of signs. Some signs like "Traction tires required" are triggered in the same fashion except just using DTMF tones to turn them on or off. In either case, simple monitoring of these frequency can result in the knowledge of what needs to be done to control them, or maybe just record what was sent and play it back later. ------------------------------ Date: Mon, 18 Dec 1995 19:17:17 EST From: Andrew Koenig Subject: Re: a well-managed risk (Whittle, RISKS-17.47) Jerry Whittle points out that aircraft fuel gauges may be inaccurate. I must confess I didn't press on this particular issue, because (a) there are several reasons the original estimates might be off and (b) there are several ways to cross-check the estimates when determining (mid-flight) what has happened. For example, an engine might be burning more fuel than it should, but if it is, one would expect that to show up in two places: the fuel flow gauge for that engine and the total fuel remaining. It might also show up in other engine instruments. Or the flight might be progressing more slowly than intended because of unforecast headwinds. But that would show up in increased time between checkpoints, not to mention the inertial navigation system. In other words, the fuel gauges might be inaccurate, but if all the relevant instruments agree, they're probably right. Of course one should treat the amount of fuel on board as being known more accurately than the gauges might report. Nevertheless, the procedure the crew actually followed must be safer than just assuming that you still have enough fuel to get there because you did when you left. Jonathan Corbet says he would prefer it if the flight plan showed the intended destination as the official one rather than the `virtual destination.' Then the crew reviews the relevant data mid-flight and diverts if things don't look right. In theory, of course, there's no difference between the two plans. In each case the crew takes off intending to land (for real) at the same place and intending to divert to the same place if there is an anomaly in the fuel consumption. There is a difference, though, and that is in what happens if for some reason a decision is not made or not communicated to the folks on the ground. In this particular case, for example, the decision to continue to the intended destination is not entirely up to the flight crew: the airline's dispatchers on the ground must agree as well. I understand that kind of thing is routine in airline operations and provides a way to ensure that the decision was actually made and not just documented :-) In this case, if for some reason the crew cannot reach the dispatcher on the radio, then there is no decision and the flight must therefore continue to its original `official' destination. So the difference between the two strategies is that in the case I described, the flight does the safe, inconvenient thing unless everything can be proven to be working properly. In the strategy Jonathan Corbet describes, the flight does the convenient, potentially unsafe thing unless something can be proven to be working improperly. Those two strategies may be the same in theory, but they sure differ in practice. --Andrew Koenig ark@research.att.com PS: This follow-up is very late for an amusing reason: I mistakenly sent it to risks@csl.cri.com instead of risks@csl.sri.com. Normally I would have found out about that immediately. In this case, however, cri.com picks up its mail by polling its service provider, and has not been doing that very often. As a result, nothing happened for a week or so and then I got back automatic mail saying that my message had not been delivered and it would keep trying. So I compounded my error by trying to send mail to postmaster@csl.cri.com, which had the expected effect: no response for a week and then another automatic message. Finally I mentioned the problem to Steve Bellovin, who said he had no trouble reaching RISKS. I sent him a copy of the bounce message and not having my particular set of blinders, he saw the problem immediately. ------------------------------ Date: 6 September 1995 (LAST-MODIFIED) From: RISKS-request@csl.sri.com Subject: ABRIDGED info on RISKS (comp.risks) The RISKS Forum is a moderated digest. Its USENET equivalent is comp.risks. SUBSCRIPTIONS: PLEASE read RISKS as a newsgroup (comp.risks or equivalent) on your system, if possible and convenient for you. BITNET folks may use a LISTSERV (e.g., LISTSERV@UGA): SUBSCRIBE RISKS or UNSUBSCRIBE RISKS. [...] DIRECT REQUESTS to (majordomo) with one-line, SUBSCRIBE (or UNSUBSCRIBE) [with net address if different from FROM:] INFO [for further information] CONTRIBUTIONS: to risks@csl.sri.com, with appropriate, substantive Subject: line, otherwise they may be ignored. Must be relevant, sound, in good taste, objective, cogent, coherent, concise, and nonrepetitious. Diversity is welcome, but not personal attacks. [...] ALL CONTRIBUTIONS CONSIDERED AS PERSONAL COMMENTS; USUAL DISCLAIMERS APPLY. Relevant contributions may appear in the RISKS section of regular issues of ACM SIGSOFT's SOFTWARE ENGINEERING NOTES, unless you state otherwise. RISKS can also be read on the web at URL http://catless.ncl.ac.uk/Risks RISKS ARCHIVES: "ftp ftp.sri.comlogin anonymous[YourNetAddress] cd risks or cwd risks, depending on your particular FTP. [...] [Back issues are in the subdirectory corresponding to the volume number.] Individual issues can be accessed using a URL of the form http://catless.ncl.ac.uk/Risks/VL.IS.html [i.e., VoLume, ISsue] ftp://unix.sri.com/risks [if your browser accepts URLs.] ------------------------------ End of RISKS-FORUM Digest 17.56 ************************