The Risks Digest

The RISKS Digest

Forum on Risks to the Public in Computers and Related Systems

ACM Committee on Computers and Public Policy, Peter G. Neumann, moderator

Volume 22 Issue 11

Thursday 6 Jun 2002


Impact of inadequate software testing on US economy
Rick Kuhn
"Truncation error" found in GPS code on Int'l Space Station
George White
FBI's Carnivore hampered anti-terror probe
Marc Rotenberg
Sex, Truth and Videotaping
Gary Marx
Kursk submarine: to test or not to test ...?
Ken Knowlton
Deja vu: Stockholm power outage hits high-tech companies
Ulf Lindqvist
Inadvisable instructions from Sun on StarOffice 5.2
John Sullivan
Confirming cricket score reason for delay
R. Jagannathan
Students provide bulk of tech support in schools
More on typos and homographs
Martin Wheatman
Please ignore the anti-shoplifting device!
Mario Hendricks
Re: The Klez Effect
Paul van Keep
Greg Searle
Re: Klez and mail loops
A. Harry Williams
More on Klez
Hal Lewis
Info on RISKS (comp.risks)

Impact of inadequate software testing on US economy

<Rick Kuhn <>>
Wed, 05 Jun 2002 14:53:35 -0400

NIST has released a new study conducted by the Research Triangle Institute
that should be of interest to readers: "The Economic Impacts of Inadequate
Infrastructure for Software Testing".  Comments and discussion are welcome.

Rick Kuhn

 From the summary:

NIST engaged the Research Triangle Institute (RTI) to assess the cost to the
U.S. economy of inadequate software testing infrastructure.  Inadequate
testing is defined as failure to identify and remove software bugs in real
time. Over half of software bugs are currently not found until downstream
in the development process leading to significant economic costs. RTI
identified a set of quality attributes and used them to construct metrics
for estimating the cost of an inadequate testing infrastructure. Two in
depth case studies were conducted. In the manufacturing sector,
transportation equipment industries were analyzed. Data were collected from
software developers (CAD/CAM/CAE and product data management vendors) and
from users (primarily automotive and aerospace companies). In the service
sector, financial services were analyzed with data collected again from
software developers (routers and switches, financial electronic data
interchange, and clearinghouse) and from users (banks and credit unions).
...the annual cost to these two major industry groups from inadequate
software infrastructure is estimated to be $5.85 billion. Similarities
across industries with respect to software development and use and, in
particular, software testing labor costs allowed a projection of the cost to
the entire U.S. economy. Using the per-employee impacts for the two case
studies, an extrapolation to other manufacturing and service industries
yields an approximate estimate of $59.5 billion as the annual cost to the
nation of inadequate software testing infrastructure.

"Truncation error" found in GPS code on Int'l Space Station

<George White <>>
Thu, 30 May 2002 08:45:11 -0300 (ADT)

The following excerpt describes the resolution of a problem in the
GPS attitude control for the International Space Station.

> From Thu May 30 08:28:31 2002
Date: Wed, 29 May 2002 06:43:17 -0700
From: Jacques van Oene <>
Subject: ISS On Orbit status 28-05-2002

"The troubleshooting of the GPS (global position system) attitude
errors has been successful. At first thought to be due to the high
solar Beta angle (where fewer GPS satellites are in sight), the cause
has now been traced to the software, which did not calculate attitudes
correctly due to a truncated parameter. The error was eliminated, and
the system is now working fine, even at high Beta angles, supplying
state vector and attitude data. With Russian concurrence, GPS will now
also provide the periodic updates of the SM's BINS strapdown
navigation and guidance system, instead of requiring lengthy manual
data takes by the crew using the optical PUMA system, thus saving
valuable crew time."

Not to mention the psychological benefits to the crew of one less "area of
concern" in a setting where resolving glitches in a critical subsystem can
easily become an exercise in "staying alive".

George White <> Halifax, Nova Scotia

FBI's Carnivore hampered anti-terror probe

<Marc Rotenberg <>>
Tue, 28 May 2002 17:03:50 -0400


Washington, DC -- An FBI anti-terrorism investigation possibly involving
Usama bin Laden was hampered by technical flaws in the Bureau's
controversial Carnivore Internet surveillance system.  The incident, which
occurred in March 2000, is described in newly-released FBI documents
obtained under court order by the Electronic Privacy Information Center
(EPIC).  A written report describes the incident as part of a "pattern"
indicating "an inability on the part of the FBI to manage" its foreign
intelligence surveillance activities.

An internal FBI e-mail message dated April 5, 2000, and sent to M. E.
(Spike) Bowman, Associate General Counsel for National Security Affairs,
recounts how the Carnivore "software was turned on and did not work
correctly." The surveillance system captured not only the electronic
communications of the court-authorized target, "but also picked up E-Mails
on non-covered" individuals, a violation of federal wiretap law.  According
to the Bureau document, the "FBI technical person was apparently so upset
that he destroyed all the E-Mail take, including the take on [the authorized

The botched surveillance was performed by the FBI's International Terrorism
Operations Section (ITOS) and its "UBL Unit," which refers to the
government's official designation of bin Laden.  The Bureau document
indicates that an official at the Justice Department's Office of
Intelligence Policy and Review (whose name has been deleted) became aware of
the problem, and "To state that she is unhappy with ITOS and the UBL Unit
would be an understatement of incredible proportions."

The reported problem apparently was not the first to arise during the course
of FBI implementation of the Foreign Intelligence Surveillance Act (FISA).
The internal document concludes its report of the "UBL Unit" incident by
noting, "When you add this story to the FISA mistakes covered in [another,
unreleased document], you have a pattern of occurrences which indicate to
OIPR an inability on the part of the FBI to manage its FISAs."

Two Bureau documents written one week later discuss Carnivore's tendency to
cause "the improper capture of data," and note that "[s]uch unauthorized
interceptions not only can violate a citizen's privacy but also can
seriously 'contaminate' onoging investigations" and that such interceptions
are "unlawful."  An FBI lawyer (whose name has been deleted) writes that the
Bureau must "go out of our way to avoid tripping over innocent third party
communications."  The lawyer concludes, "I am not sure how we can proceed to
test [Carnivore] without inadvertently intercepting the communications of
others, but we really need to try."

The Bureau lawyer notes that "missteps under FISA lead to mandatory
reporting to the President's Foreign Intelligence Advisory Board, and such
errancies must be reported/explained/justified to Congress."  The documents
do not indicate whether the "UBL Unit" incident was reported to either body.

Since its existence became public in 2000, the Carnivore system has been
criticized by EPIC and other privacy groups, as well as members of Congress,
because it gives the FBI unprecedented, direct access to the data networks
of Internet service providers.  The FBI has publicly downplayed the system's
potential for over-collection of private communications, although internal
documents released earlier to EPIC confirmed such a risk.  An independent
review of Carnivore commissioned by the Justice Department also found that
the system is capable of "broad sweeps" and recommended technical changes to
address the problem.  Neither DOJ nor the FBI has indicated publicly whether
those recommendations were ever implemented.

The newly-released FBI documents were provided to EPIC on 24 May 2002, in
response to a court order issued by U.S. District Judge James Robertson in
the privacy group's ongoing Freedom of Information Act lawsuit seeking the
disclosure of material concerning Carnivore.  The order directed the Bureau
to conduct a second search for relevant documents after EPIC successfully
argued (over the Bureau's objections) that an initial FBI search was
inadequate and likely overlooked responsive records.

The case is being litigated by EPIC's General Counsel, David Sobel, who
said, "These documents confirm what many of us have believed for two years
-- Carnivore is a powerful but clumsy tool that endangers the privacy of
innocent American citizens.  We have now learned that its imprecision can
also jeopardize important investigations, including those involving
terrorism."  Sobel added, "As we suggested when it first became public,
Carnivore's use should be suspended until the questions surrounding it
finally can be resolved.  Our FOIA lawsuit shows that there's a great deal
about Carnivore that we still don't know."

The newly-released FBI documents are available at:

Contacts: DAVID SOBEL 202-483-1140 x105 WAYNE MADSEN x104

Sex, Truth and Videotaping

<"gtmarx" <>>
Tue, 28 May 2002 20:45:20 -0700

At-Home Spying: Privacy Wanes as Technology Gains
Surveillance may be legal, but is that the only standard?
Commentary by Gary T. Marx, 28 May 2002, *Los Angeles Times*
  [Reprinted with the permission of the author]

Recently in France, a father who was concerned about the possible
mistreatment of his 3-year-old son by a baby-sitter's boyfriend hid a
miniature camera in his home to record any suspicious behavior. The father
found some, but it was not abuse of the child; the camera revealed the
baby-sitter and her boyfriend amorously entangled while the child slept
soundly in the next room.

The father paid a penalty. In France, such videotaping is a violation of
criminal and civil law. The father was arrested and ordered to pay a fine
for invasion of privacy.

Did the father do something wrong? Is there a victim here? As the ubiquitous
advertisements for cameras concealed in teddy bears and other unlikely
places remind us, parents have an obligation to protect their children. A
hidden video camera offers an easy way to do this--to the extent that seeing
is believing. If nothing is found, the responsibly vigilant parents rest
well knowing that the child was not harmed. If something is found, there is
tangible evidence for taking protective and even legal action. Different
privacy standards characterize home and work, as well as areas within
these. The baby-sitter after all was not filmed in her own home but at her
place of work.

Yes, the French parents had alternatives. They could have discussed their
concerns with the sitter, checked to see whether other employers had had
problems with her, banned the boyfriend or simply found another
baby-sitter. If there are some grounds for doubt, why take a chance by
spying? And, as it turns out, the law was on the baby-sitter's side.

In the U.S., the employer largely sets the conditions of work. To use the
French baby-sitter as an example, the camera was in the living room, not in
a bathroom where the expectation of privacy would have been greater. The
place and the video equipment belonged to the parents, and the baby-sitter
willingly came to the home. The images weren't sold on the Internet, used as
blackmail or stolen.

To cast the best light on the father, he wanted to have the evidence in hand
before deciding to fire the sitter. The use of hidden cameras is hardly an
uncommon or exotic means for this. And, had the father been living in the
U.S. instead of France, in most jurisdictions he would have broken no law.

Yet this videotaping, even if well-intentioned and revealing nothing
incriminating, is patently offensive. E.M. Forster captured this well in
noting: "For it is a serious thing to have been watched. We all radiate
something curiously intimate when we believe ourselves to be alone."
Secretly recording people violates their dignity and can put the individual
at an unfair strategic disadvantage.

We assume, or at least morally expect, that under ordinary circumstances
behavior behind closed doors, in darkness and at a distance will be
protected from the eavesdropping of third parties. We also have a right to
assume that interaction and communication are ephemeral and transitory and
are not subject to being captured and preserved through hidden video or
audio means without our knowledge. Another unintended consequence is that
sometimes people seeking specific information--i.e., whether their child is
being mistreated--may observe something even they didn't want to know.

In addition, remotely transmitted signals might be picked up by others in
the vicinity. The behavior of the spy is thus doubly troubling. Not only is
he invading the privacy of those he is watching, but he may unwittingly
enable others to invade as well.

The fact that there is still a legal right to secretly record images in the
U.S. does not mean that it is the right thing to do. We would do well to
learn from the French the general principle of respect for private life, a
principle that holds no matter what new technologies are offered to us that
allow us to spy on others.

Gary T. Marx, a Massachusetts Institute of Technology emeritus
professor, is the author of "Undercover: Police Surveillance in
Comparative Perspective" (Kluwer Law, 1995). Web site: garymarx. net.

Kursk submarine: to test or not to test ...?

<Ken Knowlton <>>
Thu, 30 May 2002 08:23:45 EDT

Another Hubble Telescope story, of sorts: *Time Magazine* (3 Jun 2002)
reports that when the Russian submarine Kursk sank, 23 of the momentarily
surviving crewmen rushed to the floating rescue capsule located in the rear.
But it failed to disengage.  It had never been tested.

Deja vu: Stockholm power outage hits high-tech companies

<Ulf Lindqvist <>>
Thu, 30 May 2002 09:08:17 -0700 (PDT)

In RISKS 21.27, Mar 15 2001, I reported about about a large and
long-lasting power blackout in Stockholm, Sweden:

> A fire in a tunnel containing power cables caused a long-lasting
> blackout for 50,000 people and a large number of high-tech
> companies in several Stockholm suburbs. [...] The largest employer
> in the area, Ericsson, told 11,000 employees to stay at home
> Monday as their workplace had no power.

Guess what - something very similar happened again on May 29 2002 in the
same tunnel, with roughly the same consequences. In a press release, the
utility company Birka Energi claims that this is actually not exactly the
same situation as last time. In March 2001, it was an 11 kV cable in the
tunnel that caught fire, and they claim that the cause of that particular
problem has been eliminated. This time, however, they received a ground
fault indication on a 33 kV cable right before the fire. To their customers,
who will be without power for a couple of days, that information will not
make much of a difference.

The risk is a familiar one: addressing specific symptoms instead of the
underlying vulnerability. PGN's comment about tunnel vision is unfortunately
still appropriate.

Ulf Lindqvist, System Design Lab, SRI International, 333 Ravenswood Ave,
Menlo Park CA 94025-3493, USA +1 650 859-2351

Inadvisable instructions from Sun on StarOffice 5.2

<John Sullivan <>>
Wed, 29 May 2002 23:29:30 +0100

I heard today that Sun are withdrawing the free download of StarOffice 5.2,
so I went over to their website to investigate. Having downloaded the
package (in this case a single .bin file which is actually an executable) I
went to the installation instructions at:

Here we are told:

    Note for Linux and Solaris[tm] customers only:

    It will be necessary for a chmod command to be executed to set
    read, write, and execute permission.

    For example:

    % chmod 777 *.bin

    Once you have set the permissions, you can run the setup, and
    StarOffice 5.2 will do the rest.

One hopes that anyone seriously likely to be at risk of exploit due to this
would not be inexperienced enough to actually follow their instructions

However a sufficiently inexperienced user, even though they are most
probably not running a system shared with an adversary capable of attacking
them because of this, may well absorb that command as some sort of "magic
rune" and issue it at other times in the future when the window of
opportunity is much wider.

Confirming cricket score reason for delay

<"R. Jagannathan" <>>
Tue, 4 Jun 2002 00:48:23 -0700

An interesting article from a "dependability" standpoint.  Mismatch between
what a computer thought was the revised target and what the umpire manually
computed caused a 20-minute delay during which the teams had to play without
knowing the revised target.  The game between India and West Indies was won
by India and the West Indian captain later rued the fact that their team did
not know the score for 20 minutes, which affected how they played.  All this
for a one-run difference between the manual and computed calculation.

The Duckworth-Lewis method is an empirically-derived table (by the two
mathematicians) to compute revised targets when a one-day cricket match gets
shortened by rain or other similar disruptions.

Students provide bulk of tech support in schools

<"NewsScan" <>>
Thu, 06 Jun 2002 08:48:37 -0700

Fifty-four percent of U.S. schools rely on students to provide technical
support for their computer systems, according to a report titled "Are We
There Yet?" (, released yesterday by
the National School Boards Foundation. In 43% of the 811 districts surveyed,
students troubleshoot for hardware, software and other problems, and 39% of
the districts, students are tasked with setting up equipment and
wiring. Nearly as many districts also report that students perform technical
maintenance. The fact that students are providing so much hands-on
assistance is viewed as a "win-win" situation by John Bailey, director of
education technology for the Department of Education. Their tech savvy helps
compensate for a dearth of tech support funding in school budgets and
teachers who are "unevenly prepared for using technology as a tool for
teaching and learning," according to the NSBF, which reports that 69% of the
survey respondents rated new teachers as average or novices in computer
skills. The role reversal signals a shift in the relationship between
teachers and students as online lessons become integrated into the school
curriculum, says Anne Bryant, executive director of the National School
Boards Association: "Teachers become the guide on the side, instead of the
sage on the stage."  [AP 5 Jun 2002; NewsScan Daily, 6 June 2002]

More on typos and homographs

<Martin Wheatman <>>
Tue, 28 May 2002 11:52:02 +0100

I use a national bank in the UK called Lloyds TSB (which is a recent
amalgamation of Lloyds Bank and the Trustee Savings Bank), and I'm naturally
keen to use their online services. However, I misspelt the name earlier this
year (, and was taken to the real  I
didn't immediately spot the typing error until I noticed that the Webmaster,
rather incompetently perhaps (if it were a real scam), was using a free Web
host service (, which displays the redirected Web site
underneath their banner advertising free Web hosting (I use the same company
as they do provide a good, cheap, "no frills" service). Writing a program to
redirect traffic to the end service is trivial (no need to even provide Web
content!), as would filtering usernames and passwords.

It's slightly different to the homograph scam as it relies on the typist to
make errors (ad htat nerver hapens!), rather than providing the links
already embedded in Web content, and feeds off , and probably the fact that there will
(I suspect) be a lot more combinations of typos than homographs. Also, the
simple (but probably unheeded) solution to display mixed character sets
wouldn't work... :( One protection, in the UK at least, is to access
commercial sites thought the "" domain which is managed by Companies
House (effectively the Government), who will only allow registered companies
to use their registered names. Ok, so you could register a company wth the
misspelt name, but I suspect that Companies House (and the company being
scammed) will no doubt be interested in the reasons for the similar name
(probably using the same mechanism as protection for trademark laws /
passing off?).

P.S. As a responsible citizen, I notified my bank (it was rather scary!),
and although the site is still registered it is no longer used to redirect
traffic (whether these two events are linked, I still don't know -
presumably the Bank is still bothered by adverse publicity).

Please ignore the anti-shoplifting device!

<Mario Hendricks <>>
Thu, 06 Jun 2002 13:26:46 -0400

The other day I went into my local Apple store to buy an external floppy
drive.  On my way out of the mall to my vehicle, I decided to take the
shortest route, through a nearby Eddie Bauer store.  As a I walked into the
store, the anti-shoplifting device sounded an alarm.  As there were no other
people within 5 or 10 meters, I realized that my recent purchase (from
Apple) must have set off the alarm.  I also realized that the alarm would
likely sound as I exited at the other end of the store.

When I got near the parking lot exit, I encountered an Eddie Bauer employee.
Rather than to get accosted as attempting to shoplift by this employee, I
decided to warn him that the anti-shoplifting device would probably sound as
I exited.  The employee saw that I was carrying an Apple Computer bag (they
have very distinctive shopping bags) and responded along the lines of "Oh,
yeah, stuff from Apple always set off the alarm."  Sure enough, when I left
the store the alarm sounded.

The risks of such a false positive alarm are obvious.  The fact that
employees know to ignore the alarm (in at least this one case) raises even
more risks.  Of course, if anyone should ever want to shoplift from this
store, there seems to be a tested procedure.  You would think that either
the store's management or the anti-theft device manufacturer would be
concerned about such issues.

Re: The Klez Effect

<"Paul van Keep" <>>
Tue, 4 Jun 2002 10:43:54 +0200

Yesterday I received a bounce from because I 'allegedly'
sent a Klez.H infected e-mail to Peter. Of course I didn't (I thoroughly
removed Outlook from my system and only use Polarbar, a Java mailer). But
the interesting side effect is that the virus firewall nicely bounces the
message to me, first stripping the infection payload. But what doesn't get
stripped is the document that Klez attaches to each mail it sends out. So
with each bounce I get a nice bonus.  The Klez e-mails with me as originator
come from somebody who seems to be a designer. I now have a collection of
three different jpgs with nice pictures of jewelry and home appliances.
(S)he must also be a RISKS reader, the only link between me and the RISKS
e-mail. The risk is that virus firewalls, by not stripping all attachments
from infected e-mails, not only block viruses from spreading, which is good,
but also help in distributing potentially sensitive documents to third

P.S. I'm still waiting to get a nice Word document with good takeover
information so I can do some serious insider trading

  [Classic case of address spoofing.  There are also lots of
  spoofs allegedly from RISKS.  Don't Believe a Word!  PGN]

Re: The Klez Effect

<"Greg Searle" <>>
Tue, 4 Jun 2002 16:36:39 -0400

Listmasters are really getting hit, but you don't need me to tell you that.
Just today, another list that I subscribe to urged its members to rid
themselves of the virus.

There's good news, however.  According to MessageLabs
(, Klez is finally starting to slowly
recede.  It peaked two weeks ago at about forty thousand viruses intercepted
by the company per day.  Now it's around 27K/day.

So far, I've only had a couple of firewalls send me warnings that I "sent"
the virus.  No live person ignorant of Klez's forged header has complained

Re: Klez and mail loops (Pool, RISKS-22.10)

<"A. Harry Williams" <HARRY@VM.MARIST.EDU>>
Tue, 28 May 2002 09:26:18 EDT

In discussing the Klez virus/worm, Martin Pool makes a common mistake among
UNIX admins, assuming that what is available on their system in TCP/IP
utilities accurately and completely implements RFC specs.  X- headers are
user-defined headers, and therefore cannot be an RFC network standard
header.  I just searched, and can only find Precedence: in RFC 2076, where
it is defined as "non-standard, controversial and discouraged".  The real
problem is that these auto-responders, including many Out of Office
programs, try to use the email Headers, rather than the envelope headers.  A
simple MAIL FROM:<> that was correctly honored would stop many loops, and
that is its purpose.

More on Klez (Garfinkel, RISKS-22.10)

<hal lewis <>>
Mon, 27 May 2002 19:27:19 -0700

 >It makes you wonder if there should be any liability for these individuals.

Who needs to wonder? This is theft, pure and simple.

Please report problems with the web pages to the maintainer