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 23 Issue 66

Friday 14 January 2005


A Comedy of Errors
Leslie Lamport
30,000 personal records stolen in GMU server compromise
James Bauman
New FBI software not usable
Wal-Mart Stung in $1.5 Million Bar-Code Scam
Evan Schuman via Monty Solomon
Attack on T-Mobile
Risks of /pseudo-?/random alphanumeric generation
Joe Thompson
Risks of lenient parsing
Jim Horning
Heisenberg at work? Ranking cardiologists
David Lesher
Ticket not in computer system: your insurance rates may increase
Joyce Scrivner
eBay open invitation to phishing scammers
Thomas L. Jones
Honest, General, it was only a little glitch
Jeremy Epstein
Microsoft AntiSpyware beta - quick review
Rob Slade
Re: Why adding more security measures may make systems less secure
Ron Bean
Re: New Year's Privacy Resolutions
Erling Kristiansen
REVIEW: "Net Crimes and Misdemeanors", J. A. Hitchcock
Rob Slade
Info on RISKS (comp.risks)

A Comedy of Errors

<Leslie Lamport>
Wed, 5 Jan 2005 09:55:36 -0800

The TLA+ specification language uses a common notation in which
strings are enclosed in double-quotes.  Special characters within a
string, such as double-quote and backslash, are quoted by preceding
them with a backslash.  Thus, the string consisting of the five
characters a " b \ c is written "a\"b\\c".  The string "a\"b"
appearing in a specification should traverse the following circuitous
path if it is printed by the model checker:

1. The parser should convert the string "a\"b" to the list
   ('a', '"', 'b') of the three characters a " b.

2. The printing routine should convert ('a', '"', 'b')
   to the string "a\"b".

3. The pretty-printer, which inserts spaces and line breaks to format
   the output, should leave the string "a\"b" unchanged.

In the current version, none of these steps is performed correctly.
Instead, they do the following:

1. The parser converts the string "a\"b" to the list
   ('a', "\", '"', 'b') of the four characters a \ " b.

2. The printing routine, which should covert ('a', "\", '"', 'b')
   to "a\\\"b", instead converts it to "a\"b".

3. The pretty printer, which should leave the string "a\"b"
   unchanged, instead prints it as "a\".

What is remarkable is that these three transformations are performed
by code written by three different people.  The programmers are all
PhD researchers at different institutions--one at a university and the
other two at two different industrial research labs.

Quoted characters are one of only two special cases that occur in
handling strings.  (The other, the empty string, should be handled
automatically by any competently-written code.)  Yet three different
people forgot about that case when writing code.  (I presume that the
first transformation is done in two steps--the first identifying the
string and the second converting it to a list--and the programmer
forgot about quoted characters only in the second step.)

This comedy of errors shows that smart people can overlook even
obvious cases.  One hopes that, had this been production code written
by professional programmers rather than research-project code written
by researchers, then so simple an error would have been caught.  But
programs contain many problems that are not so simple.  This incident
points out the limitation of code reviews.  It also calls into
question the use of multi-version programming as an alternative to
verification for building highly reliable systems.

30,000 personal records stolen in GMU server compromise

<James Bauman <>>
Tue, 11 Jan 2005 14:59:32 -0500

The server at George Mason University in Virginia was compromised by
crackers who stole personal information ("names, photos, Social Security
numbers and (campus ID) numbers of all members of the Mason community who
have identification cards") on 30,000 students, faculty, and staff.

The mega-risk here is obvious — tens of thousands of people who may become
victims of identity theft, one of the fastest growing crimes in America.

A system alert is posted on the University web site -->

For more details, see

New FBI software not usable

<"NewsScan" <>>
Thu, 13 Jan 2005 08:49:59 -0700

A new FBI computer system called Virtual Case File, designed to help agents
share information to ward off terrorist attacks, may have to be discarded
because it doesn't work as designed. The agency will be soliciting proposals
for new software from outside contractors for new software. Sen.  Judd Gregg
(R-N.H.), chairman of the Senate appropriations subcommittee, calls the
development "a stunning reversal of progress" and adds: "If the software has
failed, that sets us back a long way. This has been a fits-and-starts
exercise, and a very expensive one for a very long time.  There are very
serious questions about whether the FBI is able to keep up with the
expanding responsibility and the amount of new dollars that are flowing into
it. We have fully funded it at its requested levels." Science Applications,
the company that developed the system, says it "successfully completed"
delivery of the initial version of the Virtual Case File software last
month.  [*Los Angeles Times* 13 Jan 2005, NewsScan Daily, 13 Jan 2005],1,2171776.story?

  [VCF is part of a four-year, $.5 billion overhaul of the FBI's
  ``antiquated'' computer systems.  PGN]

Wal-Mart Stung in $1.5 Million Bar-Code Scam

<Monty Solomon <>>
Fri, 7 Jan 2005 01:49:49 -0500

By Evan Schuman January 5, 2005

In a scheme that leveraged a little technology but relied on inattentive
cashiers, Tennessee authorities have arrested two couples on charges that
they used bogus bar codes to steal at least $1.5 million from hundreds of
stores--some belonging to Wal-Mart--in 19 states. The group is slated to
appear in court Wednesday.

Although the accused are said to have spent a lot of time and effort
organizing colleagues in various parts of the country, the technology
portion of their scheme was quite simple. They are accused of visiting a
retailer and purchasing a low-priced item. The group would then scan the bar
codes and simply print out duplicate bar codes, said Thomas Dean, the
assistant Sumner County (Tennessee) district attorney who is assigned to the

The accused--Michael Poore, 29, and Julie Marie Simmons, 35, also known as
Julie Poore; and Dewey Howerton, 39, and Laura Howerton, 39--would then go
back to the store, tape the duplicate bar code on a higher-priced item and
purchase the more expensive item at the lower scanned price, Dean said in an interview.

One of the accused, according to the police complaint, would then remove the
bogus tag and try to return the item to the store for the full purchase
price. Instead of cash, the defendants would often ask for gift cards, Dean
said. "Wal-Mart will more quickly put it on a gift card than hand you cash,"
he said.  ...,1759,1748274,00.asp

Attack on T-Mobile

<"NewsScan" <>>
Thu, 13 Jan 2005 08:49:59 -0700

A network vandal broke into the network of wireless carrier T-Mobile over a
seven-month period and read e-mails and personal computer files of hundreds
of customers — including those of the Secret Service agent investigating
the hacker himself. The online activities of the vandal, 21-year-old
computer engineer Nicolas Lee Jacobsen of Santa Ana, were traced to a hotel
where he was staying in Williamsport, N.Y. Although Jacobsen was able to
view the names and Social Security numbers of 400 customers (all of whom
were notified in writing about the break-in), customer credit card numbers
and other financial information never were revealed, and T-Mobile says it
"immediately took steps that prevented any further access to this system."
[AP, 12 Jan 2005; NewsScan Daily, 13 Jan 2005]

Risks of /pseudo-?/random alphanumeric generation

<Joe Thompson <>>
Thu, 6 Jan 2005 13:50:16 -0500

The Cabbage Patch doll, known to legions of kids from the 80s on, comes with
"adoption paperwork" including a unique, randomly-generated serial number.
However, apparently the company's random number generator was, from a
certain point of view, a little too random:

Girl Gets Cabbage Patch Doll With Obscene Message

Apparently, purely by chance — or so the company insists, at any rate --
one serial number included a six-letter string commonly considered obscene,
especially so in the context of children's toys.  The slide show on the
linked new story page has a shot of the string in question (with the first
letter helpfully covered up to avoid further offense).  That it was given as
a Christmas gift seems particularly unfortunate.

Many are the tales of large computer systems where user account names
are fully or partially randomized to avoid embarrassing formations
from pieces of names or other data.  Even in such cases, some basic
filtering is needed; in fact given the audience for the product I'm
surprised no one at Play-Along has realized this in twenty years.
People easily forget that 0000, 1111, 2222, etc. are as random as any
other four-digit string in certain contexts.

Risks of lenient parsing

Fri, 07 Jan 2005 4:26 PM

Yesterday I had a frustrating experience trying to help track down a
problem in a post to a blog I subscribe to,
<>. It ended happily enough
when we were able to locate a syntax error in the HTML of the post, but
not before we had explored several blind alleys.

For the gory details, see

So what is the lesson? There was clearly a syntax error, so what we got is
what we deserved, right? I think not.

Given the frequency of errors in HTML, it would be unreasonable for
renderers to refuse to display pages with errors. (There is an error
somewhere in my blog post on this that I am still looking for.) However, we
stumbled around blindly because *none* of the browsers we were using gave
any hint that there was a syntax error on the page. Each just silently
"corrected" the error. Unfortunately, but predictably, they didn't all
"correct" it in the same way, meaning that Peter and his testers were
consistently getting one result, and I was consistently getting another.

I contend that *all* of the browsers were wrong not to indicate clearly the
presence of a syntax error. A friendly browser would even have made some
attempt to indicate the approximate location on the page of the error.

Although it was published more than thirty years ago, I think my advice on
"What the Compiler Should Tell the User" (in *Compiler Construction, an
Advanced Course*, F. L. Bauer and J. Eickel (eds.), Springer-Verlag,
pp. 525-548, 1974) is still pertinent to those who build compilers and other
formal language interpreters. Those who do not study the past are very
likely not to learn its lessons, and therefore to repeat old mistakes, such
as silently "correcting" syntax errors.

Heisenberg at work? Ranking cardiologists

<David Lesher <>>
Tue, 11 Jan 2005 07:52:25 -0500 (EST)

Cardiologists Say Rankings Sway Choices on Surgery, Marc Santora, 11 Jan 2005

  An overwhelming majority of cardiologists in New York say that, in certain
  instances, they do not operate on patients who might benefit from heart
  surgery, because they are worried about hurting their rankings on
  physician scorecards issued by the state, according to a survey released

I am reminded of the {alleged} quota scheme in the FBI. For years, no agent
would willingly undertake terrorism cases, as such are long, time consuming
and fraught with failure. Stick to the meat & potatoes of doofus
bankrobbers, and you'll get promoted.

So the PHB's made terrorism cases carry more weight... Result?  Now
jaywalking and felony mopery are charged as terrorism...

Ticket not in computer system: your insurance rates may increase

<Joyce Scrivner <>>
Thu, 06 Jan 2005 16:19:07 -0600


Mark Frauenfelder: In November, my wife, Carla Frauenfelder, was driving
home when a Los Angeles Police officer pulled her over. He ticketed her for
making an illegal left turn.

The next day, with her ticket in hand, I entered the url for the website
listed on the ticket ( I wanted to pay the fine and
sign her up for driving school so our car insurance rates wouldn't
increase. The website couldn't find the ticket. I tried searching for it
both by entering the ticket number and by entering my wife's driver license
number. No luck. So I called the phone number on the ticket. The woman who
answered said there was no record of the ticket. She said my wife would
have to drive to the ticket office on Penfield St, in Chatsworth to take
care of it.

So, my wife drove there on January 5th and showed the woman at the counter
the ticket. The woman entered the ticket number and nothing came up. She
scratched her head for a minute, and then noticed that the police officer
forgot to write a date on the ticket. Apparently, that screws everything up.

The woman told my wife what the fine is (about $135), but told her that she
could not accept payment for the fine, because the ticket is not in the
database. My wife is not allowed to attend driving school, either, because
the ticket isn't in the database.

The woman instructed my wife to call the court every day week, to find out
if the ticket had been entered into the computer yet. Once it shows up, she
is supposed to drive to the ticket office the very next day to take care of
it. And once the ticket has been entered, she is going to be hit with a
penalty and possibly a warrant for her arrest, because once the information
goes into the computer it'll see that she hasn't paid the fine yet, and it
will be flagged as delinquent. My wife will then have to explain the
situation to another helpful city employee.

My wife asked the woman how it long will take for the ticket to be entered
into the computer system. The woman said she had no idea. My wife asked her
if she is going to have to call every day week for the next several years.
She shrugged and said "Well, it might take a week, it might take six
months, I don't know."

My wife asked again if she could just pay the fine and have it apply to the
ticket when it finally does show up. Woman: "Nope."

I'm at a loss for what to do here. If you have a good idea please
<>email me and let me know if you do!

eBay open invitation to phishing scammers

<"Thomas L. Jones" <>>
Sun, 2 Jan 2005 05:23:20 -0500

Recently I received an e-mail from involving something called = "My
Messages." The e-mail directs the user to click on a link and enter = his or
her password. Thus it is indistinguishable from a phishing scam, = and will have only itself to blame when its customers fall = victim to

Thomas L. Jones, Ph.D., Computer Science

cc:  Federal Trade Commission <>

Honest, General, it was only a little glitch

<Jeremy Epstein <>>
Thu, 13 Jan 2005 8:18:22 -0500

As has been widely reported, the DoD's missile interceptor test failed
miserably in December, building on a rather impressive history of failures.
According to Pentagon brass, the problem was "with an automated pre-launch
check of the communications flow between the interceptor and the main flight
control computer. Detecting too many missed messages, the system shut down
automatically, as designed.  [so] the Pentagon will increase the pre-launch
tolerance for missed messages. [General] Obering said the tolerance level
was set too low; increasing it will not risk a flight guidance failure".

Well, that makes me feel better.  The system ran into problems, so it
generated errors.  Rather than figuring out what the problem was, let's
ignore the errors.  Not unlike turning up the radio in your car so you can't
hear it falling apart.

The general went on to say "Statistically, it's a very rare occurrence and
most likely would not happen again."

Gee, I feel safer every minute.

Microsoft AntiSpyware beta - quick review

<Rob Slade <>>
Fri, 7 Jan 2005 13:49:15 -0800

The beta version of Microsoft's Anti-Spyware program (purchased from Giant)
is available at

The beta version is about a 6.4 meg download, and can be downloaded as a
file in order to copy and install it onto other machines.  That's very nice,
and a departure from Microsoft's often heavy-handed approach.

Installing offers you a few options: do you want to set it to automatically
update (I did), do you want it to install real-time protection (I did: so
far it hasn't interfered with much, but I haven't used it much, either), and
do you want to join Spynet.  Spynet is not an invitation to join a kind of
corporate CIA, but will report on "suspect" files.  There is a fair amount
of information on what it collects, if you ask for it.  It seems to send
information about file sizes and an MD5 hash back to HQ, but not, seemingly,
the suspect file itself.  In any case, there wasn't enough information on
what it *doesn't* collect for me to feel comfortable, so I turned it off.
(I hope.)

The installation seems to default to a reasonably protected mode: the
defaults are for auto updating, real-time protection, and scheduled scans
(although the schedule is for 2 am).

When you start up the program, it is initially set for a quick scan.  I
changed that to a full scan, which took about half an hour on my machine.

>From a quick test, the MS antispyware, at least in beta, falls between
Spybot S&D and Adaware in terms of detection.  Spybot is fairly
conservative, and only deals with stuff that is pretty certain to be
spy/adware, whereas Adaware will detect a bunch of other stuff.  The MS
product detected one copy of BackWeb (inactive) that Spybot had not, and
detected about 38 copies of 15 versions of other stuff from my samples
directory.  (Adaware quarantined about 60.)  The items detected all seem to
have a least some remote access component, even if it is rather limited
(such as BadTrans.B, that drops a keylogger).  Oddly, it only detected two
of my extensive collection of Bagles.

You can ask the program to deal with individual threats in different ways,
although seemingly not individual files.  (As a researcher, I like that.  In
terms of protection, I'm not as sure.)  The options are to remove,
quarantine, ignore, or always ignore.  The program usually defaults to
quarantine, although some threats are considered more serious, and marked to
remove.  The explanation of "always ignore" is not detailed enough, as far
as I am concerned: does this mean always ignore this particular file, or
always ignore this threat?

You can also specify certain directories to scan, or to ignore.  Again, as a
researcher I really appreciate the ability to tell it so ignore my sample
directory.  Unfortunately, this option doesn't work properly: it scans
directories you tell it to ignore, regardless.  When I told it to scan
*only* my sample directory, it seemed to scan a fair amount of other stuff
as well.  Again, from a protective standpoint, this is probably a good

At the moment, after a very quick test, I'd provisionally recommend the use
of the MS/Giant antispyware program, at least in fairly restricted and
manual mode.  I'd be interested in hearing from others who have tested the
real-time operations more extensively, and particularly from anyone who has
tested the Spynet capabilities, and what information is returned thereby.    or

Re: Why adding more security measures may make systems less secure

< (Ron Bean)>
Wed, 29 Dec 2004 00:17:01 -0500

"Don Norman" <> writes in RISKS-23.63:

> 4: The Dedicated Worker problem.  If the security or safety requirements
> get in the way of doing the work, then the most dedicated workers will
> defeat them.

> Note, the most obvious response of security and safety people is "more
> training is necessary."

There is training, and then there is operant conditioning.  If defeating a
safety and/or security system results in an "attaboy" for increased
productivity, the employee is being "trained" to defeat the system,
regardless of anything that may have been said in the official training

The only way I see around this is to _continually_ emphasize the value of
safety/security systems, and to somehow downplay the all-too-obvious value
of defeating them (possibly by making them less awkward to use-- don't
assume this is impossible).

A lot of safety training focuses on reviews of "one in a million" accidents
that actually happened, so people will remember why the safety systems are
there in the first place. The book "Construction Failure" by Feld & Carper
(Wiley 1997) notes that many construction accidents are caused by
"unexpected" weather conditions, that statistically are not "unexpected" at
all. This may be true of computer security issues as well.

It would also help if the security message were coming from the people
responsible for production rather than from a separate department. In many
companies, the people in charge of safety and/or security are seen purely as
obstacles, because they don't seem to be of any help when it comes to
actually getting things done.

Re: New Year's Privacy Resolutions (Peek, RISKS 23.65)

<Erling Kristiansen <>>
Thu, 06 Jan 2005 21:14:53 +0100

Bernard Peek sets as a goal "Every ad a wanted ad". But he thereby excludes
those people who feel that "Every ad an unwanted ad".  For those of us who
belong to this group, Rotenberg's Resolution seems a better starting point.

Peek advocates the view that the more complete trail of your actions you
leave, the better can advertising be targeted. I would like to ask: What
kind of trail am I supposed to leave in order to get the message across that
I do not want any unsolicited ads at all? The only answer I can think of is:
Leave as little trail as possible. Which is exactly what Rotenberg suggests.

REVIEW: "Net Crimes and Misdemeanors", J. A. Hitchcock

<Rob Slade <>>
Mon, 10 Jan 2005 15:33:56 -0800

BKNTCRMD.RVW   20041016

"Net Crimes and Misdemeanors", J. A. Hitchcock, 2002, 0-910965-57-9,
%A   J. A. Hitchcock
%C   143 Old Marlin Pike, Medford, NJ   08055
%D   2002
%G   0-910965-57-9
%I   Information Today Inc.
%O   U$24.95/C$37.95 609-654-6266
%P   359 p.
%T   "Net Crimes and Misdemeanors"

This book is not about net crimes in general, but about cyberstalking
and online harassment.

Chapter one details Hitchcock's own experience with cyberstalking and
harassment, an extremely unpleasant case of deliberate personal attack
by fraudsters she had exposed.  Three other cases are briefly
described in chapter two, along with some basic advice on header
analysis.  Spam is delineated, and some helpful sites for dealing with
it are listed, in chapter three (which also contains the usual, not
terribly useful, suggestions for keeping your address off the net).
Chapter four lists some urban legends and chain letters, which are
hardly criminal material.

Chapter five lists various types of online scams, but really only
addresses credit card theft.  The utility of the advice varies: the
book suggests that you only deal with vendors with a professional
looking website (hardly a guarantee of virtue), but also gives fairly
detailed descriptions of indicators for a secure HTTP (HyperText
Transfer Protocol) session.  Online auction fraud is covered in
chapter six, from the perspective of both buyer and seller.  The story
of adoption fraud, in chapter seven, is particularly distressing.
Chapter eight give some account of identity theft, but the initial
"case" is more related to harassment, and the material never really
looks at more usual identity theft situations.  More cases of
cyberstalking are listed in chapter nine, with not as much helpful
content.  Chapter ten discusses trolls, flames--and more harassment.

Chapter eleven examines chat and harassment.  Other means of
harassment are discussed in chapter twelve.  Child exploitation is
reviewed in chapter thirteen.  Chapter fourteen looks at various
issues in the workplace.  Statements from various law enforcement
personnel are given in chapter fifteen--along with an odd mention of
the Sam Spade program.  Harassment at universities is covered in
chapter sixteen.  There is a terse mention of the PGP program in
seventeen.  Chapter eighteen describes viruses and firewalls, but not
very well.  Tips on investigating harassment are in chapter nineteen.

The book does provide some helpful resources on certain topics.  It
could have provided more, if it didn't keep returning to the same
topic over and over again.

copyright Robert M. Slade, 2004   BKNTCRMD.RVW   20041016    or

Please report problems with the web pages to the maintainer