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 45

Saturday 10 July 2004


$500 million and counting
Tom Gray
Keyless remotes to cars suddenly useless
Paul Saffo
Stolen: one-third of the world's software
Obstacles to Net phone service
Zinc whiskers
Craig S. Bell
Friends don't let friends use Microsoft Internet Explorer
Tom Van Vleck
Bev Harris crusades to expose e-voting flaws
Fredric L. Rice
E-voting concerns
Perils of Database Matching, Chapter 47,061
Paul Wallich
Private-sector firm maintains dossiers in U.S.
David Marston
Re: Web ads threat to bank security
Rich Kulawiec
E-mail non-privacy is a good decision!
Craig DeForest
VoIP hacks gut Caller I.D.
Monty Solomon
Using google against google
Peter Parker
Re: Coca-Cola Cans as Security Threat
Nick Brown
REVIEW: "Network Security Jumpstart", Matthew Strebe
Rob Slade
Info on RISKS (comp.risks)

$500 million and counting (from Dave Farber's IP list)

<Tom Gray <>>
July 10, 2004 6:00:25 AM EDT

It is not mentioned directly, but this $500 million dollar computer system
owned by the Ontario government is incapable of adjusting the social
assistance rates.  Recipients will be receiving two lump sum payments, one
this summer and one in the fall, to make up the 3% increase that the
government has decided on.

How can a computer system cost $500,000,000? Just how can training by itself
consume hundreds of millions of dollars? Why if a program cost $500,000,000
to produce does it not work?

  [I guess you have not followed the IRS and FAA modernization fiascos,
  each over $4 Billion, discussed in RISKS...  The more money spent,
  the less likely it seems success will follow.  PGN]

How costly computer sparked a `nightmare':
Social services system `inflexible from Day 1,' expert says;
Government estimates fixing flaws could top $10 million
Richard Brennan and Robert Benzie, Queen's Park Bureau

It seemed like a good idea at the time.  An ideologically driven,
cash-strapped Conservative administration wanted to reduce social
assistance costs and increase the role of the private sector in
government.  In Jan 1997 then-premier Mike Harris contracted with
Andersen Consulting to revamp the Ministry of Community and Social
Services' outdated computer system.  But a two-year independent study
of the $500 million computer system has concluded that it has been
seriously flawed from the very beginning and virtually incapable of
making timely changes.  That became clear when it was learned the
system, responsible for distributing welfare and disability benefits
to 670,000 Ontarians, is unable to calculate a 3 per cent increase,
the first rise in 11 years.  It's going to cost at least $10 million
to fix the problem -- $3 million to correct the computer system and an
additional $7 million to test it.

Keyless remotes to cars suddenly useless

<Paul Saffo <>>
Mon, 5 Jul 2004 11:55:26 -0700

Increasingly we hear sagas of entire shopping center parking lots (and less
widespread cases) in which keyless remote entry devices for automobiles are
inoperative.  For those of you with such systems, an article by Joshua
Partlow with the above caption may be of particular interest.  Waldorf (15
miles from Andrews Air Force Base in Charles County MD), Bremerton
Washington, Las Vegas, etc.  (The Florida Panhandle garage-door openers
jammed from Eglin Air Force is also mentioned, noted previously in RISKS.)
[PGN; county/state correction in archive; mea culpa]

"Keyless entry remotes have become standard in new cars in recent
years.  Of the more than 14 million cars and light trucks produced in
the United States last year, 77 percent came with the remotes, up from
32 percent in 1996."  "... unlike other more powerful radio signals,
keyless entry remotes are not licensed by the Federal Communications
Commission. They are allowed to operate on frequencies used by
licensed customers as long as their signals are sufficiently weak and
don't interfere with others. But because of this outlaw status, their
own signals can be jeopardized."

[Source: Joshua Partlow, *The Washington Post*, 5 Jul 2004; PGN-ed]

Stolen: one-third of the world's software

<"NewsScan" <>>
Wed, 07 Jul 2004 08:26:19 -0700

The Business Software Alliance, a trade group, says that 36% of all
the software in the world has been pirated, costing the industry $29
billion in lost revenue. The five countries with the highest incidence
of pirated software are: China (92%), Vietnam (92%) and Indonesia
(88%), Ukraine (91%), and Russia (87%). (AP/*San Jose Mercury News*, 7
Jul 2004; NewsScan Daily, 7 Jul 2004]

Obstacles to Net phone service

<"NewsScan" <>>
Tue, 06 Jul 2004 08:29:13 -0700

AT&T says it expects to have 1 million customers for its
voice-over-Internet-protocol (VoIP) phone service by the end of next year,
and cable-TV company Comcast expects to offer VoIP all its customers by the
end of 2006; however, Mark Main of the British consulting firm Ovum warns
that -- although everyone will be using VoIP 10 or 15 years from now -- the
road to that point "will be quite varied, quite torturous [tortuous?] and
not at all clean." Some obstacles in the way: only 27% of U.S. online users
have even heard of it; a VoIP subscriber needs a broadband connection, and
phone service will be only as good as that broadband connection; prices may
go up in the future due to increased regulation and taxes; and VoIP service,
which depends on the regular power grid, will fail if grid should fail.
[AP/*San Jose Mercury News*, 6 Jul 2004; NewsScan Daily, 6 Jul 2004]

Zinc whiskers

<"Craig S. Bell" <>>
Thu, 08 Jul 2004 15:37:17 -0700 (PDT)

The zinc coating on the underside of datacenter floors can emit tiny
metallic whiskers, which could lead to abnormally high rates of
electronic equipment failure:

The State of Colorado recently suffered a non-trivial outage,
apparently due to zinc whiskers:,1413,36%257E33%257E2245069,00.html

Friends don't let friends use Microsoft Internet Explorer

<Tom Van Vleck <>>
Tue, 6 Jul 2004 08:41:29 -0400

Describes an attack on IE where a file named img1big.gif
installs and runs an IE Browser Helper Object that
steals information before SSL transmission and sends
copies to

Visit the wrong website and your IE is invisibly bugged.

Bev Harris crusades to expose e-voting flaws

<"Fredric L. Rice" <>>
Tue, 06 Jul 2004 22:20:59 -0700

  Ambushing registrars and tracking down executives at their homes
  and offices, Harris, 52, has uncovered conflicts of interests and
  security laws inside the companies that make electronic ballot
  machines.  Searching the Web and poring over newspaper clippings,
  Harris has unearthed obscure arrest records, ties to conservative
  political groups and other embarrassing secrets of senior executives
  at voting companies.

Her conclusion: there will be so many problems with the more than
100,000 paperless voting terminals to be used in the November
presidential election that the fiasco will dwarf Florida's hanging
chad debacle of 2000.

E-voting concerns

<"NewsScan" <>>
Thu, 08 Jul 2004 07:57:24 -0700

California's Secretary of State has won a victory in federal court and
new agreements from counties with touch-screen machines to make extra
security arrangements. U.S. District Judge Florence-Marie Cooper
denied requests by disability rights activists and four California
counties to overturn the Secretary's conditional April 30 ban on touch
screens for the November election. In the suit, disability groups
argued that banning electronic voting will deny hundreds of thousands
of people the right to vote in private, but the judge ruled the
Americans With Disabilities Act requires only that disabled voters be
given the opportunity to vote.  [Bloomberg News/*San Jose Mercury
News* 7 Jul 2004; NewsScan Daily, 8 Jul 2004]

Perils of Database Matching, Chapter 47,061

<Paul Wallich <>>
Sat, 10 Jul 2004 09:48:37 -0400

(if a bunch of people haven't sent this already)

The New York Times reports this morning
that the reason for the extraordinary paucity of hispanic voters on the
latest Florida felon-purge list was the lack of a "hispanic" category in
the felon database, so that race-matching against voters who identified
themselves as hispanic would automatically fail. (This on top of not
having bothered to check whether any of the felons had has their rights

"The method uses race as one of several factors in determining whether a
felon has registered to vote. If a voter's first name, last name and
date of birth are the same as those of a convicted felon but the race is
different, the name is not put on the list for potential purging.

But the database of felons has only five variables for race: white,
black, Asian, Indian and unknown. And a voter registered as Hispanic
whose name and birth date matched a felon's would be left off the purge
list unless his race was listed as unknown.  ...
The paucity of Hispanic voters on the felon list was first reported
Wednesday, by The Sarasota Herald-Tribune, but officials said then that
the problem was not systematic. After The New York Times examined the
data, state officials acknowledged that the method for matching lists of
felons to those of voters automatically exempted all felons who
identified themselves as Hispanic.

...The exclusion of Hispanics from the purge list explains some of the
wide discrepancies in party affiliation of voters on the felon list,
which bears the names of 28,025 Democrats and just 9,521 Republicans,
with most of the rest unaffiliated."

Pretty much anyone who has ever tried to match items in one database
against those in another knows that your have to get the record
formats and categories right for the results to mean anything. In this
case, even the simplest of properly-prepared test data sets would have
uncovered the screwup. (And now I'm trying to decide whether it's
scarier to think that there was malice involved or that a bunch of
ostensible paid professionals with more than two years two work on the
problem could hand over a list like this with straight faces.)

PS. And these are some of the same folks who are supposedly evaluating
the quality of paperless voting machines?

Private-sector firm maintains dossiers in U.S.

<"David Marston" <>>
4 Jul 2004 22:08:58 -0400

The 28 Jun 2004 issue of Mass. High Tech ("The Journal of New England
Technology") has a page 1 story about LocatePlus of Beverly, MA. The news
angle is that a unit of the Massachusetts State Police will upgrade their
use of LocatePlus from CD-ROMs of vehicle records to wireless access (via
Blackberry) to all LocatePlus data.

Checking reveals that they offer some data (public records)
to the public and more to licensed private investigators and the like, but
law enforcement customers get everything, including non-public information.
The website does not say (where I could see it, at least) how to opt out or
correct erroneous information about oneself.

The MHT article says that LocatePlus claims to have information on 98% of
the U.S. population (note: not just adults), including residence data,
"court filings" and "restricted government data." Banks can "verify"
Social Security numbers against whatever data LocatePlus has, and can use
the service to screen for illegal money laundering or funneling. There
are nearly 16000 LocatePlus customers in all.

The RISKS are many. Wireless communication can be intercepted, and what
happens if a cop's Blackberry falls into the wrong hands? Data aggregation
depends on correctly matching the person across databases. If a private
company has "restricted government data", how did they get it, and are
they obligated to protect it and cleanse it as zealously as the agency
from which it came? Are the customers obligated to provide their data to
LocatePlus to help it amass information? As we know from credit bureaus,
the standard for data accuracy is set by the satisfaction level of the
paying customers rather than the data subjects; in the case of credit data,
those paying customers seem to tolerate a 20% error rate year after year.
They believe someone is a law enforcement agency on the strength of a
printed letterhead. Add on the standard set of security concerns.

Re: Web ads threat to bank security (From Dave Farber's IP list)

<Rich Kulawiec <>>
July 7, 2004 8:06:10 AM EDT

> Security experts said updating virus software was the best protection.

But it's not.  Oh, not that it isn't a good idea for people who are
running operating systems which are susceptible to viruses, but like
most security problems, a multi-layered approach is more likely to work,
especially if one of the layers fails.

For example, in this case:

- use a robust browser, such as Mozilla: never use IE
- use a robust mail client, such as Thunderbird: never use Outlook
- subscribe to the -announce list for that browser and
  get in the habit of downloading new versions whenever
  there's a major new release OR a significant security fix
- disable pop-ups
- restrict use of cookies
- run a web proxy (I use Privoxy, that adds
  another layer of defenses
- run anti-adware software (because it will catch things that
  AV software won't)
- if you insist on using a web browser to read your mail, then
  disable Javascript in it
- (better) turn off HTML interpretation in mail
- and so on.

Aside: it's fascinating how many of these articles say "email virus"
when the proper term is "Microsoft Outlook virus" and "web threat" when
the proper term is "Microsoft Internet Explorer security bug".

E-mail non-privacy is a good decision!

< (Craig DeForest)>
Mon, 5 Jul 2004 11:39:32 -0600

In the 3 Jul 2004 issue of RISKS, there were several articles about
the recent court decision that e-mail is not private.  Surprisingly, I
mus take a contrarian position: this is a good decision.  One cannot
legislate privacy; one can only give most users a false sense of

Consider, for example, the results of another legislated-privacy
"solution": analog cellphones use normal, unencrypted broadcast radio
to transmit conversations, so it is now illegal (in the U.S.A.) to
build or buy a radio scanner that receives on those frequencies.  As a
result, a black market developed in radio scanners that were capable
of receiving that frequency range.  Meanwhile, people with analog
cellphones continued for many years to hold private conversations on
the airwaves, with occasionally hilarious results as scofflaws
listened in anyway.  The problem of broken privacy would be much less,
if the cellphone companies had instead been forced to educate their
users that they were essentially bellowing their conversation from a
rooftop.  (Of course, most people now use some sort of encryption as a
side-effect of using digital cellphones.)

Long-time readers may recall that, as a younger and angrier man, I
have ranted about similar discrepancies between the law and reality in
the satellite-communications market: it is illegal (again, in the
U.S.A.) to build a radio receiver that can receive satellite
television.  This despite the fact that (for many years) the
television signals were being beamed directly into your backyard,
unencrypted.  If the proprietors don't want you to receive their radio
signals, they should not be bombarding you with them in the first
place -- or should encrypt them.

It's best to call a spade a spade: non-secure communications are
non-secure, and people who use them should be aware of what they are
getting.  After all, truly private solutions (PGP, GPG and the like)
do exist, and folks who expect privacy should use them.

VoIP hacks gut Caller I.D.

<Monty Solomon <>>
Wed, 7 Jul 2004 13:27:35 -0400

Implementation quirks in Voice over IP are making it easy for hackers
to spoof Caller I.D., and to unmask blocked numbers.

By Kevin Poulsen, SecurityFocus Jul 6 2004 1:54PM

Caller I.D. isn't what it used to be.

Hackers have discovered that the handy feature that tells you who's
calling before you answer the phone is easily manipulated through
weaknesses in Voice over IP (VoIP) programs and networks. They can
make their phone calls appear to be from any number they want, and
even pierce the veil of Caller I.D. blocking to unmask an anonymous
phoner's unlisted number.

At root, the issue is one of what happens to a nugget of
authentication data when it leaves the tightly-regulated realm of
traditional telephony, and passes into the unregulated domain of the

On the old-fashioned phone network, Caller I.D. works this way: your
local phone company or cell phone carrier sends your "Calling Party
Number" (CPN) with every call, like a return address on an envelope.
Transmitted along with your CPN is a privacy flag that tells the
telephone switch at the receiving end of the call whether or not to
share your number with the recipient: if you have blocking on your
line, the phone company you're dialing into knows your number, but
won't share it with the person you're calling.

This arrangement relies on telephone equipment at both ends of the
call being trusted: the phone switch providing you with dial tone
promises not to lie about your number to other switches, and the
switch on the receiving end promises not to reveal your number if
you've asked that it be blocked. In the U.S. that trust is backed by
FCC regulations that dictate precisely how telephone carriers handle
CPNs, Caller I.D. and blocking. Most subscribers have come to take
Caller I.D. for granted, and some financial institutions even use
Caller I.D. to authenticate customers over the phone.  ...

Using google against google.

<"Peter Parker" <>>
Fri, 09 Jul 2004 03:05:30 -0700

Good news for the spammers!!

As most of us are aware that Google provides various options/operators
for writing effective queries. One of the operator is the "site:"
option, which restricts the search to the website specified with this
tag. Just tried googling for some gmail accounts with and the results were a list of urls with the
title "Link Already Used". The area of concern is that all these pages
are actually error pages with a valid gmail user accounts.... so with
a small script its very easy for some one to glean a list of _valid_
gmail accounts.

Do you have a gmail account? ....check if your name is already harvested

Re: Coca-Cola Cans as Security Threat (Dominey, RISKS-23.44)

<BROWN Nick <>>
Mon, 5 Jul 2004 14:05:41 +0200

The immediate RISKs to US national security are minimal - apart from
anything else, the GPS signal won't work inside most office buildings - but
taking it on trust that Coke has issued exactly 120 of these cans seems
optimistic, and assuming that more than 100 or so will ever be found, even
more so.

One would hope that either this promotion, or similar ones in the future,
will have some sort of self-limiting features to the technology - for
example, limiting the time validity of the SIM cards.

REVIEW: "Network Security Jumpstart", Matthew Strebe

<Rob Slade <>>
Tue, 6 Jul 2004 09:25:12 -0800

BKNTSCJS.RVW   20030604

"Network Security Jumpstart", Matthew Strebe, 2002, 0-7821-4120-X,
%A   Matthew Strebe
%C   1151 Marina Village Parkway, Alameda, CA   94501
%D   2002
%G   0-7821-4120-X
%I   Sybex Computer Books
%O   U$24.99/C$39.95/UK#18.99 800-227-2346
%P   365 p.
%T   "Network Security Jumpstart"

The introduction states that this book is suitable for anyone from the
home user to the network administrator to the CEO.  Which is a pretty
tall order.

Chapter one has a decent overview of why computers aren't secure, a
scant computer security history, a few security concepts, and a fairly
trivial set of "review" questions.  There is a media level exposition
on "hackers," in chapter two, a rough outline of intrusion procedures,
and a list of specific attacks that I'm not sure the author fully
understands.  (Immediately following "Denial of Service" comes a
separate entry for "Floods": flooding being a type of denial of
service.)  There is a terse introduction to cryptography, and not much
more than chapter one gave us about authentication, in chapter three.
The suggestions for policy creation, in chapter four, aren't bad for
simple cases, but seriously understate the difficulty of establishing
a full policy, even for home users.  Chapter five describes firewalls
(and seven tells a little bit more about using them at home).  Chapter
six makes the common mistake of assuming that all VPNs (Virtual
Private Networks) are about confidentiality: some are merely about
managing communications configurations.

There is some correct and useful information about viruses in chapter
eight, but it is unfortunately mixed in with a lot of garbage.
Windows NT and its subsequent versions are *not* immune to viruses,
although a rigorous set of file permissions can reduce your risk of
file infectors (which are no longer a major category anyway).
Signature scanners are *not* the only type of antiviral software.
Viruses were *not* invented by accident, BRAIN *never* had an onscreen
display and didn't infect program files, and neither Stoned nor
Jerusalem (Friday the 13th is one variant) were based on BRAIN.
Neither Stoned nor BRAIN relied on program sharing to propagate: data
disks were quite sufficient.  Viruses that only replicate are *not*
benign (anybody ever have problems with Stoned?  Melissa?
Loveletter?), *will* be discovered, and scanning signatures *are*

Fault tolerance, in chapter nine, is not quite business continuity
planning (BCP), but does go beyond the usual UPS (Uninterruptible
Power Supply) and backup recommendations.  Although chapter ten lists
a number of security mechanisms in Windows, a practical understanding
of their use is not presented.  The UNIX tools in eleven are described
more usefully--but they only relate to file permissions.  The network
security tools for UNIX are in twelve--but are only enumerated.
Chapter thirteen has good suggestions for Web server security--but
doesn't say how to implement them.  A random collection of e-mail
security tools and threats makes up chapter fourteen.  IDS (Intrusion
Detection System) concepts are not explained very well in chapter
fifteen: Strebe apparently doesn't understand that all forms use audit
data of one type or another, and doesn't list the major distinctions
between either the engine type or sensor location.

Even given all the faults, one has to admit that Strebe has not done a
bad job with his ambitious intent.  Certainly home users and CEOs can
find better explanations here than in many of the other works aimed at
them, however much I might wish that the book as a whole was more
accurate.  And, yes, even the network administrators might find some
helpful points in the more conceptual material at the beginning of the
book: most of them could do with a better understanding of the need
for policy.  This work isn't great, by any means, but it can fulfill a
need for a quick guide to network threats, for a variety of audiences.

copyright Robert M. Slade, 2004   BKNTSCJS.RVW   20030604    or

Please report problems with the web pages to the maintainer