ECHELON: Deaf Ears

Date Tue, 1 Aug 2000 12:31:42 -0400

[: hacktivism :]

Deaf Ears 

By William M. Arkin
Special to
Monday, July 31, 2000

You send an Email, fax a friend, talk on your cell phone or participate in
an Internet chat session; you can be sure that the National Security Agency
is listening.

Get a grip. . . .

The Internet might be all Echelon, all the time, but in the real world, the
NSA can barely track all the foreign leaders, diplomats, governments,
military units, terrorists, international drug dealers and criminals it is
directed to snoop on. Add to that the explosion of fiber optics and wireless
devices, which presents eavesdroppers with new challenges of interception at
the very times when there is a quantum increase in the overall volume of
communications. Throw in the increasingly difficult task of "breaking" ever
more sophisticated codes, and you might ask yourself the question as to why
any sane person would conclude that Big Brother has the resources, let alone
the interest, to keep tabs on you or me.

The Know Nothings
A ha, I can hear the paranoids responding, that's where Echelon comes in.
This five-government "global surveillance system" is the very computer
system that allows huge volumes to be sifted without human intervention.

Echelon, controlled by the NSA at Fort Meade, MD., in cooperation with
Britain, Australia, Canada and New Zealand, allows governments to specify
the subjects they are interested in, and Echelon automatically routes
material - faxes, Email traffic, telephone calls - from the big ear to
prying eyes.

Is Echelon, as the American Civil Liberties Union says, "the most powerful
intelligence gathering organization in the world"? Or is it, as the equally
liberal National Security Archive says, "a more limited program"?

I tend to fall in the archive's camp. One of the reasons the archive thinks
Echelon's assault on privacy is overblown is not just the web of oversight
that exists in the post-Vietnam, post-Watergate era. They and other sane
skeptics of Echelon-mania point to technological barriers that stand in the
way of an all-hearing, all-seeing network.

Sifting Through
While I agree that Echelon is only symbolic of American imperialism and
privacy denied, the latest anti-McDonald's rant from Europeans who should be
asking questions about their own governments, the Internet and portable
device explosion has propelled the intelligence listeners to obtain new
hardware and software to satisfy even the worst nightmares.

Enter the Multi-Intelligence and Information Exploitation (MIITE) research
program of the Air Force Research Laboratory in Rome, N.Y. The laboratory
describes the $25 million three-year effort as focusing on "identifying and
developing technology to support global awareness, data fusion, dynamic
planning, data warehouse functions, and force execution."

In English? Synetics Corp., a subsidiary of BAE Systems and the prime
contractor, coordinates 45 teammates ranging from the largest defense
contractors (Lockheed Martin, Computer Sciences Corp.) to niche intelligence
companies such as Delfin and QuesTech, all of whom specialize in signal
processing and data management to develop some of the very capabilities to
do what Echelon opponents believe is already being done.

According to Air Force documents associated with the May 1999 contract,
MIITE brings together the efforts of 79 disparate information technology
research projects, many of which have shown promise in dealing with huge
amounts of information to create predictive intelligence on the battlefield.
One goal is to solve the bottlenecks associated with intercepted material.

You're Hearing Voices
Projects 32 through 39 of MIITE are developing new automated audio
exploitation techniques to improve the speed and breadth of intelligence
collection. Intelligence agencies have identified a "serious shortfall in
ability to identify and handle large volumes of communication messages," the
Air Force documents say.

Project 32 is developing the capability to identify languages and dialects
"independent of speaker and what is being said." In the first phase of
research, language databases were developed and tested against thousands of
transmissions. The result? Eighty-one percent recognition of five languages
by listening to less than a half a second of speech. Peruvian and Cuban
dialects were distinguished in Spanish language traffic with 80 percent

Project 33 has demonstrated the ability to identify and sort 41 separate
male speakers with 86 percent accuracy within just two seconds of hearing
each speaker. Project 34 has demonstrated identical accuracy in recognizing
60 speakers after training on about six seconds of speech.

The goal? Message sorting by speaker, aircraft identification by pilot's
voice, automatic language to language and then voice to text transcription.
Project 38 is even working on recognizing cellular phone and frequency
hopping data with only one second or less of transmission so digital signals
in the air that are sliced up into tiny packets for transmission to cell
towers can be reconstructed both accurately and quickly without having to
intercept the transmission at its source.

MIITE is a rare opportunity to get a look at the state of art in the
business of collecting and sifting huge amounts of difficult-to-process
information. The technologies being researched should give some indication
of both the enormous difficulties the intelligence monitors face in a world
where their targets are so varied and the technologies so uncooperative.

MIITE hardly scratches the surface of NSA's direct research efforts, to be
sure. But the notion that the government can simply listen in on any
communication in any mode and that someone in the government has the time or
would actually be interested in analyzing stray traffic when extracting the
target material is so difficult should demonstrate how utterly and totally
off-base Echelon paranoia is.

[: hacktivism :]
[: for unsubscribe instructions or list info consult the list FAQ :]
[: :]