AN ARMY of automated detectives is getting ready to track down pirated books,
pictures and music on the Net. Unlike some existing schemes, they won鈥檛 need
digital 鈥渨atermarks鈥� to find pirated files. In music files, for example, the
marks can change the music.
Every year, music, text and image files that are worth an estimated 拢10
billion to their copyright owners are illegally copied over the Internet,
according to analysts in the US. To disguise this digital booty, pirates give
files false names or even split and spread them among many different
servers.
Attempts to reach agreement on the best anti-piracy technology have so far
failed and no one has managed to come up with digital watermarks that are not
only invisible to legitimate users, but are also indelible.
Advertisement
Now Netsertion, a start-up in Yarnton near Oxford, believes it has found the
answer. Instead of labelling files with a watermark, Netsertion has worked out a
way to analyse the actual contents of the file and from this generate a unique
ID code or fingerprint that can be registered to the copyright holder. Pirated
versions of a book or a song will generate the same fingerprint as the original
even if the file has been modified or renamed.
The fingerprint is created by analysing the contents of the original file and
extracting features that make it unique. It uses the same sort of
characteristics that people use to recognise a piece of music or a picture. For
music files, they might be the pitch, loudness and harmonic structure. For
pictures, colour balance and the juxtaposition of different objects capture the
essence of an image. In each case, the combination stays recognisable in pirated
versions.
Once the characteristics of the original file have been sampled, captured and
compressed, the resulting digital fingerprint is stored on a database. Programs
known as Web spiders then continually sample the Web, following links from one
page to another and taking the fingerprints of any files they encounter as they
crawl around. If a fingerprint matches one held on Netsertion鈥檚 database, it
indicates that the file has similar content, and so may be breaching the owner鈥檚
copyright. It鈥檚 then up to the copyright holder to decide what action to
take.
Digital piracy has become one of the most troubling aspects of the growth of
the Web. Earlier this year, Napster, the music file-sharing site used by an
estimated 50 million people, was forced by the recording industry to end its
free download service.
Netsertion has begun trials of its digital fingerprinting method with the
music publisher V2, part of the Virgin group, and with the publisher
HarperCollins. But Hugh Smith, a director of Netsertion, says that the
technology is not aimed solely at multinationals: 鈥淲e plan to offer digital
fingerprinting to anyone.鈥�
The first trials have tracked down thousands of pirated files, and Netsertion
says it is already seeing signs of copyright pirates trying to evade its
fingerprinting spiders. 鈥淭he situation has advanced to a game of hide and seek
between the automated detectives and pirates,鈥� says Netsertion鈥檚 Paul Creaser.