>Very nice piece, Charles. Three full days of pho very eloquently
>summarized. ; )
>The Day After: Exactly How Will Napster Remove Copyright-Protected Songs?
>Charles C. Mann
>Here's a quick quiz: Which of the following is the correct spelling of the
>1962 nonsense surf classic by the Rivingtons? Is it a) ''Pa Pa Ooh Mow
>Mow,'' as one authoritative record collectors' Web site has it; b) ''Papa
>Oom Mow Mow,'' as claimed by another; or c) the socialistically correct
>''Poppa Ooh Mao Mao,'' as avowed by a third?
>Until Monday's less-than-swift-but-no-less-than-stunning appeals court
>decision against the file-sharing service Napster -- in which the court
>mostly backed a district court judge's decision to shut down the site while
>the case is brought to trial -- the correct response would have been d) what
>are you, a rock critic? But, given the court's ruling that Napster must now
>find a way to identify and block copyright-infringing song files, the proper
>spelling of song titles will likely be more than a fanboy's parlor trick: it
>could be a key in determining the future of Napster.
>Judge Marilyn Hall Patel, who fashioned the original injunction against
>Napster that the appellate court has now deemed ''overbroad,'' has probably
>already begun to take up where she left off in June. (For an account of the
>legal strategizing in the immediate aftermath of the decision, click here.)
>But lacking any specific guidance from the three-judge panel on how to flush
>infringing songs from the Napster system, Judge Patel faces a number of
>potential landmines on the road to Injunction II.
>Among the difficult questions that Judge Patel must untangle: Since song
>files are frequently misnamed, as the court recognized, how can Napster be
>expected to know precisely what is on its users' hard drives? And even if a
>system were found to purge out copyright-protected songs, how many
>infringing copies -- ''leakage,'' in copyright-lawyer parlance -- would be
>allowed to float around the service without putting Napster in legal
>''It will all depend on how Judge Patel re-crafts the injunction,'' says
>Professor James Boyle, an intellectual-property lawyer at Duke University
>Law School and the author of Shamans, Software, and Spleens: Law and the
>Construction of the Information Society. ''In theory, it could swing either
>way, which would have a huge effect on what Napster actually has to do to
>In simple terms, the appellate court told Napster only that it must obey a
>two-step procedure: The copyright holder -- the record labels, in this case
>-- must tell Napster which of their songs are being traded and then Napster
>must make every effort to block its users' access to those songs.
>The decision itself speaks only of parsing file names -- looking for MP3s
>with names like ''Shaggy-It-Wasn't-Me.mp3'' -- a method that poses problems
>for both Napster and the RIAA. But the labels are closely evaluating other
>screening methods, especially a technique known as ''audio fingerprinting,''
>and may demand that file-swapping services employ this method as well.
>HOW MUCH TO RELY ON A 'BLUNT TOOL'
>Even so, the RIAA is already gearing up for the cruder, file-name-based
>screening system. According to RIAA attorney Cary Sherman the organization
>has already prepared a database of 2.5 million label-owned songs in
>anticipation of having Napster continually sweep its directory for these
>cuts, booting off each and every user who has even one of them. Or,
>depending on the details of Patel's amended injunction, Napster might only
>be obliged to prevent particular titles from appearing on its directory, and
>for a limited period of time.
>As both sides concede, neither tack would totally eliminate illicit
>file-swapping on Napster. But there is wide disagreement about whether any
>screening based on file name could, in the end, do the job with reasonable
>accuracy. Napster interim CEO Hank Barry said on Monday that given the wide
>inconsistency in file names, keeping all offending tracks off the central
>directory by checking their titles would require ''technology that we don't
>believe exists at this time in any manner technologically feasible.''
>Gnutella developer Gene Kan argues that ''One could conceivably filter out
>files with specific keywords in the names. But files are inconsistently
>named and keyword filtering would probably in the end be too blunt a tool to
>use.'' Moreover, Kan says, such efforts will just encourage ''users to
>switch to something which is more difficult for the recording industry to
>manage'' -- like Gnutella.
>Dan Farmer, an independent security researcher who has served as a
>consultant to the RIAA, says, somewhat more hopefully, that screening files
>by their names would certainly ''hamper unrestricted copying, especially at
>the beginning.'' But, he quickly adds, ''it won't stop it completely.''
>Farmer explains, ''If people still want to exchange music that's on the
>blacklist, they'll find a way around it -- putting the titles in Pig Latin,
>for example. They'll be harder to search for, but they still will be on the
>system, and people will still be trading them, although at a lesser level.''
>''I've been reading the discussion of the Napster decision on Slashdot,''
>says Gene Hoffman, CEO of eMusic.com, which has the exclusive online license
>to thousands of songs, mostly from small- and mid-sized artists like Elvis
>Costello and They Might Be Giants. ''They brag about all the ways they say
>they can get around file-name searches. But as long as the core search for
>Elvis Costello turns up nothing, we don't care. If the 'warez d00dz' want to
>spell it 'elvs cstllo,' nobody will find it.''
>No matter what the ultimate effectiveness of such takedown orders, pushing
>ahead with them has pitfalls for both the labels and Napster. The obligation
>to notify, according to Lawrence Lessig of Stanford Law School, is ''exactly
>what the RIAA didn't want,'' because making the labels initiate the hunt for
>infringing Napsterians will set them up as the ''bad guys.'' Equally
>troubling for the RIAA, he says, is that the appellate decision seemingly
>awards the labels the perpetual duty to monitor and notify Napster of
>infringement -- a ''war of attrition'' that will keep this issue alive in
>the public mind for months and years to come.
>MAKING 'I DISAPPEAR' DISAPPEAR
>For Napster, the pitfalls are both more obvious and more complicated. On the
>simplest level, the service risks losing the content that attracts a large
>part of its user base. Perhaps equally importantly, zealous copyright
>policing may end up blocking wholly legitimate users -- a potential PR
>nightmare. Imagine searching by song title for Metallica's ''I Disappear,''
>the Mission Impossible:2 theme whose appearance on Napster last spring led
>to the band's widely publicized lawsuit against the file-sharing company.
>Such a search would find a) copies of the CD cut, which Metallica doesn't
>want on the service; b) copies of live performances of ''I Disappear,''
>which the band has encouraged fans to trade on a nonprofit basis, as occurs
>on Napster; c) a small number of homemade mixes -- blends of ''I Disappear''
>and tunes by, say, Britney Spears, some of which are parodies protected by
>the First Amendment; and finally d) a small group of mislabeled files that
>have nothing to do with Metallica at all. In addition, the search will not
>find versions of ''I Disappear'' that are e) spelled incorrectly -- a recent
>search for the track under alternate names produced several files called
>To both comply with the ruling and avoid antagonizing its users, the company
>will want to warn users away from song type ''a,'' to minimize the
>inconvenience to patrons of ''b,'' ''c'' and ''d'' -- and avoid the wrath of
>the record labels for not ticketing users of version ''e.'' But this will
>not be possible in every case. ''There is going to be some legitimate
>content that is removed if you just (search) by file name,'' asks Travis
>Kalanick, erstwhile VP of strategy for Scour, a now- defunct Napster
>competitor. ''How many of the legitimate users are you willing to lean on to
>get the infringers? If you sweep with too broad a brush, you drive away your
>To avoid the multiple problems associated with screening for file names,
>Napster and the recording industry have several options, says Kalanick. When
>each Napster user contacts the company's central servers, the user's
>computer sends out the file names of the MP3s that it will make available
>for swapping. It also sends out each music file's ''MD5 hash'' -- a sort of
>digital fingerprint that uniquely identifies it. In simple terms, an MD5
>hash is an algorithm that generates a string of 32 letters and numbers that
>is unique to each file run through the algorithm. Thus if any two MP3 copies
>of ''I Disappear'' have the same MD5 hashes, they are copied from the same
>source. Generally, any copies made of a particular digital track, using the
>same ripping software with the same settings, will create files with the
>identical MD5 hash. Thus, Kalanick says, the RIAA could identify particular
>MD5 hashes and instruct the Napster servers not to include files with those
>hashes in the company's index.
>FINGERPRINTING COULD BE THE KEY
>This measure would be no panacea. It would block selected files with 100
>percent accuracy -- without having to kick off any users at all. But it
>would not remove all infringing files, and clever programmers could figure
>out ways to get around it. Moreover, it would require continual monitoring
>for new MD5 hashes. For its part, Napster has argued that it does not
>believe such measures, even if they were acceptable to the RIAA, would
>satisfy its legal obligations.
>A second method would involve identifying songs either through unique codes
>called watermarks or by assaying their characteristics in a so-called
>''digital fingerprint.'' Again, Napster would simply search for files with
>blocked watermarks or digital fingerprints and not allow them to appear on
>its central directory.
>Watermarks are notoriously vulnerable to technical attack, according to
>Bruce Schneier, head of the Net security firm Counterpane. But digital
>fingerprints, a newer technology that is now exciting the attention of the
>labels, may not be so easily hacked.
>One leading music-fingerprinter, Cantametrix, runs music files through
>software that measures the mathematical properties of their sound waves.
>''The song identifies itself,'' says Cantametrix marketing VP Ridge Nye.
>''You could distort the music to make it have a different fingerprint, but
>that would be self-defeating.''
>In a recent demonstration for Inside, the software was able to pick out
>versions of Eric Clapton's ''Layla'' that had been encoded in a variety of
>ways, and even distinguish between electric, acoustic and live versions of
>the song. Impressed by a similar demonstration last Thursday, according to
>Nye, the RIAA's Sherman and a team of technology experts will spend much of
>Wednesday looking ''under the hood'' of the software in a quickly assembled
>meeting in Washington, D.C.
>Also in there pitching its wares is Audible Magic. According to CEO Vance
>Ikezoye, the company's software can uniquely identify a song by measuring
>the properties of a snippet less than 20 seconds long. Like Cantametrix,
>Audible Magic has been doing dog-and-pony shows for the labels.
>Both companies argue that Napster could put the fingerprinting software in
>its ''client'' -- the program downloaded by the user. As songs come onto the
>user's computer -- either by personally ripping a CD or by downloading --
>the software would ascertain their ''fingerprints'' and send them to
>Napster's central servers. (The fingerprints are recorded in small files --
>less than 1K, in Audible Magic's case.) Napster's computers would then
>compare the batch of fingerprints to a master list supplied by the record
>labels, and unauthorized tracks would simply not appear on its central
>index, making them effectively unavailable for trading.
>According to Hoffman of eMusic, though, the complex fingerprinting
>technology may be unnecessary. A bitter opponent of Napster, eMusic has
>notified the file-swapping company of ''between 20,000 and 50,000''
>infringing users with ''about 80 percent of (the eMusic) catalog.'' To be
>certain about the identity of the files, it has checked MD5 hashes. ''In our
>experience,'' he says, ''the likelihood of false positives with a name
>search is very low -- there's not that much reason to get involved in the
>complexities. If they do a simple name search and block the results, they'll
>get what we want.''
>Hoffman, who has seen his company's offerings of downloadable music unable
>to gain an audience because of the free file-swapping onslaught, was
>delighted by the court's order to Napster, which boosted his company's
>sagging stock price by almost 70 percent. ''I had a bottle of Jack Daniels
>and a bottle of champagne for tonight,'' he says. ''Which one I was going to
>drink depended on the decision, but I was going to get drunk one way or the
>With Napster forced to change their ways, he says, ''Champagne is on the
>This is the pho mailing list, managed by Majordomo 1.94.4.
>To send a message to the list, email phoATonehouse.com.
>To send a request to majordomo, email majordomoATonehouse.com and put your
>request in the body of the message (use request "help" for help).
>To unsubscribe from the list, email majordomoATonehouse.com and put
>"unsubscribe pho" in the body of the message.
Rumori, the Detritus.net Discussion List
to unsubscribe, send mail to majordomoATdetritus.net
with "unsubscribe rumori" in the message body.
Rumori list archives & other information are at
[an error occurred while processing this directive] N© Detritus.net. Sharerights extended to all.