'

CDDB tug of war--users are the losers?

More than 885,000 users have helped build CDDB, a online music database that's been maintained by Gracenote. Who's in control of the database? The courts will decide.

Earlier this month, when he decided to record a new compact disc, Joe Smith followed a procedure he's done hundreds of times: Insert the CD, get into the online music database and select songs.

Yet this time, instead of displaying the usual song information, the popular CD-burning software made by Roxio delivered a different message. Roxio products were no longer supported by Gracenote's CD information service, which housed the database Smith had been using to match song and album titles with a disc placed into his computer--a collection entered in large part by individuals on the Net.

"I blame Gracenote," Smith said, for "taking what is essentially an open-source database and closing it off." Because so many people voluntarily helped build the database, being locked out is a slap in the face, said the software engineer who asked that his real name not be used.

The data blackout is the result of a copyright battle between the two companies that may have far-reaching consequences. Already, thousands of customers have been cut off from the largest such online database, a collection of 885,000 records called CDDB. While the legal dispute centers on whether Roxio can use online databases created by people on the Internet, many people are concerned that CDDB--maintained for the last three years by Gracenote but built from individuals' submissions--can be owned.

The issue is central to a lawsuit that could potentially affect a variety of communities throughout the Internet, from so-called open-source programming projects to comments archived on online bulletin boards.

If Roxio wins, open-source competitors to Gracenote will be legally free to build databases that can't be co-opted by others. If Gracenote wins, other databases relying on individuals' submissions could be considered the property of whatever company collects and formats the data.

More immediately, the case will determine the fate of two alternative CD databases--Freedb.org and MusicBrainz.org--that were created after several programmers on the Internet realized that Gracenote had essentially fenced off the original CDDB by changing the terms of the licenses.

Like Gracenote's CDDB, the two open-source services allow people to download information from the Internet on just about any audio CD they insert into their computer. While MusicBrainz.org uses a different system that is incompatible with the Gracenote database, Freedb.org uses the same kernel of software and data--information that Gracenote claims is its own.

Roxio originally intended in late April to shift its software from Gracenote's CDDB to the database run by Freedb.org. However, the suit aims to prevent any of Gracenote's 1,800 commercial licensees and 2,200 free licensees from jumping to competitors.

The outcome of the suit will determine whether CD-listing services owned by the public can exist.

"Obviously, we would like more companies to integrate their software with MusicBrainz," said Robert Kaye, the lead programmer behind MusicBrainz.org, which recently changed its name from CDIndex.org. "A lot of companies are looking at us, but they hesitate because they are afraid of being sued."

While Gracenote claims the collection of CD information at the heart of the CDDB system belongs to the company, many cyber-rights activists and open-source advocates are raising eyebrows. Because much of the database was entered in by the public, they argue that the public should own the result.

"The most egregious part of this lawsuit is that the company has taken the public's work and propertized it," said Robin Gross, an attorney for the Electronic Frontier Foundation, which fights for public rights on the Internet.

Furthermore, Gracenote's copyright is backed with a software patent, which in some part gives the company a lock on the organization of such a database.

Gracenote's case will clarify how current database laws should be interpreted.

Today, few laws protect a database. While two bills aimed at protecting a collection of data are wending their way through Congress, companies with proprietary databases typically rely on copyright to protect their archives of data.

In the past, the protection has only been granted for data organized in a way that is considered "original." Yet, the creator of the database doesn't have to be Picasso, said David Marglin, general counsel for Gracenote.

"If your arrangement of facts (has) even a small modicum of originality you can get a copyright on your arrangement," he said, adding that in Gracenote's case, the company's CD database has originality that extends beyond the fact that people have submitted to the archive.

While several cases have established such copyright protections for databases, the CDDB is fundamentally different from other collections of information created by companies, said Eben Moglen, professor of law and legal history for Columbia Law School and general counsel for the Free Software Foundation.

Although other companies have paid employees to enter information into a database, the public typed in many of the song titles and album names contained in Gracenote's CDDB, said Moglen.

"It's as though you assembled the telephone white pages by asking everyone to call you up and give you their information--and then turned around and copyrighted the work," he said.

A proprietary database or a publicly created one? The answer lies largely in the law's reading of past events.

When it was first created by programmer Ti Kan in 1993, the CDDB was a part of XMCD, Kan's application for playing CDs on Unix-like operating systems. At that time, the program didn't use the Internet. It merely saved the CD information to the computer on which the software ran.

Within a couple of years, with the help of buddy Steve Sherf, Kan quickly created a system for taking track information entered by an individual and uploading it to a database on the Internet. If someone placed a CD known to the database in his or her computer, the player could download the information from the online database, saving the person the few minutes it would take to type in the information.

Sherf is credited for most of the work on the CD database component of the system. "Ti and I worked together for years," said Sherf, now chief architect for Gracenote. "While he invented XMCD, he was only interested in writing his program. I made the service myself and wrote the code."

Without people submitting song titles to the database, the CDDB would be only marginally useful. Kan, long a supporter of Free Software, released the source code of his program under the GNU General Purpose License. Such a license gives people the right to modify and use the software as long as the source code of the application is released under the GPL.

With the code--and many assumed, the data--released as a public resource, submissions to the CD database took off.

"The CDDB project out of which (Gracenote) grew was an open project," said Bruce Perens, a longtime open-source evangelist. "Because of their positioning as an open database, they got thousands of people entering in titles and songs on their own."

Yet, unbeknownst to most people, Kan included in the code for XMCD commands to append copyright notices on people's submissions. Kan did not answer repeated requests for comment, so it's unknown when the copyright notice was added and how visible it was to users of the system. But Gracenote has claimed that about 100,000 records were copyrighted by Kan and that copyright was then transferred.

In 1998, high-technology venture firm Escient bought out Kan's and Sherf's work on the CDDB and combined the duo's software with other technology from multimedia firm Ion. Two years later, the combined firm, now called Gracenote, patented the software solution used by the CDDB to locate matches in the database.

When Escient claimed ownership of the database and meted out strict licenses, the open-source community hit back.

The most successful competitor to date is Freedb.org, an open-source project started by Michael Kaiser in 1999 from the last publicly mirrored version of the CDDB (from 1997) and the open-source server program.

"There were no copyright notices included in the archive," said Joerg Hevers, one of the project leaders of Freedb.org. Today, there are more than 330,000 titles included in the database.

Others were equally upset by Escient's move to a proprietary license.

"CDDB had adopted a draconian license," MusicBrainz's Kaye said of his decision to start his own CD information database. "You had to do what they say or you didn't get a license and you were screwed."

Unlike with Freedb.org, Kaye started a CD database from scratch. Called CDIndex, and now MusicBrainz, the database calculated the "magic number" identifying CDs differently than Gracenote's formula. Moreoever, the program code has been created from scratch.

The key to both databases is the lesson learned by the public: Keep it open. Both databases not only have open-source software, but the content of the database is copyrighted under the Open Content License, the open-source equivalent for data.

"One person getting ripped off is not a big deal," said Kaye. "If you are ripping off 10,000 people that have put in the data for 20 CDs, that is not a trivial contribution for the community at large."

Kaye should know. He claims that he entered at least two dozen records in Gracenote's CD database.

While he's resigned that he no longer has any control over those records, he vows that it won't happen again.