Friday, July 14, 2006

gracenote graces us with its database

i spotted this blurb in the indy star:

Gracenote Inc., a tech firm founded and chaired by Indianapolis businessman Scott Jones, has landed a deal with dozens of music publishers to offer digital copies of more than one million song lyrics over the Internet, according to the Wall Street Journal.

That paves the way for Apple Computer and other companies that use Gracenote's services to offer lyrics with song downloads.

The deal is also likely to lead to legal action by music publishers against a multitude of Web sites that post lyrics without permission from copyright holders.

Gracenote, based in Emeryville, Calif., did not disclose financial terms of its agreement.

reuters has more:

Until now, consumers' access to song lyrics has been largely through unauthorized sources, which usually provide inaccurate content, the company said.

Publishing industry officials cited Web sites like ( and ( among those who provide their catalogs' lyrics without their authorization. These sites could not be reached for comment.

inside indiana business has a post up about the story, with a couple soundbites from scott jones. check out the first audio clip; its caption is "Jones says one day you won't have to know the artist or title of the song to be able to find it and buy it." and that sounds great, except that you can already do this on google, and the reason you can is because of volunteer lyrics sites like go ahead and try it. type a phrase from the lyrics of a song into google, and chances are you'll find it immediately. you'll improve your chances if you add the search term "lyrics". i do this all the time to look up hip-hop lyrics, since my local blazin' hip-hop & r&b station doesn't seem to believe in back-announcing.

so basically, jones is bragging that he will shut down these free services and replace them with a pay service that does the same thing. that's SOP for the music industry, but who does jones think he's fooling here? sure, he also mentions having such capability on an ipod, and a sort of "name that tune" program where you "hum a few bars" and the program can identify the song, both of which would be cool and don't currently exist, but there's no technological reason why these services couldn't co-exist with free user-driven lyrics sites, or why they couldn't have been created without a precious license from gracenote. (they could be done just as easily using freedb, for example.)

but if you know gracenote's history, you'll spot the real irony here. once upon a time, gracenote was itself an open-source database called CDDB with content created by users. if you were playing cds on your computer in the mid-90s, you probably used CDDB and might have even submitted a few tracklists to the database.

but then, once CDDB was full of content (user-created content that was as inaccurate as the stuff on, its founders weren't so interested in open-source anymore. from wikipedia:

In 1998, Kan and Scherf incorporated CDDB into a privately held company with investment from Escient, a high-tech venture firm. CDDB was then renamed Gracenote. The maneuver was and remains controversial, because the CDDB database was and is built on the voluntary submission of CD track data by thousands of individual users, who received no compensation for their work. Initially, most of these were users of the xmcd CD player program. The xmcd program itself was an open-source, GPL project, and many listing contributors assumed that the database was free as well. However, at some point the code for xmcd was modified to append copyright notices to all submissions. How visible or open this was to contributors remains a matter of debate. Many contributors of track listings were angered at the transfer of these listings to a profit-making entity which proceeded to make money by charging license fees for access to a database of track listings which individuals had contributed for free.

As of 2005 Gracenote claims that its database contains information on almost 4 million CDs. The reliability both of this statement and of the database itself have been challenged. Because the information going into the database has not been subjected to quality control, duplicate entries are very common. David Jennings, in an article entitled "How many CDs are there in the world?" gives an example of a six-CD set in which "two of the six CDs appear twice in the database, and one appears three times." An article on the AtomicPop website cites Ty Roberts, chief technology officer of Gracenote, as saying that there are approximately 500,000 individual CD titles commercially released and available for sale today in the United States.

in short, gracenote started out as the same kind of free, unlicensed, user-driven database service as sites like, and has the same kinds of accuracy problems as those sites. but gracenote was the first to "go legit" and is now helping in the fight against other free, user-driven services.

yet you won't see these facts in the major news coverage of the gracenote deal. these stories are full of quotes about how inaccurate unlicensed sites are, but you won't find any mention of the similar inaccuracies in gracenote's own database, or of gracenote's past as exactly the kind of service the industry is trying to demonize.

No comments: