What would the be-all, end-all music Web site be? Imagine AOL's chat community, combined with Rolling Stone's content and Sidewalk's local concert listings, mixed in with CDNow's buying power and streaming audio to listen while you surf. Now picture all of that personalized to your taste.
MyLaunch, created in October by the publishers of the music CD-ROM zine Launch, is hoping to be a site just like that. And in order to get a jump start, they bought Firefly's 2-year-old Bignote - the showcase site that introduced collaborative filtering to the world and accidentally spawned a thriving music community. But while Bignote was a "just for fun" community, built to demonstrate Firefly's filtering technology, Launch wants to turn it into a profitable service that's simultaneously very general and individually tailored. That is a daunting, and possibly misguided, task.
"We're going to make this work with personalization for the consumer, or die trying," says Launch CEO David Goldberg. "It's very powerful for people if it does work."
Launch was founded in 1994 when David Goldberg, a former marketing executive at Capital Records, created a CD-ROM music magazine focused on indie rock. Chock full of video sequences and audio clips, Launch now has about 200,000 subscribers and is chugging along nicely. Launch's entry into the noisy arena of online music publishing is intended to tap into that base and create a more personalized service - complete with news, features, chat, audio clips, and album sales - for a more diverse group of music-lovers. Firefly, a software company looking to unload a site that they'd never really intended to create, was happy to sell them Bignote as a base for the project.
Bignote leverages a community knowledge base - the engine was built around users rating and reviewing musical artists and albums, then spitting back uncannily accurate recommendations of different musical artists based on ratings by other users with similar tastes.
Launch announced their purchase of Bignote in October, after nine months of tweaking the site to fit their editorial mission. The Bignote site came complete with about 500,000 members, 14 million album ratings, and built-in community tools; what Launch added was a fresh design, and extensive editor-created content.
Eclecticism sidelined
While Bignote was focused on the evaluation of music - an addictive process that encouraged users to rate hundreds of artists - the MyLaunch site has moved user-recommended music to the sidelines. When new users join, they are asked to pick a local radio station they like (primarily mainstream stations) and then to judge albums from the playlist of the radio stations they select.
Those playlist ratings then generate the personalized content on the user's "MyMusic" page, which offers up general music news and "personalized" features, album reviews, concert reviews, and upcoming releases based on your musical preferences. If the radio station you select only has 20 songs on its playlist, that's all you're going to get to rate - and that's not enough to create a personalized profile. (If you don't like any local radio stations, you're out of luck.)
The inherent problem is that MyLaunch is attempting to be an all-in-one music destination for lovers of too many kinds of music. You can listen, buy, read, and chat about "your music" - but "your music" is defined by those 30 or so albums you rated at the beginning. You may like Aphex Twin, but you're going to end up with Celine Dion.
"We have some bugs with the thing - you have to do a fair number of ratings. If you don't do enough, you get these generic things," explains Goldberg. "We still have some teething issues."
Musical eclecticism, unfortunately, is one of those issues. If a user likes pop, and just pop, they'll do fine. But if you happen to be a lover of techno with a side interest in indie rock, plus a quiet passion for '50s jazz and '70s soul, you're going to run into some problems. Unless there are lots of people out there with similarly diverse tastes, your personalized page could end up with a hodgepodge of music features you care nothing about.
MyLaunch executive producer Douglas Glen suggests that users with eclectic tastes create multiple aliases. "If you're going to like multiple types of music, you run the risk of the algorithm getting mushy."
Judging from the attention span of your average click-a-holic, though, it's unlikely that many will take time to create multiple aliases for each of their musical interests. Glitches such as these are fine for a toy - which is essentially what Bignote was - but if the tool is being made into the basis for a large-scale service, the bugs are going to have to be fixed.
The Launch creators know they're tackling an ambitious project, and are scrambling to fill in the gaps. They're about to add an area called "MyRatings," allowing users to retroactively rate albums in genres other than those defined by their local radio stations. They're trying to expand the editorial staff to generate the extensive content necessary for personalized sites for hundreds of thousands of very different users. They're extending the size of those radio playlists, to enable more initial ratings.
Expert opinion
Even with the additional work, whether or not an audience of music lovers will feel compelled to read from such a generalized source remains a big question.
Clifford Nass, professor of communication at Stanford, defines a theory of "specialists vs. generalists" in his book The Media Equation (co-written by Byron Reeves): If a person or media form is labelled as a "specialist," people will almost blindly trust its insights. For example, subjects who watched a TV carefully labelled "news television" gave its news programming higher rating than a TV labelled "entertainment television" with the exact same programming.
According to Nass, personal agent tools like MyLaunch will likely fail to garner the trust of those who have strong musical likes and dislikes, but might serve well for what he calls "incidental music listeners" (aka musical ignoramuses), who would consider even an all-encompassing site more of a specialist than they are. Says Nass, "People passionate about a genre will find this unsuccessful, but general people who are not passionate about specific genres will find the service good, because it will be seen as a specialist in 'music.'"
Still, the Firefly and Launch creators insist that MyLaunch can be both a specialist and a generalist. "The point of the Web is that special-interest publications can go into the depth and detail a general publication can't," says Saul Klein, senior VP of brand and strategy at Firefly. "We designed these tools to go as niche or as broad [as users want]. You know that Launch can offer you all kinds of things, but it can also go very deeply into one area."
More content?
But that depth in one area currently comes only from the reviews of users with similar tastes who have contributed to the site's database; the "MyMusic" main page, on the other hand, is still being generated magazine-style by a staff of editors. And if Launch plans to create what are essentially personalized mini-magazines for eight or nine different genres, they're going to have to generate a small mountain of features, reviews, and news for everything from hip hop to jazz to country to techno. As Goldberg says, they've got an editorial staff of five (plus 20 or so freelancers), and half a million pages of content - and that's still not enough.
A saving grace, however, could well be the one feature that Firefly had from the start - the community. Although it may be a daunting, even impossible, task to create "specialist" editorial content to satisfy every unique user's tastes, eclectic users may at least be able to find a friend who shares the same bizarre passion for gospel, techno, and heavy metal.
"The way it works is that whatever your idiosyncrasy is, you'll always find people within the community with the same taste, even if it's not a hundred thousand people," explains Firefly's Klein. "That's the beauty of the system - you'll always find an affinity group."
