Recent Harvard Graduates Revolutionize Downloading of Internet Music

During reading period two years ago, three self-proclaimed “radio nerds” at Harvard Radio WHRB stumbled onto an idea which led them away from the frenzy of exams and into a world of patents, venture capital and big bucks.

In an attempt to make digital music easier to navigate, Clifford Chen ’00, Gabriel Dorfman ’02-’04 and Michael Papish, class of 2000, co-founded MediaUnbound, a service that creates playlists based on its users’ tastes.

Just last Tuesday, the Cambridge-based company signed a three-year deal with pressplay, a legal version of Napster backed by the world’s three largest record companies––Universal Music Group, Sony Music Entertainment and EMI Recorded Music.

The site will use MediaUnbound’s technology to recommend music to its subscribers.

While the dot com world has proven harsh for many who left school on its promise, Papish says MediaUnbound is thriving, although he will not specify the amount of the deal.

“This deal is huge for us,” Papish says, “Pressplay has an enormous amount of money behind them.”

The Beginnings

MediaUnbound was born two years ago, Papish says, when he and his friends began to notice the difficulty of navigating the music options available through the Internet.

After creating a WHRB website which would allow listeners online access to the radio station, Papish says he began to consider the computer’s untapped music potential.

“We were bored and didn’t want to study,” he says. “We saw the computer was a horrible thing for music, and it seemed like an exciting area to explore. But beyond that, our ideas were still fuzzy.”

According to Papish, it took the three students only 10 days before they began to receive funding for their project, from WHRB alums and Harvard Business School grants.

“It was an easy time to start a company because at that time every business student’s dream was to drop out of school and make it big,” Dorfman says.

The students then began to research how to best market their idea.

According to Dorfman, they identified a need for a more accurate system of online music recommendations.

“MediaUnbound solves the problem that there’s too much stuff on the Internet for people to know about it all,” Dorfman says.

The group discovered how to personalize a set of recommendations from seemingly limitless lists of songs by combining computer methods with the expertise of a number of music analysts.

On a typical day, music analyst Alex Helsinger ’98-’00 says, he spends his time expanding obscure playlists, like the hardcore ska list.

Helsinger is one of 10 MediaUnbound employees whose job is to know everything about popular music from its inception to the present day, he says.

Without these music analysts, “the purely technological is too predictable, you miss the excitement of how people relate to music––the mystery part,” Helsinger adds.

MediaUnbound’s group of employees comes nearly exclusively from the ranks of WHRB.

“What we looked for was people who know what makes people tick. And that’s basically what people at WHRB were doing already,” Papish says.

One of their early recruits, for example, is Matthew D. Pakulski ’94, who began performing in a punk rock band during his undergraduate years in Dunster House and owned a record company for a few years after graduation.

“College radio DJ’s are all up on obscure musice,” Helsinger says. “Our ultimate goal is to create a virtual college DJ.”

MediaUnbound

At the start, Helsinger says, the students worked to discover the reason people love particular songs and often arrived at a surprising set of associations.

For instance, says analyst Joshua Mann ’00, fans of alternative musician Beck tend to appreciate Prince.

“[Beck’s] whole popular image is derived from Prince,” Mann says. “We found that given exposure to both artists, people tend to like both. And they’re usually surprised by it.”

For the moment, Dorfman says, MediaUnbound leads the competition in personalization. Its competitors rely on computer algorithms, which are much less accurate.

Following Media Expo in October 2000, a convention for Internet media, Dorfman says he realized that though “other people were trying to do the same thing, it didn’t work as well.”

Helsinger agrees.

“A number of users were saying we have this uncanny ability to figure out what they liked,” Helsinger says. “They think their tastes are too complex and nuanced to be understood. And while it’s true that they’re complex and nuanced, it’s not true that no one can understand it.”

But it can be a hard idea to get used to, Helsinger says.

“The idea can make some people bristle a little bit,” Helsinger says. “They think, ‘How can a computer tell me what I like?’”

Pakulski, a MediaUnbound music analyst, says he is not sure how much real change the burgeoning company can effect, questioning how receptive the public will be to suggestions of less mainstream music.

“I feel like if you want to reach a wider audience, you have to dumb it down,” Pakulski says. “What I like about MediaUnbound, though, is that they try to make people a little better informed than just the most popular thing. But I’m too jaded to believe it can really change anything.”

Papish is more optimistic about how his company will benefit independent artists.

And there are those, even at MediaUnbound, who still espouse old-school values.

“My preferred way of listening to musice is to throw on a CD or an LP,” Pakulski said. “I still prefer the actual physical object.”

—Staff writer Eugenia B. Schraa can be reached at schraa@fas.harvard.edu.