NEW YORK (TheStreet) -- Alright, this is, by bounds and leaps (not even leaps and bounds), my favorite portion of the whole behind-the-scenes experiment with Pandora's (P - Get Report) Music Genome Project.
In Part One, we covered what, when you consider the entire series as a whole, amounts to some basics about the MGP, focusing on the ways Pandora personalizes and nurtures music discovery on a track-by-track basis.
In Part Two, we geeked out just a little bit more as VP/Playlists and Chief Scientist Eric Bieschke detailed how Pandora breaks new bands and breaks in music you might not be familiar with (that's discovery). We also covered a bit of Pandora's history, which has a lot to do with startup culture, the attendant struggles, stops and starts and Apple's (AAPL - Get Report) iPhone. It also focused on Bieschke, who started working at Pandora, as the company's second employee before Pandora was even Pandora, at 19 years of age.
Here in Part Three, Bieschke "debugs" my account. The information revealed right alongside the name of the track playing on the desktop version of my Pandora account is so proprietary I'm not even allowed to show it. So, please pardon the fact that you can't see the screen.
Must Read: Old Dominion Freight Just Keeps on Trucking
But you can hear what we're talking about and it goes as far as anybody has ever gone into uncovering the inner workings of how Pandora does what it does. It's telling. And instructive. Because it debunks the simplistic notion that Pandora is nothing but a jukebox randomly spitting out songs. The level of human consideration and scientific complexity that goes into what the Music Genome Project does is truly staggering. And it rests any doubts about Pandora's veracity and sustainability as a business, one that has fended off all competition, including Apple's iTunes Radio.
Here in the first video segment, Bieschke and I start going through my shuffle. Most of this is self-explanatory. Next to each song title, there are seemingly random letters and numbers that denote statistical markers and the algorithms that make up the MGP. Bieschke tells us what they mean and how they operate independent of and in collaboration with one another.
The second song that comes up piques our interest because it's by an artist we had discussed earlier -- Michael Mazochi. He's a local LA musician I like a lot and whose acoustic/folk work appears frequently on my Pandora shuffle:
In this video, I fire up an artist-specific station -- First Aid Kit Radio -- and Bieschke takes us inside more MGP particulars. What's most interesting to me is the MGP's obvious track-by-track analysis. As Bieschke explains, Pandora was trying to figure out which musicological style I most prefer within this band's catalog. That's fine grain stuff. And it's precisely the sort of thing that sets Pandora apart as an Internet radio platform:
In another article that will follow this one, I include one more segment of song-by-song analysis of my Pandora account. We switch to Taylor Swift Radio and watch something very interesting unfold ...
--Written by Rocco Pendola in Santa Monica, Calif.