There is a hole in our current system of music distribution. As the cost to physically or electronically reproduce a song approaches zero, there is less and less of a need to "own" music. After all, you can own the copyright, or you can own a physical recording of a song, but you can't "own" the song itself - once someone else has heard it, whether or not they have paid you, they "own" a copy in their head, too.

This shift favors a market where you pay to experience the music instead of to own a physical or digital recording. I foresee a service that has a vast library of songs, and charges you some small amount (three cents, perhaps?) per song you listen to. At three to four minutes per song, this amounts to around fifty cents an hour - well below the "who cares?" price point for most of the music industry's target demographic. The industry would favor this change, as they get to steadily collect money every time you listen to their artists. From the consumer standpoint, today's current teenagers (and tomorrow's young adults) are a wired generation who have grown up multitasking - the ideal social group to listen to new music while working or studying and then recommend it to peers either verbally or electronically.

This would be facilitated greatly by a massive social networking computer program - something of a cross between iTunes and Myspace. I'll call it MusicBuddy here, because you know it will have some idiotic name like that. MusicBuddy would have a huge library of readily available songs, and would be able to accurately suggest songs you might like based on what you've been listening to the most, what your friends listen to, and what people who share your musical tastes also like. There are some programs like this now, but advances in social networking over the next several years should help MusicBuddy to determine what someone who replays "The Devil Went Down to Georgia", "The Phantom of the Opera," and "Rock Around the Clock" might like - current programs would assume you like all country songs, musicals, and oldies. With a large enough database and enough processing power to sift through it, MusicBuddy would have no need for categories.

When this shift happens, current advertising and producing models would also become obsolete. Artists could still be promoted by the huge labels, but the majority of song "sales" would be from friends wanting to share their new finds with other friends, or from people who are willing to take suggestions from MusicBuddy. (If the suggestions are good, then more people would be willing to use it, which gives a larger database to pull from, which makes the suggestions even more accurate, etc.) Each song gets by on its own merits, so "filler" songs for CDs will be less necessary. I don't expect physical CDs to disappear entirely, but they won't be the primary method of music distribution.

So now that artists are essentially advertised by MusicBuddy and competing programs, and the cost to produce the music is virtually nothing (you no longer have to hire 20 people to make your song's background sound like a 20-piece band), the current record labels lose their bargaining power. They currently can threaten artists with virtual obscurity if the artists don't give them a large portion of the fees, but MusicBuddy would allow artists to gain reasonable popularity just based on local performances - a few people hear their music and search for it on MusicBuddy, and word gets around through social recommendations, internet memes ( "Hey, that's the Numa Numa song!" - Editor's note: Dragostea din tei by O-Zone), and MusicBuddy recommendations. Prices for physical CDs drop somewhat as the middlemen are cut out, and even though physical CDs are harder to find locally, they're still available through numerous websites.

Individual "collection" sizes will also grow significantly - MusicBuddy (or your iPod-of-the-future) can remember everything you've played at $0.03 each, and since you're not physically storing your own copies of these songs, you could have a "favorites" list tens of thousands of songs long. The current iTunes random-but-weighted shuffle criteria would fit this model as well - you listen to a mix of songs you love, songs your friends love, songs MusicBuddy thinks you would love, and songs you've requested before but not recently.

Personally, I organize my music by mood - I listen to this group when I'm sad, this group when I'm frustrated, this group when I'm cheerful, etc. Given enough data, MusicBuddy could predict "these are the songs people like to listen to all at the same time," and would suggest only "my girlfriend broke up with me and life sucks" songs when your musical tastes show you're in that mood, and only "wow, life is wonderful!" songs if you request similar ones.

So this is my prediction for the future of music. Any aspiring programmers out there want to take a crack at creating my fictional MusicBuddy?