A Forward-Looking Model for Online Music Distribution

First Draft Proposal
Michael Carmack
June 22, 2000

1. Current state of affairs

Numerous companies today, notably those such as MP3.com, have hailed themselves to be the future of music distribution. Indeed, in a recent press release, MP3.com stated that they consider their future role to be similar to that of a utility provider, like a water or electric company--except they will be the providers of music. In a similar context, executives at ListenSmart.com have envisioned a great "celestial jukebox" (of which, presumably, they will control) that will catalog and provide music to all who want to listen. However, if one considers the future of the Internet, these visions seem very short-sighted. If the state of the Internet today was anything like what the Internet will be like tomorrow, then these might be worthy goals. But the Internet is evolving in such a way that the need for a centralized content provider will be non-existent within a few years. And this being the case, one can see that far better models are on the horizon.

First, it needs to be pointed out what is wrong with the centralized model. The biggest problem is that by their nature, centralized databases introduce a middleman between the musician and the consumer. The online music sites trumpeted their way into the picture with announcements of how they were the solutions to the evils of the record companies; that they would put an end to the era of $15 CD's and shady contracts. Now the primary reason that the record companies were able to get in such a position was that they had control over the distribution. If an artist wanted to get his or her music sold in a record store outside the hometown, getting a record deal was about the only way to do it, because only the record companies had the cash reserves to duplicate the music in quantity and ship it, and the clout to convince stores to sell it. But with the Internet, the artist is able to bypass that middleman. Without a need for mass-production or distribution, the artist can get music from the studio directly into every living room in the world. Even the elite access to a recording studio is becoming a thing of the past. It may have at one time cost over $100,000 to get a quality production because of the millions that had to be invested in a professional studio, but with the tremendous increase in computing power and the development of sophisticated software, people can build a $50,000 studio that rivals those costing 100 times as much. And it probably won't be too long before anyone with the expertise can create a professional environment right in their living room for under $10,000. What this means is that the $100,000 it once took to record an album (and for which the record companies were needed to fund) is quickly dropping to a price that is reasonably affordable to any artist with a day job.

If the record companies fear the prospect of online music distribution, it's rightly so. They won't be needed to fund the recordings, they won't be needed to distribute the music, and they won't be needed to duplicate the CD's. The only card they still have in their pockets is that someone still needs to fund promotion. But if the record companies' promotional expertise on the Internet is anything like their distribution expertise, they'll have a long way to go. In short, an artist would be far better off doing everything themselves (recording, getting the music online, and promoting it) with at least some guaranteed return while retaining the rights to their music, rather than signing over all the rights to their recordings to a record company that promises the world but doesn't deliver a dime.

But, where are things headed with at the moment? If MP3.com or IUMA have their way, they will be the new middlemen. Before, it was Sony Music, UPS, and the retail stores. Now it will be MP3.com and Sprint. MP3.com and Sprint may never achieve the power that the record companies once had, but whenever there's a middleman money is going to be lost-- money that should rightfully be the artists'. To put in perspective just how much is lost, consider that the total costs that MP3.com incurred on the internet connection and network engineers last month alone likely topped $5 million. In comparison, all of the artists on the site combined got a meager $1 million to split between them. If MP3.com had some way of elimitating the network costs, artists should suddenly start earning six times what they are now. But now it's money being lost to the middleman.

2. A solution in waiting

The solution to this problem with the middleman is actually already laid out before us, but it has yet to be taken advantage of. This solution lies in the technology behind file-swapping programs such as Napster or Gnutella. With these programs that allow users to swap files from one personal computer to another, there is no going through MP3.com and there are no network costs to anyone beyond that of a phone or cable line. And basically, since people would probably have their Internet connections anyway (i.e. they don't just have the Internet so they can download music) this "cost" is only nominal. What this means is that all of the money that is being lost to companies paying premiums for ultra-high-bandwith redundant internet connections is pointless, because there already exists a free high-bandwidth redundant network. That network is simply the Internet as it exists between personal online computers. To be sure, any particular node is likely going to have far less bandwidth than a site like MP3.com, but what the overall network looses in point-to-point bandwidth it makes up for a million-fold in redundancy. The trick is getting this network to work in favor of the artists, but in a way that still gives (essentially) free music to the consumer.

3. Content delivery

The first hurdle is ensuring quick response and delivery. The current Gnutella model would never work. Gnutella is chaos--there is simply no reliable way to quickly find and retrieve a file. Searches can take from minutes to hours to return a hit, and even then there is no guarantee that you will actually get what you ask for, or that the file will be complete, or that it will be coming from a nearby site with a fast connection. Something like Napster is far more efficient, where users poll a central database and the database returns IP addresses of computers that have (or at least once had) the file. But Napster still doesn't provide the convenience of a site like MP3.com, nor could it the way it currently works. Even resolving the issue of quickly retrieving and automatically streaming a file (to get the same convenience of a site like MP3.com where it's just point-and-click on the "play" link), there is still the chaos issue. There's little one can do besides perform random searches and hope that a filename matches and that the whole song is there. Without resolving this issue, Napster stands little chance of moving mainstream. As long as it works like a hacker's tool it will continue to be regarded as a hacker's tool. And consequently, it will be a great piece of technology going to waste.

A better idea is something along these lines: A central server is set up so that artists can upload their music. On the outside, this will appear to work essentially the same as MP3.com or IUMA, but on the inside something very different would happen. Once on the site, the song is given a unique filename that would serve as the song identification, and the song information would be incorporated into the central server's main database. (Each database entry would consist of one song, the title of the song, the author, the songwriter, the lyrics, etc.) Now the first few streams or downloads that were requested from the central server might well come directly from the central server (since we might be the only ones with the file at first), but after the music had been sent to a few people (who's IP's would be indexed), for future requests the server would simply act as a proxy. What would happen is that people who reqested a song from the server would have the song automatically and transparently downloaded to their computer and stored in a predetermined directory, the size of which the user could configure based on how many songs he/she wanted locally stored. Users would of course have the option of not being a "distributor", meaning their computer would never get proxy requests to deliver a file, but those that did would have the "storage" directory's contents routinely polled by the server so that the server always had a fresh list of available IP's for the requested songs.

4. Securing payback

With the delivery issue resolved, the next hurdle is providing payback to the artist. Since it would sell advertising and essentially function as an MP3.com-type site, the central server would be able to pass on advertising revenue directly to the artists just as MP3.com does. But it is the hope that this would ultimately be a tiny fraction of the total money earned by the artists. A much greater share would eventually come from independent radio stations, shopping malls, companies that requested background music for their offices or web sites, and so on. It works like this: the server is the input point for all music, and point at which music is cataloged and given it's unique ID for rapid retrieval. The server maintains the database that allows one to find songs based on title, author, etc., but it becomes the role of independent providers to actually compose playlists (hereafter referred to as 'stations', even though it may be a company creating a playlist for the office intercom, not a conventional "radio station"). Now, in order to be a station provider, one would have to open an account on the server and download custom software that would report back to the server the songs that were included on the station as well as how much the station was agreeing to pay the artists for each song. The station provider would also have to agree to immediately remove from the station any artist who requested removal. Aside from that, the station would be free to put any songs it wanted on the station and pay the artists anything it wanted (even pay different amounts for different songs). What this does is that it ultimately gives the artists themselves the direct control over where their music gets played. For example the artist might see that they're being played on Cheesy Radio Station that only pays .0005 cents for every download, but it is apparent that the site is generating millions in advertising revenue. Guessing that is making a lot of money but little is being passed on to the artist, the artist has the choice to discontinue their music from being available from the site. The choice can be made by simply clicking a button that says "Request Removal". The central site would then send an automated message to the station and request that the song be removed. Stations that did not honor the request within a certain time period would face fines or perhaps having their station licence revoked.

Now, it's entirely likely that you'll end up having people running hacked software, or simply running the software without paying the artists. But for small-scale sites, the lost revenue will probably be inconsequential and large-scale sites simply won't be able to hide. Any site that got large enough to attract significant attention and generate any significant revenue would not be able to hide the fact that they weren't paying for their music. If they were sufficiently prosperous, it might be worth a class-action suit on behalf of the artists to go after lost revenue. But I'd imagine this would be a rare event. In most cases, simply blacklisting the site, mass petitioning to advertisers and threatening boycotts, and simply spreading the word that the site was not giving anything back to those whose music it was using would probably be enough to pressure the site into going legit, simply because it would be to it's economic advantage to avoid being blacklisted by the artist community and it's sympathizers. And assuming that people, given two equally reasonable choice (free music from station X or free music from station Y), would more often choose the one that was morally unambiguous, the threat of digital piracy doesn't seem so strong. It would be analagous to meeting two men on the street, both offering you a Rolex--one is a Rolex salesman giving it away as a promotional tool, the other is a crook who stole it. Which would you take? Either way you get a free Rolex; but in one case you know you're risking indirectly stealing someone else's property, and (worse?) perhaps not even getting an authentic product. Isn't the trivial advertising that comes with the genuine, legit Rolex worth the peace of mind? It would be for me, and I think it would be for the majority of users also. I don't doubt that there will pop up some kind of underground network, especially for songs that are still trying to be sold for $15 (i.e. everything by the record labels), but the artist who makes their music freely available to the consumer will surely be rewarded by listener loyalty-- listeners making a point of only listening from those sites that had met with artist approval.

5. The incentive to run the file-sharing software

What is the incentive to run the file-sharing software? Well, the first incentive is simply knowing that you are doing something good for both artists and other listeners just by helping the system work smoothly. After all, Gnutella doesn't do a thing for artists and doesn't earn anyone a nickel, but for some reason people still open up entire directories on their computers for sharing. And it's not that Gnutella requires people do this; for whatever reasons they just choose to. The artists themselves would be particularly motivated to participate, since it would be their music that they were helping distribute. And other people might not mind simply because the cost is so little: Hard drives are getting down around the $5/GB range and in a few years will probably be nearing the $1/GB range. With 1GB storing around 250 4-minute songs, it becomes evident that even this modest contribution quickly adds up when there are millions of people participating.

A far greater concern is that of bandwidth. Today, most people are still on 56k modems and even most cable modems are quite slow in transmitting data (the blazing download speeds don't seem to carry over to the other direction). But this rapidly changing. Cable and DSL are spreading like wildfire and DSL does offer high speeds in both directions. If consumer demand dictates, high speeds in both directions will surely become commonplace. At this point, the cost in providing the bandwidth will also drop to a negligibly low point for the end-user. It's true, this is still a few years away, but this proposal is not aimed at building around today's technology: that's better left to companies with large reserves of money to burn.

Now beyond the goodwill of people, we'll also offer an implicit incentive for companies running websites that want background music or shopping malls that want a background stream or just individuals who want to make their own personal radio station. Since large companies are often running on dedicated servers with high bandwidth (fractional T1 to OCx), and since it's a rare occasion that their bandwidth is maxed out, they are excellent candidates to target for help distribute the music. And again, it's best to pass this onto the individual artists as a choice. For instance, along with a statement of all the stations that they were on and a number telling the artist how much each station is paying them for downloads, artists could also be given a number representing how many files the server was providing each day, along with, say, a figure representing the average over the last month. For example:

Macy's, Kansas Store 131    0.05 cents per song    2 play    2014/2150 songs provided
Joe's Radio Station         0.1 cents per song    17 plays    104/90 songs provided
What this says is that Macy's played this particular artist two times that day, paying them 0.05 cents each time one of their songs was played. Joe, on the other hand, pays them more per song and also plays them more. One artist might look at this and say "Macy's is a rip-off, I'll remove myself out of protest to their petty payback". But another might reason that "Well, I do get a lot of exposure by being played at Macy's, and they are helping the community by providing a lot of songs, so I'll stay with them even though they are not paying as much." For something like a non-profit organization that the artist supports (e.g. Greenpeace) the artist might agree to accept no money at all. For a company that the artist does not want to be associated with (e.g. Hustler Online), no amount of payment may be sufficient. But the point is, it all comes back to the individual artists to make these decisions. No longer is there any middleman standing in the way of payment, fixing prices, or determining who gets what. That is all turned over to the artist.

This is also a naturally adaptive model, and in much more a position to change with changing times than a model that relies on a central body making these decisions. As more and more individuals join the program and having large corporations help supply the songs becomes less relevant, the companies will naturally be forced to shift their payment options to keep artists on their site. For example, where a company might get away with paying an artist only 0.1 cent per play next year because they offer the availability of much of their bandwidth, in 10 years when high-bandwidth is more common and many people are distributors, the companies' role in being a provider will become less valuable and the artists themselves will begin to place higher monetary demands for their music. And since each artist can make that choice by clicking a "Request Remove" (or perhaps a "Request Payment Increase"), as soon as masses of artists start demanding more money, companies will have little option but to meet the demands. But this will all work out economically in the end--a balance will naturally be established based on the quality and desireability of an artist's music and how much someone else is willing to pay for it. If any individual artist starts trying to charge outrageous prices and the price/desireability ratio get's too high, stations will naturally start picking up songs by other artists. Similarly, if a station starts paying too little to the artists it plays, the artists will start dropping one by one. Eventually, unless the station agrees to up it's payout, there will be few artists remaining.

6. Implicit "gaming" protection

Since ultimately the payback directly from the central server's advertising would represent a relatively small fraction of the total payback, there seems little risk of gamers or bot-runners who play their own music over and over in an attempt to earn money. Why? Well, for starters, any given IP address could be limited in how many downloads it was allowed per day. For example, MP3.com only allows one download per song per pay for each unique IP address. For another, by not offering any user-created stations on the central server, there would be none of the station gaming where hundreds of users agree to play a given station over and over for the privledge of being on it. And since the payback from any external station must come from that station, the only people that would end up paying station-gamers would be the gamers who created the station. There's simply no incintive for anyone to station-game; the payoff is implicitly removed.