Chris Anderson says “almost anything is worth offering on the off chance it will find a buyer.” I think it is important to know that even though the long tail model has exciting implications, we cannot forget the “almost” in this statement.
Don’t get me wrong. I agree that there exists a long tail. I also think, though, that there is an even longer tail that nobody will buy from. If I may, I will call this body the “crap tail.” Consider the example of “crap tracks.” Anderson says, sure, there are a lot of crap tracks out there, and he is not worried because they “just sit harmlessly on some server.” I believe we must take this assertion with a grain of salt. Sure, the aggregate sales of obscure music has large implications for how we should organize the industry, and clearly there are worthwhile profits to be made. However, there are real costs associated with producing and making available these crap tracks. Suppose a population of really crappy teenage singers decide to sell their stuff on iTunes, and suppose that nobody wishes to buy their music. Perhaps the the server space necessary to store the crap songs costs more than revenue from the crap songs. I think the aggregate effects of scenarios like this should not be forgotten in the quest to milk the long tail for all of its revenue. The crap tail, if it is sufficiently large, could at the very least erode the long trail. I firmly believe that there is some music out there that has no musical merit or camp value, and will never be bought. If the volume of this music is sufficient enough, it can be an invisible hole in the profits bucket. Thus, I believe it is most beneficial NOT to commit fully to the long tail model. I think that to truly optimize the online music industry, there needs to be some sort of filter so that the online retailers know that the costs of providing a track will be covered by sales.
I think that Rhapsody illustrates the effects of the crap tail. At the time of printing, the library consisted of 735,000 tracks. Of those, the author states that only its top 400,000 tracks are streamed at least once a month. That’s cool and all…but you still have 335,000 (45%) of your tracks that presumably don’t get streamed at least once a month. It must cost RealNetworks some sum of money to keep these tracks that nobody listens to. I suppose the danger is in overestimating the length of the long tail, and being stuck to pay server fees for music literally nobody wants to listen to. For industry optimization to work completely, I think providers need to calculate the length of the long tail and cut down on server space taken up by truly crap tracks, because they cost money too.
Again, I believe that we should NOT commit fully to the long tail model. Like many models, it is useful for explaining certain phenomenons. I like the long tail model because it truly gives a new insight on how the music, book, and film rental industries should operate. However, it is mildly problematic because the model says that we should make available every single piece of media we can in the hopes that there will be a buyer. On the whole, the approach may be profitable, but I think a little consideration and caution must be exercised so that the costs of maintaining truly crap tracks that nobody will buy do not erode into profits.