The Long TailChris Anderson started this with an article in Wired abouttwo years ago. He claimed that half the books sold atAmazon were obscure books not offered in anyconventional bookstore. (He’s backed off to perhaps 25%)Thus, he argues, the Internet is encouraging an enormousamount of specialized creativity for specialized interests.
Mass Media or Specialized?Some kinds of media don’t lend themselves to massproduction. Poetry reading and lectures, for example, arepretty restricted in the number of people who can attend aperformance.On the other extreme is short-wave radio, which can reachlarge parts of the planet. Enormous numbers of peoplecould listen at once.
Why mass or specialized?If there are large economies of scale – if making manycopies isn’t as expensive (per copy) as making a few, theneconomics encourages a few “blockbusters”. Forexample, radio is an example of economy of scale; theradio station incurs no additional cost when more peoplelisten to it.And if there are high startup costs – if it is difficult forsomebody to get into the business – that also encouragesa few mass-market producers. You can’t easily start anew radio station because you need an FCC license.
Consider book publishingIn the 18th century, with type held in wooden forms andprinted on wooden presses by hand, it was hard to makemore than 300-500 copies in a single press run. So booksand newspapers appeared in great variety and were oftendistributed only locally. In the 19th century, steam presses,iron presses, and machine-made paper meant that youcould have press runs of tens of thousands of copies.Nowadays, you can hardly afford to turn on a real bindingmachine unless you have 2,000 copies to put through it,and every airport bookstall has the same John Grishamand Danielle Steel novels.
But there are different modelsEven if you ought to print many copies, you can choosewhether to focus on books that will sell them quickly orover the long term. According to Jason Epstein (founder ofAnchor Books, and thus inventor of trade paperbacks),until the 1970s book publishers focussed on the backlistpotential of a new book: would it sell for many years? Thatmeant returns were not a problem: the copies wouldeventually sell. But in the 1970s, perhaps as a function ofrising interest rates, income tax law changes, and buyoutsof publishers, the book business changed to emphasizebest-sellers that would sell quickly.
So what about the Internet?Well, there are enormous economies of scale. It’s almostfree to have more people access your web page (not quitetrue, since if the volume gets high enough you need moreservers and your ISP charges you more, but it’s prettycheap).But there are no startup costs. Anybody can get a websiteand start serving up pages.So the Internet could go either way, and some years ago Iwould ask this question and say I didn’t know the answer.
Consider AmazonAmazon, as part of its competition with Barnes & Noble,tries to advertise how many books it has for sale. Itscrounges through every publisher it can find, plus everyout of print book it can find, and claims to sell more than5,000,000 books.No physical bookstore, realistically, can sell more thanabout 100,000 books. The whole town of Hay-on-Wye hassomething like 1,000,000 books for sale.Chris’ discovery: perhaps ¼ of the books sold by Amazonare not in the top 100,000; and would thus not be sold byany “brick-and-mortar” bookstore.
Anderson’s answerThe Internet is going for many diverse sources, rather thana few “blockbusters” (a somewhat unfortunate term; itderives from a bomb used by the RAF in World War II,which literally destroyed an entire block of houses).It thus represents a chance for obscure voices, whetherpeople who have interests in the railway history of Albertaor your local teenagers starting a rock band, to find the fewpeople who want to listen to them and distribute their work.
VideoIn the 1950s, at its peak, 70% of American householdswould be tuned into an episode of “I Love Lucy”. Todaythe top show (CSI: somewhere or other) has an audienceshare of about 18%. As recently as 15 years ago, thatwouldn’t have made the top ten.Instead, people watch selections from the 500 cablechannels we now have, or from YouTube.
MusicThe biggest music retailer in the United States is Wal-Mart,and they sell some 2,400 CDs, with perhaps 40,000 trackson them.iTunes has 2 M tracks – P2P has perhaps 9M - there areprobably 25M tracks in total. 40% of Rhapsody sales arebeyond the first 40,000. People are, in aggregate,listening to a lot more than top-40; even though top-40 isall you can find on the radio or in Wal-Mart.
MoviesBlockbuster Video gets some 90% of its rentals from thenew theatrical releases.Netflix offers 40,000 DVDs; and its sales are 30% newreleases and 70% others.They claim that recommendations, searching, and othertechniques encourage people to expand their choices inmovies.
TVYouTube is now seeing 100M downloads a day. Itsviewership is comparable to a network.Counterargument: if each download is watched for aminute or so, that’s 2 M hours of watching; ordinary TVgets 500 times that much attention.Personally I find it hard to see the difference betweenYouTube and “Amazing Home Videos”. But YouTube isgrowing, and network TV is dropping.
Books againCompetition between Amazon and Barnes and Noble hasgreatly increased the number of books you can readilypurchased; Amazon now has 5M books for sale andAbebooks has 45M.There are now over 80,000 publishers in the US and theypublish more than 200,000 books per year. Both numbersare rising rapidly as the low-grade books can find anaudience.
The other sideAccording to the Wall Street Journal, there may be a lot ofthings sold, but all the profit is in the top 100. In fact,satellite radio, which started off like cable with the goal ofhaving many specialized channels, is now sending outmore and more top-40 style services.Similarly, cable networks are becoming more and morehomogenous; there was a time when Bravo and A&E werechannels devoted to things like classical music, and nowthey do reruns of cop shows.
Lee Gomes v. Chris AndersonLee Gomes wrote in the Wall St. Journal that 2.7% ofAmazon’s titles are 75% of its revenues. Thus, he argues,there is no “long tail”. Anderson replies: if Amazondoubles the number of titles, 2.7% of its titles becomestwice as many books.Some economists (Brynjolffson, Smith and Hu) report thatAmazon gets 52% of its profits from the tail and 47% fromthe first 100,000 titles: this results from the heavierdiscounting on the best-sellers compared with the tail.
The 98% rule?Anderson wrote that almost everything sells at least onecopy; specifically that Ecast, an online jukebox, says that98% of its songs is played at least once a quarter.According to Anderson, again, every song on iTunes hassold at least once.Gomes disagrees: in one month, Rhapsody reports that22% of its 1.1 million songs did not sell at all.Note that there is no contradiction: if songs are selling atrandom, and 22% don’t sell each month, 1% won’t selleach quarter and all but one in a million will sell each year.
The tail gets longerAmazon and others now do print on demand. Thisminimizes the demand needed to keep the book available,and it lets each purchaser choose the typefont or the paperstock. This is similar to having many more different booksfor sale.Similarly we now have vast numbers of personal videosbeing made that would never have found a broadcastoutlet.
What is the impact?More choice; you can read, watch, or listen to the specificthings you want.More junk out there. (although whether the TV networksare better or worse than home-made cat videos isarguable).More stuff is free. The Web lets you easily choose to reada free book or listen to free music.
Again, there are naysayersPublishers subsidize good books with the trashy best-sellers; if every book is being sold separately, thesesubsidies go away, and the good books may not bepublished.Except that most publishers don’t deliberately publishbooks that they think will lose money. Although they dosubsidize books, it’s not intentional. So whatever theinvestment needed to publish, it would still be provided(assuming publishers continued to guess that the bookwould sell).
Advertising?96% of web advertising goes to 50 sites, mostly Google,AOL, MSN and Yahoo (Richard Eldeman). So isadvertising not a long tail example?Half of the ad revenues get redistributed to other sitesunder programs like “AdSense”. So in fact, lots of sites geta tiny amount of ad revenue. (AdSense is a program bywhich Google puts ads on other sites and shares therevenue; imagine that the search-targeting ads appear notjust on the search results page, but also on the pagespointed to.By the way, each dollar of advertising that newspapers arelosing is turning into 33 cents of revenue online; i.e. onlineadvertising is perceived to be more efficient.
Social cohesionOnce upon a time everybody in college was taught thesame books, and they were standard literary works. Untilafter World War II it was common for college students toknow Latin and have read standard texts (Caesar, Livy, …)and in Britain they would also have known Greek.And recently everybody would have known the same TVshows. And followed the same news stories.Will it be a problem if there are no common social threads?Would our society fragment?
Niches, tribesAnderson says we’ll have social cohesion, just in smallergroups. He says there will be more interaction withindividuals following genuinely shared interests than justfollowing the mass market.The opponents say the “long tail” encourages individualismor “egocasting”.Your call.
Culture and societySo would the Civil War have been avoided if all 19thcentury Americans had watched the same TV shows?Seems doubtful. There was no Internet in either the 1950sor 1960s, and one was conformist and the otherfragmented.If one looks at some measures of social disorder such ascrime, strikes, or votes for 3rd parties, it would seem that inthe last ten years they have declined.