'Long Tail' May Not Wag the Web Just Yet

Carl Bialik from WSJ writes "Expanding on an article he wrote in 2004 (and discussed on Slashdot), Wired magazine editor Chris Anderson argues in his best-seller 'The Long Tail' that the web is changing commerce from a hit-driven business to one focused on niches. But Wall Street Journal columnist Lee Gomes questions Anderson's data, and adds, 'I don't think things are changing as much as he does.' Gomes writes, 'At Apple's iTunes, one person who has seen the data -- which Apple doesn't disclose -- said sales "closely track Billboard. It's a hits business. The data tend to refute 'The Long Tail.' " ' On his blog, Anderson responds that Gomes 'stumbles over statistics and more, and in the end simply makes a muddle of what might have been an interesting debate over the magnitude of the Long Tail effect.'"

missing the point

[blamanj]

The point of the "long tail" isn't that the sales stop being weighted towards the "hits". You still have a power-law distribution, that doesn't change. The point is that today, the distribution is artificially restricted, because items are dropped from availability. When you extend the number of products available, you make more money from the "tail" products and the portion that came from the original set of hits is smaller.

Flawed Argument

[twofidyKidd]

Citing the iTunes store as representative of anything but a "Hits Business" is flawed. I think consumers who represent the statistics in the long tail don't shop at the iTunes store. While I know it's not a vaild argument to cite what my own purchasing practices are, I for one spend a lot less on music at iTunes and more at places like Om Records [om-records.com], Defected [defected.com], and other independent label online stores. In fact, if I do purchase at iTunes, it's usually a very popular song which is consistent with iTunes being in the "Hits Business". The arguer is right about that, but wrong about who it accurately represents.

Re:Duh

[Brickwall]

"How do you advertise the other ~90% of your catalog?"

The following snips are from Slate's recent article on the long tail:

"At Slate, our inventory is our articles. We publish 20 or so stories every weekday, but we also have a backlog of about 33,000 pieces in our archives. Because those stories are freely available to our readers, a chunk of our traffic each day comes not from our "hits"--current pieces that are promoted on our home page, which typically draw tens of thousands of readers--but from older pieces with narrower appeal. "

"Why did the piece pull in such consistent numbers? Google. When readers type "Girls Gone Wild" into Google's search box--seeking intellectual succor, no doubt--Levy's piece is the fourth hit. The story here, of course, is not that Levy's dispatches got 2,354 hits last Tuesday; Slate as a whole pulled in about 1.9 million hits that day. The story is that that traffic was free--we paid for the piece years ago, and we didn't expend any additional man-hours last week assigning, editing, or producing it. That means Levy's dispatches provided 2,354 chances for our advertisers to reach our readers--and pay us for the privilege of doing so--without costing us a thing. " Whole article is here: http://www.slate.com/id/2146301/ [slate.com]

Is that meaningful?

[yankpop]

So what? How many copies do you need to sell to get an Amazon rank of 500,000? How many for 1,000,000? I'm guessing that to qualify for a rank of one millionth you'd have to sell somewhere in the 0-1 copies range. So if two people bought your book in the last week you might 'spike' to 500,000th, then drop right back down until your next sale in 2020. Hardly a compelling argument to support the importance of the long tail.

yp.

Reality is in between

[ursabear]

I think the whole discussion about the long tail is interesting and worthwhile.

I think the truth between the two discussions (Anderson vs. Gomes) is more likely to be something in the middle, not at one extreme or the other. I don't think hits are going away, and I think hits make their mark on most any marketable thing/meta-thing. With that said, I think that niches are more meaningful and valuable than ever before.

The book example is great - I get more access to niche publications because of the long tail concept, largely because of funding and popularity of hits. Said a little differently: the niche stuff generally sits alongside the hits, and generally benefit from some of the hits' halos.

My music isn't hit music. That's OK, it's just stuff made from my soul, and I am not planning to quit my day job. The money made from niche availability on the internet (for me) fund dinners out, an occasional instrument upgrade, or a small household bill or two. Why is the long tail beneficial to me? Because when someone is browsing James Blunt [numberonemusic.com], they'll often see me on the front page in a promo, and sometimes (well, briefly) listen to my stuff too. Similarly, iTunes/Rhapsody/Emusic/Yahoo! Music/etc. browsers often buy the latest hits, but will splurge on a Jimmy Bear tune or two - how do they find my tunes? Because my niche music is available with the hits, and because searches sometimes come up with one of my funky little musings.

My point is, that niche stuff isn't taking over the world, and hits aren't all there is. I think the niche markets of the world have been greatly enhanced by Internet access, and that they also benefit from proximity to the hits.

Re:How strange

[Eivind]

If you're storing a million songs online, then frankly, $10.000/month is a very small cost. If you sell these songs for an average of $0.69 then you need to, on the average, sell one copy of a song every 5 years or so to break even.

It's true, songs that sell less than 1 copy every 5 years don't turn a profit, but that's still very different from a record store. Try stacking a record-store with all songs that sell atleast 1 copy ever 5 years....

Besides, sysadmin-costs and so on do not go scale O(n) with the number of records stored. In other words, if you are already storing 100.000 songs that sell well, then adding a million songs that sell less will *NOT* multiply your costs by 10. It'll be more like a factor of 2.

Raw discspace is ridicolously cheap. About $1/GB *including* RAID, powersupply etc. (i.e. you can build a fileserver holding 3 terabytes in a RAID and pay $3000 for it) Given that the average song is on the order of 5MB, that means your 1penny/song budget is double the hardware-cost. Migth be realistic, when you factor in backups, sysadmins etc.

The storage is only going to get increasingly trivial too. Today storage costs on the order of $1/GB, so the storage of this million songs costs on the order of $5000. If capacity/cost continue to double every 18 months, then in 10 years the cost will be about $50. In other words, the entire collection will fit trivially in a home-computer bougth at Wal-Mart for $399.