« The Conservation Law of Transparency (you can't be open in all things all the time) | Main | Guest post: Acting as a non-monetary economy »

December 23, 2008



"Because Will hasn’t published the data or said where it came from, we can’t really know what it means."

80 million sales.

I think we know what it means, Chris: you've failed.



I was reading the TIDBITS on the right side and found them interesting as you highlight stuff about your next book (i know my post has nothing to do with this article).

I thought you might be interested in adding something about the Free Zones. Here in Dubai, we have a lot of Free Zones like Dubai Internet City, Dubai Media City, Dubai International Financial Center ..etc

The concept is to attract International companies from different sectors to open regional offices in Dubai without charging them taxes, and then benefit from the business and money circulation those companies produce.

I take it as providing the companies a FREE business environment (tax free) and getting their money back through different methods. Do you see this qualify to be an example of FREE the way you present it?


Chris Anderson

@Murshed: That's a great idea. I was just working on the international chapter and will include it there.


Interesting but I still believe that niche markets will always suffer to traditional buying trends.

Apart from those wishing to have or give a non-traditional gift, it is human behavior to maintain the thinking of the masses. 'Sheep' mentality but at the same time we must realize that others tend to frown on brands and products with which they cannot identify.

And let's be honest, we often buy to compare, show and feel like we can have what others have. Buying trends are not going to change anytime soon just because niche products are more easily available via the internet.

Angus Batey


I don't know anything more about the Will Page research than what was published in The Times, but I do know a little about how the music business works in the UK. The MCPS is the Mechanical Copyright Protection Society, and the PRS is the Performing Rights Society - the MCPS/PRS Alliance is precisely what it suggests, an alliance of these two collecting societies. They work on behalf of rights holders, like ASCAP and other collection societies you have in the States, and collect royalties from, respectively, mechanical reproductions of recorded works (CD sales, digital downloads) and public performances of members' compositions (radio airplay, live performance, radios playing background music in coffee shops, etc.).

Given that Page works for these entities, might it not be possible he has used MCPS as his source of data? While not every composer is a member of PRS, every record label that sells through retail outlets in the UK has to be a member of MCPS. If so, it's a fair bet that when he talks about percentages of tracks that don't sell any copies at all, he's drawing on a complete catalogue of all material available from all legitimate online music retailers in the UK. Indeed, it would seem odd for him to allow his and his employers' names to be put to research that did not use their own internal data as its foundation, as his affiliation with them certainly implies as much.

If that is in fact the case - and like I say, while I don't know for certain, it's a fair bet, so, please, humour me - how would you then interpret the data he has collated? If it refers to the entire legitimate UK music download market, not just one retailer, isn't that something of a surprise to you?




Regarding this issue, I agree with Chris Anderson mostly.

I think the most effective way for a poor startup (including artists) is to create a market, not just a product in an existing market. What does it mean to create a market? Create a product that's different enough so that users cannot associate with an existing category.

Chris Anderson

@Angus: If that's the case--that it's a blend of iTunes and other online retailers, some selling singles, some selling subscriptions, possibly including ringtones, etc--then it would be a disaster from a statistical analysis perspective, since each one of those marketplaces has very different characteristics. Let's be generous and assume he didn't make that beginner's mistake, and it is a single online retailer such as iTunes. That would be very interesting and I'd love to actually see the data and find out when it was taken: before iTunes had recommendations, or after, etc. But sadly it seems we'll never know.

Angus Batey

Thanks Chris,

OK, I take your point that we're not sure about what the data represents or where it has come from, so I accept that any speculation on what it means will be of limited value. But then again, this whole blog deals with speculation and theorising to a great extent, so I'm still interested to know what you make of this data *if* we allow ourselves, for the sake of debate, to assume that it refers to the whole of the UK paid-for per-track download market. Here's why.

There are other retailers aside from iTunes who sell tracks individually and don't use a subscription model. And in terms of income to artists, composers and other rights-holders, I'd be happy to include sales of ringtones in the same set of statistics, as ringtone sales represent something analogous to single-track download sales (ie, someone paying for a single digital copy of a single piece of music). As far as a time frame is concerned, the Times report, as woolly as it is, does say that these figures relate to sales made "last year". Let's assume the writer is entirely correct, and isn't jumping the gun as people often do in December and referring to the year we're still in; therefore let's consider that the data covers sales in calendar year 2007 - ie, with iTunes including recommendations.

I appreciate that that's a lot to take on trust, and that if any of those assumptions are off then we're wasting time speculating; but I would be curious to hear how the Long Tail model takees account of the possibility that lots of music available for purchase online isn't purchased at all, and while I respect and understand your reservations about the report, I think it's reasonable to ask you to speculate in that direction.

Regards, and compliments of the season,


Chris Anderson

@Angus. No one in their right mind would try to compare ringtones sales with iTunes sales, so I won't even try. As for recommendations, iTunes didn't add them (the Genius sidebar) until late 2008.

Angus Batey

Thanks Chris,

I hadn't realised that the recommendations iTunes have run ever since the service was launched in the UK weren't the same ones you've been referring to in your writing on this topic. Presumably this means that when you wrote your book, iTunes didn't have a sufficiently effective recommendation engine operating to substantiate your theory, which I guess only makes your work all the more prescient.

Am still curious as to what this may say about the Long Tail theory if we assume (as Benoit Felten seems to infer) that it relates solely to iTunes sales of single tracks. Is there any chance you may be willing to speculate on that basis?



Chris Anderson

@Angus: I have no idea. I try not to speculate, and prefer to work with hard data. I've never had iTunes data, so I only analyzed the datasets I did have: Rhapsody (streaming plus track sales), Netflix, Amazon, SourceForge software, search, social networks, etc. All followed powerlaw distributions for at least half of their "inventory" (as I've written before, all filters tend to break down in the deep tail, where the network effects are thin. You can often see a powerlaw fit the top half of the curve (straight line on a log-log scale) and then a lognormal fit the remainder better.)

I'm a regular iTunes user and I don't remember any functioning recommendations before the Genius sidebar. What are you referring to?

Angus Batey

Thanks again Chris.

It does occur to me that if the research was based on iTunes, we may never be told that, because Apple's antipathy towards answering any questions that involve actual numbers is legendary, and if Page did get hold of their figures via the MCPS back door, he may not be at liberty to say that that's what he's done.

Anyway, we still don't have a Genius sidebar on iTunes UK - only the very basic recommendations we've had since the service launched: "Listeners Also Bought" and suchlike. So in that sense, regardless of the period the research covers, that's likely to always have been the way. (Unless, of course, I'm using an out-of-date iTunes and thus missing out on the Genius sidebar, which is entirely possible as I have an older Mac which can't run OSX 10.5, which means I'm not able to use the most recent versions of many apps).

I think the distinction between a "deep" tail and a "long" tail is important - I did read your book, though some time ago, so apologies if I'd not kept up to speed on that aspect. I think that would explain how your theory would work in this instance, if the figures do turn out to be reliable.

What I feel this suggests, restricting my thinking solely to recorded music, is that some level of marketing and promotion is essential in order to draw even tiny amounts of attention to the least popular items in an online/"infinite" inventory. Recommendations of the type used by Amazon and iTunes will only work to a limited degree, whether directly user-generated or purely algoritmic. To over-simplify, even if iTunes only stocked two albums - Pet Sounds and The White Album - it's unlikely that, say, Revolution No 9 would be flagged up in an "if you like this, try..." way, because it isn't anything like any of the other tracks on offer. With over a million tracks, there are bound to be some that won't ever get recommended by a computer program; and if no actual user has heard a track, they're not going to recommend it to anyone else. So the tracks that haven't been purchased at all are unlikely to be purchased until someone takes a chance on them, which they won't do, usually, unless they maybe read a review of them, or hear/see them played somewhere, or have them drawn to their attention in some other way, such as via advertising. This suggests a continued need for some aspects of the "starmaking" machinery, which in turn means there's still going to be a need, if not for blockbusters, then at least for significant successes, to allow the people making them to have sufficient resources spare to help draw attention to their less popular/newer wares.



Salvador Laddaga

Dear Chirs,

I went to a conference that you made in Mexico City on October 08, I’ve been following your blog ever since, the reason i write it's that i want to ask for your help, there was this slide in your keynote where you explain how blogs and independent sites are more or equally important as a traffic source, it was the slide were you end up talking about Adam Fuhrer, could you share that info with me?

Thnx and keep on the good work.
Greetings from Mexico

Sanne Roemen

This is all still about music sales. That's ok but I think the long tail works a bit differently for niche music markets. It's not the songs or albums that sell better, but the bands themselves that reach a bigger audience. Wigh pays of in playing more live shows, in other than their own region where they sell their independently recorded music and merchandise and maybe get a decent fee or some of the revenue from ticket sales. I don't have hard data for this development either, just a gut feeling and a strong wish for someone to do some research into a broader spectrum of ways to 'sell'.


Please take a look to this article! thank u



I just wanted to comment on Genius acting as a filter (as defined in the Long Tail). It doesn't. Unless I'm totally mistaken as a Genius user, Genius uses iTunes Store data to build set lists based on the music you own, but it does not act as a recommendation engine for purchase.

This is one further argument to express that using the iTunes store as a basis to demonstrate anything about the Long Tail is absurd. If you combine the absence of filters, the price fixing and the fact that there no margin consideration in iTunes, it's an increadibly bad choice of dataset.

If this is indeed Page's dataset, the only thing his analysis demonstrates is that the Long Tail phenomenon does not apply automatically to any online marketplace regardless of how it's set up. And anyone who's read Chris' book would already know that...

Jesselli Couture

That's a great idea. Thanks for shairing this info.

Designer Handbags


I think the iTunes Soundflavor DJ add-on may represent a long tail.

Suzie 100

Everyone has a slightly different opinion on the long tail sites and what it entails.


yus What I feel this suggests, restricting my thinking solely to recorded music, is that some level of marketing and promotion is essential in order to draw even tiny amounts


That's a good concept. Thanks for sharing this information.
Free Book Reviews

adaptateurs secteur

This post gives me the perfect scenario of the Long Tail.. In above figure the Long Tail is going to down & down.. Thank you so much for sharing this & also for update my knowledge for the long tail especially, i really appreciate this post..

Buy Viagra

hey friend excellent post about Annual “cutting the Long Tail down to size” roundup thanks for sharing!!! this blog is very interesting I love it

The comments to this entry are closed.


The Long Tail by Chris Anderson

Notes and sources for the book

FREE was available in all digital forms--ebook, web book, and audiobook--for free shortly after the hardcover was published on July 7th. The ebook and web book were free for a limited time and limited to certain geographic regions as determined by each national publisher; the unabridged MP3 audiobook (get zip file here) will remain free forever, available in all regions.

Order the hardcover now!