Fast, Cheap And Out Of Control

By london_ken Last edited 163 months ago
Fast, Cheap And Out Of Control

Sandy Pearlman - Purcell Room, South Bank Centre - 23rd June

Sandy Pearlman's CV affords him a certain authority when he speaks about his vision of the future state of the music industry. A producer and songwriter of some note, he worked with Blue Oyster Cult and The Clash. Spotting the potential of online music he founded, a downloading company that was probably too far ahead of its time, waiting for the technology to catch up with the idea. Nowadays, Pearlman is part of the Future of Music think tank, and is currently promoting his ideas on "the total re-architecting of the music business".

Pearlman's latest speaking engagement took place to a smattering of people (of which at least two-thirds looked to be hacks, the other third being made up of music industry workers) at the Purcell Room, as one of the events at Patti Smith's Meltdown Festival.

The gist of Pearlman's argument is that the proliferation of digital distribution networks means that the current model used by the music industry is doomed to fail, or at least it is doomed to fail the artists. He says the rise of the paid-for download sector is an anomaly and argues that the ratio of paid-for downloads against the huge number of illegal downloads shows how insignificant the current paid-for strucure (read iTunes) is. Pearlman's vision of the future is based on two premises:

1. The creation and maintenance of a 'recommendation engine', which will, based on a 'user' stating a preference for a band or artist, spit out a whole host of recommendations that are guaranteed to hit the right spot.

2. The creation and maintenance of zero- (or close to zero-) cost transaction servers which will allow the price of downloads to be set at 5 US cents per track.

The second premise (we'll get to the first one momentarily) is a question of economics, and anyone with a passing understanding of elasticity of demand will recognise the argument. (For those not so versed in this, elasticity of demand is, in essence, the idea that whilst any monkey can sell lots of things very cheaply, you don't necessarily get more in total revenue from it; sometimes you are better off pricing your goods a bit higher even though you sell less items. Conversely, you might think you're doing well selling things at a high price, but if you lower the price, you'd actually sell so much more that your total revenue is higher. (Wikipedia explains better and in more detail.)

Pearlman is arguing the latter point; to tap into the billions of not-paid-for downloads, you set a price that is close enough to zero that the downloaders will think, "Oh, ok, I don't mind paying that for a high-quality download; it's the full price that makes me do this naughty thing and get free lower-quality downloads." 5 cents times several billion is more than, to take the iTunes price, 79p times several million. A simple argument that can only be proven or disproven if and when it's put into practice.

The underlying assumption, of course, is that the record companies will understand Pearlman's argument and buy into it. (That's leaving aside the argument over whether consumers will pay 5 cents per track if they can still torrent the tracks for free.) Pearlman argues that the record companies have no choice, in fact, and he can see a future when the technology companies like Apple, Google, Yahoo and Amazon will eventually swallow up the record companies; these companies being premier examplars because of they are ahead of the game when it comes to a zero-cost transaction infrastructure necessary to sell works of music so cheaply. (Apple Computers are, of course, in seemingly-constant litigation battles with Apple Records regarding their rights in the music business, but that's nothing that hard cash can't solve in the long run.)

Taking the 'recommendation engine' then. Necessarily, given time constraints, Pearlman didn't go into huge details about the technology he's working with, save to say he thinks it's far better and more extensive than anything done before. He believes it will help music lovers discover music they never thought about trying before. This vision of the recommendation engine being the driver behind the discovery of new music is somewhat alarming and poses the following questions:

1. The immediate issue arising is the immense concentration of power, should this recommendation engine prove to be the gateway to these 5 cent tracks. Who is the gatekeeper and who polices the way the recommendation engine operates?

2. The recommendation engine will need to pay for itself somehow. By Pearlman's admission, vast amounts of money, far more than anyone else has invested, has gone into this technology. One assumes that whilst Pearlman talks of giving more money to the artists, this technology is not going to be run altruistically, and will need some sort of payback. Either it creams off a percentage of the price paid or it charges for entry into the database. One assumes the former option will be taken or else the database is flawed through incompleteness. If this is true, it's hard not to think that the promotion of the recommendation engine is more about making bundles of cash - and putting the record companies out of business while they're at it - than opening up musical vistas.

3. To counter the fears that there will be one omnipotent recommendation engine, Pearlman says that there will be nothing to stop the data being made available to other parties for analysis, to enable them to build their own recommendation engines. Again, the question of finance comes into play: presumably the data will not be given away for free, given the cost of collation? Then there is the question of bias. If there is no one authoritative source of recommendations, music lovers then have to spend time and money finding out which recommendation engine most closely matches their own tastes. This doesn't seem a million miles from the current situation of people finding a reviewer or publication they trust, and going with those recommendations.

4. Pearlman argues that the data does not show bias, yet it is based on the responses to surveys; surveys compiled by humans, completed by humans and analysed by humans. Somewhere along the line, even if it's the most perfect survey ever compiled, there will be a bias introduced which will dirty the end data. Pearlman places great faith in the algorithms used to overcome bias but there will always be someone clever enough (or a mole unscupulous enough) to crack the algorithms in order to skew results. It happens to Google, it'll happen to the recommendation engine(s).

5. Pearlman, at one stage during the talk, dismissed the likes of Amazon's 'if you liked that, you'll like this' and 'other people bought this' technology. Later on, he spoke about incorporating all of that kind of data into the recommendation engine, including reviews in publications (print and online), to help people make their choice. This somewhat dilutes the argument for the recommendation engine by introducing bias and the limited sphere of knowledge that is a human brain. Pearlman seems to be trying to have his 'infallible computer' cake and eat the 'human touch' cake.

Pearlman's talk was thought-provoking but it did raise more questions than it gave answers. (Something not helped by Pearlman's habit of answering a different question to the ones posed during the Q&A session.)

That the music industry needs somehow to deal with the new economics of the digital age is obvious to anyone with a vague grasp of digital distribution. That the recommendation engine Pearlman advocates is the right driver for change is open to debate.

Last Updated 24 June 2005