Tumbled Logic

A ragtag blog filled with random technical nuggets, rants, raves, occasional pretty pictures, and links to things.

Mar 18

About that Internet piracy study…

Yesterday, Richard Wray wrote up a piece in the Guardian on a study which has been backed by the TUC and claims that by 2015, losses from piracy will reach £218bn and put 1.2 million jobs in peril.

Let’s be clear about something from the outset: my purpose here isn’t to make the case that copyright infringement is fine, or even one of those mythical “victimless crimes”, however these numbers immediately struck me as being a bit on the high side.

I took a look at the report itself, catchily-titled “Building a Digital Economy: The Importance of Saving Jobs In The EU’s Creative Industries”, to try to find out what the methodology behind and basis for the headline figures were. The good news is that this was easy to find. The bad news, on the other hand, was that it didn’t inspire much confidence in the estimates.

The methodology is basically this: take estimates of current piracy levels, multiply down by a “substitution factor” (i.e., what proportion of downloads would otherwise have been sales), and scale the whole thing up to form future estimates based on anticipated improvements to speed and penetration of broadband Internet access.

As methodologies go, this isn’t at all terrible. In fact, it’s the only reasonable way of doing it. The devils in the details are twofold: everything hinges on the accuracy of both the current figures on infringement levels and the substitution factor.

For digital music piracy, a “conservative” figure of 10% was chosen. That is, for every ten tracks downloaded, one counts as a lost sale. Far from being conservative, this strikes me as being awfully high given anecdotal evidence regarding download patterns. Take a look at any BitTorrent site and you’ll be hard pressed to find a download of a single song. Thus, as BitTorrent’s popularity has risen, there’s been a shift towards downloading albums — or even back-catalogues — rather than lone tracks. In other words, even if you only wanted to sample one song, you’d have to download an album at least. If albums contained an average of ten tracks each, this would mean that every single album download (increasingly the predominant distribution format thanks to BitTorrent’s bias towards sharing large files) would translate to a lost sale of a song based on a substitution rate of 10%. As whole-album downloads become more common, this causes a bias of the effective substitution rate up towards 100%, which is clearly nonsensical.

Therefore, as usage patterns have changed, the substitution rate needs to change to account for it. If people when looking to download a single track are more likely than not to download the whole of the album that it features on (because it’s easier to do this than not), then the substitution rate needs to be adjusted downwards to account for this. In terms of convenience, it’s often easier to download a whole album now than it was to download a single track in 2003. If behaviour shifts towards downloading multi-album archives (as we’ve started to see), the same applies again. If the substitution rate doesn’t change to account for this, you will massively over-estimate the losses resulting from piracy.

As an educated guess, I would put the real substitution rate for 2008 somewhere around 0.5%, and that’s leaning towards greater losses to a fair extent. Others may have different (and more robust) estimates, of course, and I’d be interested to hear of solid research in this area.

The other piece of information — the infringement figures which are used as the basis for everything else — is even more problematic. Appendix 1 in the report details the figures and how they were obtained. In the case of the UK, digital piracy of music is pegged at 1,177M “infringements per year”, and the figure is attributed to “BPI”. The BPI’s reports on piracy, however, are only accessible to members.

What the BPI does publish (repeatedly on its site) is the figure of “some 7.3 million people engaged in unlawful filesharing”, according to Jupiter Research. Assuming these two figures — the number of people sharing and the amount of infringement taking place — are supposed to be consistent with one another, this leaves us with a few problems.

That 7.3m figure was investigated by BBC Radio 4’s “More or Less” programme, and the results were written up by Ars Technica. To quote from the article:

The number was quoted in a recent government report, but it’s not a government number; it turns out that the government commissioned a report from the CIBER research group at University College London, which contained the number. CIBER’s report cited the number four times, noting that it came from yet another report from consultancy Forrester.

Still with me? Get ready to go down the rabbit hole, because it’s here that things get weirder. The Forrester report in question does not in fact contain the “seven million” number, despite the CIBER citation. The number actually comes from a separate piece of research called the Jupiter Industry Losses Project, which attempted to quantify losses for the recording industry due to things like P2P usage. And who paid for the Industry Losses Project? The British recording industry, of course.

BPI, which represents the major labels, wouldn’t turn over the complete Industry Losses Report to Radio 4, but it did supply some of the numbers from the piece. In addition, Radio 4 talked to the author of the report, Mark Mulligan.

The report estimates that there are 6.7 million illegal file-sharers in the UK, a number that was generated by multiplying two other numbers: the total number of Internet users in the UK and the percentage of the population engaged in file-sharing.

Both numbers used in this calculation turn out to be controversial. The UK government, for instance, says that 33.9 million people are online, while the Industry Losses Report said that around 40 million were online. As for the estimate of the piracy percentage, that comes from a 2008 survey of 1,176 UK households. The survey actually found that 11.6 percent of respondents admitted to using file-sharing software, but Mulligan adjusted this upwards to 16.3 percent to account for “underreporting” (i.e., the fact that some people were lying).

The differences in all these numbers are tremendous. If the lower numbers are used instead (33.9 million and 11.6 percent), it turns out that only 3.93 million UK residents are dirty pirates—a mere 60 percent of the original 6.7 million number.

The bottom line here is that the 7.3m figure is essentially meaningless; it’s based on a survey of just over a thousand households and then multiplied up in the same way that the BASCAP report does in its predictions. If — and there’s potentially some wiggle-room here — this (still publicised today) 7.3m figure is related to the 1.1bn “infringements” figure, then it renders the UK music part of the BASCAP report as worthless as an educated guess by a journalist. If this is the quality of the data across the board, then the entire report has little merit at all from a analytics perspective.

This is most of what I’ve managed to turn up after an hour or so of digging. There are a few other incidentals, though: job losses in the “creative industries” could reach a quarter of a million, yet according to UK Enterprise, the music industry employs 130,000. Does the rest of the creative industries really have much more than double that in total? If it’s less than double, there’s a risk that the job losses could get so bad that they transcend the total number of people in the industry!

The timing of report seems a little dubious. As the ORG and 38Degrees are asking people to write to their MPs to demand debate on the Digital Economy Bill (which given it’s such a wide-ranging bill really should happen anyway), this report is released and FAST is sending out releases to its members:

John Lovelock, Chief Executive of FAST, said:

"The software industry and other copyright holders in the wider creative industry welcome the attention that is being focused on copyright issues in this digital age. We urge clear action against copyright infringers using the Bill’s provisions after Royal Assent.

Delaying the legislation unnecessarily in the Commons will not help. The Bill has had scrutiny in the Lords to excess. FAST has lobbied for intellectual property legislation since its inception in 1984.

Whilst campaigners have rightly pointed out that we shouldn’t allow politicians the power to meddle unrestrained with our copyright regime, let us be clear, there are checks and balances in place being eminently sensible given the rapid pace in digital technology and the difficulties in protecting intellectual property.”

All in all, it’s looking like rather convenient timing.

Let us not forget also that some research has suggested that losses due to piracy might actually be a negative value (i.e., sales increase), nor that 2009 saw record singles sales in the UK. Back in July, we heard that the music industry as a whole had grown 4.7%. Internationally, meanwhile, sales fell 10%.

It’s also interesting that while in other industries falling revenues are attributed to the “global economic downturn”, in creative industries there’s a tendency towards pointing the finger of blame at piracy. Charles Arthur has an interesting take.

From the various figures and reports, it’s clear that the music industry is in turmoil. What’s not clear is why and how much of it is anything but evolution. The only figures we have on negative external impacts (i.e., piracy) are dubious, to say the least, and in the UK at least the sector overall is doing well despite being in the midst of a recession.

I don’t doubt for a minute that some proportion of illegally downloaded materials translate to lost sales, but I’d love to know what the real figures are. How do they compare to lost sales through resale in charity shops (admittedly a rather flippant question, but something to ponder on)? On that note, it’s interesting that although the price of “digital” sales has started to fall, they’re still quite expensive compared to CDs, despite having very low distribution costs, no packaging costs, and — crucially — zero resale value.

Food for thought.


  1. furiousmoose reblogged this from nevali
  2. nevali posted this