IFPI blog - new paper suggesting pirate downloads help sales is flawed

25th January 2016

David Price, Head of Anti-Piracy Research and Analysis at IFPI, says a new paper from Queens' University doesn't stand up to scrutiny.

A new working paper from a doctoral student at Queens' University in Canada suggests that pirate downloads can help increase sales as much as substitute them. The paper attempts to make a link between pirated downloads on a private bittorrent tracker with album sales. Unfortunately, looked at more closely, the paper's approach and its conclusions are significantly flawed. Its conclusions are not supported by the evidence, the sample used in the research cannot be generalised to the full market, and the paper as a whole is too dependent on out of date data which invalidates its conclusions. The research tracked downloads of 2,251 albums over half a year on a single music-focused private bittorrent tracker in 2008. The researcher then attempts to correlate downloads on the private tracker with album sales in the US. His working hypothesis is that albums that were heavily downloaded might see lower sales because of a higher availability in file sharing networks - or that they might see higher sales because of a "word of mouth" effect. Analysis found that an increase in album downloads correlated to a drop of (just) 0.17% in physical album sales but a 0.21% increase in digital sales. The researcher speaks about this correlation as if file sharing was causing this effect ("I conclude that file sharing activity has a statistically significant but economically modest negative effect on legitimate music sales"), taking the activity on this single private tracker as somehow representative of all file sharing activity. The first point to note is that the data is from 2008, when the music distribution landscape - particularly on the digital side - was very different to that found today. All-you-can-eat streaming services were still embryonic and completely absent from the United States. Beyond this difficulty in taking seven-year old results as still applying to the rapidly evolving music market today, the origin of the data and its applicability to the real world is a major cause of concern.

The world of private trackers is far removed from day to day file sharing consumption. This was even more the case in 2008. It is not possible (or correct) to generalise from such a tight-knit, closed, musically fastidious community to the file sharing world in general and to the wider population. For example:

  • Private music trackers attract a particular type of music consumer, typically those who have a strong interest in particular types of music, and niche tastes. Such users do not represent the typical consumer in terms of buying habits and they do not represent the typical consumer in terms of musical taste.

  • The private tracker in question has members from across the world but the sales data came only from the US - and music tastes are different in different countries.

  • Private trackers have ratio rules: over time, you must seed (upload) as much as you download. This affects behaviour in some subtle and some not-so-subtle ways. For instance, users may choose to download a popular album simply in order to seed it back to other users, or not download at all if their ratio is particularly poor.

  • While bittorrent may be the dominant form of music file sharing today, in 2008 Limewire was extremely popular for music downloading, particularly in the United States, and there would be little crossover between users of private trackers and users of Limewire.

In summary, the findings cannot be generalised, there is no common sense link between the two sets of data, and the paper's overall approach and conclusions are therefore significantly flawed.

Link to original paper

For further information please contact

Adrian Strain, IFPI
Email: adrian.strain@ifpi.org

Tel: +44 (0)20 7878 7939