Drake IP 2013: Paul Heald, “How copyright makes works disappear, and the tolerance of infringement on Youtube”

Liveblogged from Drake IP Roundtable 2013.

Paul Heald discusses two new empirical projects:

How copyright makes book disappear

Start with a sample of 7000 fiction books on Amazon, then take 2500 titles located in the Library of Congress catalog. It turns out that there is a big drop off in books published in 1930-1990. Many books published pre 1927 public domain cut off, and many books in the last decade. There are three times more books available through Amazon in 1850s than 1950s, despite rapid growth in the number of books published.

From a sample of 581 songs in top 100 grossing movies, most songs are made contemporaneously with the movie. The proportion declines rapidly through to -70 years ago, then a slight bump at the public domain mark. When movies are chosen at random, rather than top 100 grossing movies, there are many less public domain songs. Several hypotheses about why this might be; but Heald suspects that this is because the median release date is much more recent than for the top-100 movies. So, because of copyright term extensions, public domain songs are further behind. The hypothesis is then that movies pre-1976 would use more public domain songs than post-1976 movies, and this turns out to be the case.

Study 2: Uploading hits on YouTube

Charted Number One songs in Brazil, France, US from 1930-2000. Searched for each song on YouRube and charted the first 10 uploads containing a version of the song. Asks why many songs are not monetized; A Whiter Shade of Pale has 12M views, uptime of 3 years; it’s not taken down, and it’s not monetized. Over two-thirds of Number 1 US songs 1930-67 are monetized – but a large proportion are not. Median downloads of non-monetized songs appear roughly similar to median downloads of monetized songs. The most common form of non-monetized songs are television and film performances. By contrast, monetized songs are dominated by simple recordings accompanied by an album cover.

Heald suggests that this might be because

  • Youtube requires copyright owners to show that they own all the rights; not just song, but the film clip etc. So it is much easier to monetise bare recorded songs. This would be exceptionally difficult for television clips from the 1940s and 1950s, where registration records are not generally available.

  • Composers may tolerate a lot of infringement in the context of television and movie clips. Composers can’t monetise these clips because they don’t have all the rights; but the do have a right to take them down. For some reason, perhaps because the video adds to the popularity of the work, composers don’t generally take these down.

  • Starting in the 1970s, when contracts become more formal, all the rights in a music video may be allocated to a single owner, and therefore it might be easier to monetise these clips.

Tentative conclusions

  • Non-owners play a very important role in maintaing the availability of public domain copyright works

  • Traditional book and music publishing models demonstrate how copyright stands in between works and the consuming public

  • New models, like YouTube, let the market interact with the owner, providing valuable information, and facilitating both availabiliy to consumers and revenue streams to owners.