Remix vs Shuffle

One of the key memes in free culture has been the remix. Freely licensed content combined to make something new. Remixing of books is of obvious interest and there have been many explorations of this in various forms including the Rice University project Connexions. FLOSS Manuals has had the ability to take existing manuals about free software and remix it since 2008. Its easy to use this mechanism to add a chapter or chapters of a manual with other chapters from other manuals. The output is templated HTML or customised PDF. Although the remix feature (its very easy to use with a nice drag and drop process) always gets very positive responses when demonstrated it does not get used very much.

After some years thinking about this I have come to the understanding the ‘remix’ as such has only a limited use case when it comes to constructing books from multiple sources. A remix of a book in this fashion is not a good remix as we might understand it. DJs make good mixes out of several tracks but they have various tempo, tone, and volume controls to integrate the sources. A book remixed by the (out moded) remixing tool we built is not like a music remix but more like playing back selected tracks using shuffle. The chapters are not integrated to flow well into each other, they are instead compiled into some kind of anthology.

The difference is not subtle and it's easy to understand the problem when you look at the obvious popular example of remixes in DJ culture. A DJ takes multiple sources – some complete – some snippets – and works them into a continuous whole. It is part curatorial process and part production. The curatorial process is the choosing of the works and considering where and when the selected pieces will fit into the whole. The production process is changing the tone, speed, and color of the sound and making it all work together. Without the production component its not a remix at all – its just a shuffle of sound snippets.

Text requires the same kind of shaping. If you take a chapter from one book and then put it next to another chapter from another book you do not have a book – you have two adjacent chapters. You need to work to make them fit together. Working material like this is not just a matter of ‘cross fading’ from one to the other by smoothing out the requisite intros and outros (although this makes a big difference in itself) but there are other aspects to consider – tonal, tempo, texture, language used, point of view, voice etc as well as some more mundane mechanical issues. What, for example, do you do with a chapter that makes reference to other chapters in the book it originated from? You need to change these references and other mechanics as well as take care of the more tonal components of the text.

This is why remixing in itself is not that interesting and also another argument why some free licenses should be banned for free book production. An ND license (non-derivative) renders a ’free’ work useless for combining with other works. You can separate it from its original corpus but you cannot make it fit easily within a new one. You have no licensed right even to change the mechanical components. You cannot create chapters that will smoothly exit one book and enter another. You actually have to produce the mixed material to make it all work together – there is not really much point to trying to avoid this issue.

As a consequence of these experiences we designed Bookspark to enable importing of chapters from one book to the editable environment of another book and we ditched the old remix approach. This means you can import chapters from other books and then edit the chapters to make them fit the context. That is the only sensible way we can work with this kind of re-use/remix.

Produced somewhere around 2009-2011