Which text is real?

Matthew Kirschenbaum has an interest post on the nature of the codex in a digital world. Once it's been printed to paper it's much like a codex in the world of hand-set lead type, say. But before that it's a bunch of files, and that has implications, which is what Matt's post is about.

In the course os spelling out some these implications Matt offers this anecdote:

And make no mistake, the challenges are real. Last year our colleague Martin Paul Eve received considerable attention for his revelation that there are some rather startling textual differences between the US and the UK editions of David Mitchell’s bestseller The Cloud Atlas. The explanation, which Eve received in an email from Mitchell himself, turns out to be unremarkable and will not surprise anyone familiar with the social dimensions of textual production: the manuscript had had two different editors at two different times in the New York and London offices of two different publishers, and no one had ever bothered to reconcile the changes, certainly not Mitchell himself, who had other things (like writing his next novel) on his mind. But the variants now immortalized in print (and in pixels) have genuine implications for how the text is read and interpreted, in ways Eve goes on to convincingly detail.

It’s nice that David Mitchell is still with us and was willing to answer a professor’s email. The author is not dead, yet. But once the author is, what recourse would a scholar such as Eve have had? Absent the author or some other individual with firsthand knowledge of the situation, a bibliographical explanation for the variants could only have been arrived at through an examination of the material evidence, which would have to include the digital files the different editions of the book were set and printed from. And where are those files? Well you might ask; they are perhaps in someone’s cloud somewhere, but no atlas exists to help us find them.