Reproduction and the value of noise

I’ve just been reading David McKitterick’s Old Books and New Technologies and something he says on page 15 (clearly, I’ve not gotten far yet), has made me pause. He is writing about the reproduction of early texts done by the ProQuest (then known as University Microfilms), in which the pages were copied onto microfilm.

“Not surprisingly, quality varied, of the film when it was originated in libraries across mostly Britain and North America, through to the various copying processes, right through to delivery on screen, where soiled lenses and scratched film were commonplace. Yet, understandably, researchers welcomed these surrogates, imperfect though they were.”

McKitterick is writing about the problems of remediation, of the unavoidable standardization and loss of individual detail and character that occurs in modern digitization projects. The early microfilm reproductions managed to avoid the issue because they could not help but emphasize the distinction between the original and the copy. The reproduction process, which was analog and not digital, ensured that noise entered the system frequently, appearing as scratches, fades and other artefacts which would be considered by the reproducer as errors or faults. However, no one looking at these faulty reproductions would ever be tempted to confuse them with the original. At best, viewers might be grateful that they had a glimpse of the source document, perhaps an accurate enough glimpse to develop theories or approach conclusions, but always with an obvious need to return to the source for an authoritative study.

How would this book fit in your hand? (Book of hours, Paris c. 1410. Miniature of the Annunciation, with the start of Matins in the Little Office, the beginning of the texts after the calendar in the usual arrangement.)

Digital reproduction changes that. Digital reproduction depends fundamentally on illusion for its success. Any digitization process involves sampling a source to a sufficient level of detail and accuracy to appear as though no sample had been taken. Although sampling means much is left out of the reproduction, a successful sampling process means the missing material is not evident: a photograph appears to be continuous colour, a song appears to be a continuous vibration. We accept the illusion as the reality. The human ear can not distinguish higher frequencies than about 20,000 hz, so sampling sound above that frequency is meaningless (apart from technical purposes).

The scanning of books and images is based around a standard sampling and reproduction framework. Sampling a book, we start at the beginning and work to the end. We then reproduce the result on the standard screen (typically a wide rectangular space) and the reproduction is sized to fit on that space. A large format book and a small format book appear to take up roughly the same space on this screen so that both can be viewed. The digital sampling has not recorded the impact of size. Both are flat on the digital screen. The digital sampling has not recorded the texture of the ink or the gilt on the drawings. It has not recorded the context that is part of the book or of the page.

The viewer, of course, knows this. It is obvious that the page is being viewed on a screen, not in the hand. But the reproduction is so clean and precise, the illusion of completeness so complete, that we de-emphasize the importance of what we do not have. When CDs first introduced digital recordings to the mass market by digitizing vinyl and providing cleaned copies of the music with all noise removed, audiophiles noticed the missing clicks and hiss of the recording studio. These elements had become part of the music, and they insisted it be returned. Audio recordings now routinely remove this “noise” and no on complains, because listeners less often consider it an important part of the musical event. As we become increasingly reliant on digital copies, we may learn to think of the missing parts as “noise” and believe we can understand these manuscripts and documents without it.

I'd like to hear any thoughts you have on this.