BBC News Magazine reports on the efforts of scientists to recover sound from 19th century wax cylinders using scanning microscropes and digital conversion.
While this is great – we’re managing to recover knowledge that we have lost – there is a certain irony that the old, analogue wax cylinders are, in a way, more durable than the digital copies that they are creating.
The durability problem is not like the problem the cylinders had, namely the possibility of hardware failure. The cylinders which were too delicate to play on the equipment they were designed for, which is how we lost the information on them in the first place. While the hard drives used to store the digital versions might get destroyed, the researchers have probably backed them up somewhere else (we hope). The emergence of peer-to-peer technology for durable storage (which I wrote on in my BA dissertation) is emerging, so that the risks of hardware failure are even further reduced.
Instead, the problem is of software failure. The scientists were able to get the sound off the discs because it is recorded in an analogue manner – the ripples and waves on the cylinder grooves bear a direct relationship to the sound waves produced. Decoding this was easy (in principle at least, I’m sure the work in actually doing so was quite a feat). But now they’ve digitised them, they’ve encoded them in a particular way. Who’s to say 100 years time encodings we take for granted, like JPEG or MP3, will be decipherable? Future historians and digital archaeologists will just look at the 1s and 0s and be completely baffled as to how they relate to what you can see and hear.
This has already happened, as the BBC know all too well. The 1986 BBC Domesday Project was meant to be a permanent multimedia archive of Britain in 1986 (marking the 900th anniversay of the Domesday Book). Instead, 15 years on, the discs were indecipherable, as no-one had the BBC computers needed to read them. Fortunately, the Camileon project have since managed to reproduce emulator software that can read it, but in time the emulator itself may become obsolete (one day, will we need PC emulators to run the BBC emulator software?).
Fortunately, the potential problems has been recognised now (the problem of hardware failure was never really anticipated by papermakers or wax cylinder manufacturers in the past), and so we have organisations like the Long Now Foundation (who date everything in 5 digits to avoid the Y10K bug), the Gutenberg Project, and the UK’s Digital Preservation Coalition devoted to making sure we don’t get a series of Domesdays in the future. The science of data preservation is a growing one with no one solution or consensus existing; Camileon provide a useful list of publications on the subject – the ones on the different methods used and attacking emulation as a strategy are particularly interesting to read.