Technological obsolescence is commonly seen as a big issue in digital preservation yet some people seem to have different interpretations of what it means, leading to heated discussions about the point at which formats or software become obsolete. So what's the most useful way we can define technological obsolescence within a digital preservation environment?
David Pearson and Colin Webb provided a pretty detailed background to and analysis of file format obsolescence in their IJDC paper from 2008, and they use this scoping description:
The paper also quotes me as describing:
I think it's important to remember that this is a very fuzzy concept and it means different things in different contexts. In most cases its more about inconvenience and cost rather than that extreme of absolute loss.
When I used to work at the British Library, we informally used the term "institutionally obsolete" for digital content that either wouldn't run natively on a reading room machine or wouldn't run on a "typical" users machine without installing some additional software. A case that was "institutionally obsolete" might be quite easy to fix by a technically competent individual via some extra rendering software (and hence potentially dismissable as "not really obsolete"), but might be a significant issue for the management of reading room machine setup and maintenance.
Therefore, obsolescence is a pretty meaningless concept without that crucial context.
This of course then brings us to the concept of Representation Information (OAIS) and the idea of a network of information that informs the rendering and understanding of digital information. Progress in advancing this concept has been slow, but an approach to demonstrate how it could be realised was described and tested way back in the late 1990s by the Cedars Project: A blueprint for Representation Information in the OAIS model by Sergeant and Holdsworth.
Ultimately we need more hard evidence on what obsolescence actually means in practical terms as a long term preservation risk. Andy Jackson made some first steps in analysing trends in format popularity within a web archive in this iPRES paper: Formats over Time: Exploring UK Web History. His conclusions suggest format obsolescence is not as great a risk as some have suggested (Rothenberg's famous "forever or 5 years, whichever is sooner" quote for example) but there are trends of usage and specific examples of obsolescence that require further examination.
I'd argue that it occurs when a given format is no longer in common use by its original base of users because they've moved onto something else.
Luckily, it's not a 'drop dead' date; you have time to migrate off to some other format, and you typically have enough time to do a survey to evaluate what the best format to migrate to might be.
Occassionally, though, you still run into niche formats -- some communities will keep a physical or file format as their standard even when the rest of the world's moved on to something else, as it meets their needs (and they don't want to retool everything). You have to be careful about defining what your base of users is; Betacam was considered obsolete by the consumer market more than a decade before it was in the professional market. I know people who were still using Betacam until a few (4?5?) years back, as that's what some TV stations used for people to submit commercials.
When the community that you support starts moving to some other format en masse, you should prepare to move. (but don't move as soon as some new format that's supposed to everything gets announced, no matter how 'better' it might be, because if doesn't get the traction needed, it's dead on arrival.)