I gave a talk for the Amigos “Digital Preservation: What’s Now? What’s Next” conference today in which I trotted out what I consider one of my better formulations of (one of) the model(s) I use for thinking about preservation. This model says that preservation consists in sustainable efforts, optimized over time. The change is from thinking of “preservation” and “preserved” as closed efforts or checklists of pre-defined actions to be completed. Instead, the focus is on ranking the “preservation” of things and having a process for “preserving” them.
With that distinction made, it becomes clearer that preservation/preserving is a function of strategy and management and that preserved/preservable is a function of technical knowledge about the specific types of objects in an archives’ care. For the strategic side of the process, I think Snowden and Kurtz’ Cynefin model is the most useful thing going, closely followed by Shenton and the British Library’s work on Life-cycle Collection Mangement.
For the objects themselves, I propose the following model:
- Substrate: tangible substance(s) that carries media
- Media: material(s) that record information
- Transport: means(s) for perceiving media
- Language: system(s) for interpretation of media
In this way, digital preservation and artifactual preservation can be dealt with as different instances of the same basic problem. Preservation succeeds when failures in all four factors are eliminated, corrected, or mitigated and there is a sustainable process in place that will support prevention and repair or recovery at appropriate times.
For artifactual records substrates and media are largely stable (even nitrate films and iron-gall inks last for decades), transport is simple (turn on lights), and languages decays very gracefully (compare Jane Austen and Paul Auster, for example). Substrate and media questions are most often answered by environmental controls, and transport is basically a function of the library-as-place and its associated operations. Artifactual libraries and archives are about putting things in known places and retrieving them for later use, just as Babbage did with the “store and mill” in his difference engines, or as WordPress does with PHP, MySQL, and Apache. This preservation framework also shows the importance of library efforts to advance literacy and implicitly argues for the research library’s place in the university. From kindergarten to the post-doc, education keeps the library, archive, or museum readable.
For digital records, the substrates and media are largely unreliable (the best digital lifespasn roughly match the worst instances of artifactual longevity), the transports are complex but with mixed results in reliability. The individual pieces of the internet are fairly fragile, but as a network, it is very robust. This may be a persuasive argument for co-called cloud systems of digital preservation. It is most certainly the reason LOCKSS has been the little engine that could and did in the digital preservation effort. Finally, the language of digital records can be quicker to change and tends to fail rather than decay. This is not the unviersal case, however. The TIFF format, PCM audio, and Unicode text (UTF-8 encoded, normally) show no problems and have decades of longevity behind them. Other records use less durable (more proprietary, less clearly standardized) languages.