[-empyre-] Preservation of Web materials

["Gerard Clifton" <gclifton@nla.gov.au>]

Hello to all, 

This is my initial post. Many thanks for the opportunity to contribute
to this forum. 

There have been some interesting posts which have gone straight to some
of the core issues facing digital preservation.

One of the main issues is that of technological obsolescence and our
ability to re-present the materials we are creating or collecting, which
Luke noted in terms of the overall systems of media and user interaction
over the next 100 years or more, and Adam noted in terms of proprietary
codecs within a single media type over a potentially shorter period.

The materials in PANDORA and other Web archives represent quite a
challenge over the medium to long term, due in part to the complexity of
the material - the number of files, the number of files types (around
200 MIME types alone in PANDORA), and the relationships and dependencies
between them. The issue of codecs in media files is, as Paul said, one
we identified but not one that we yet have a technical solution for.
Finding individual solutions for the whole range of file types we
collect will be a huge task. Working out what is in the archive,
relative proportions and associated risks has been a recent focus, with
the aim of prioritising which of the myriad issues to address first.

There isn't yet an overall strategy for addressing long term
preservation of Web materials. We expect emulation would be problematic
in a distributed Web environment as we would have little control over
the type of platform and configuration the end user might be using.
Migration is likely to be suitable for a proportion of items (at least
images and some document types), but more complex and interactive
objects are still problematic (just as they are for stand-alone items). 

There are also those strategies that lie somewhere in between the two -
the use of viewers/plugins to read or deliver obsolete content, or
migration at the time of access. An article that may be of interest is
by Rosenthal et al., describing LOCKSS's demonstration of on-the-fly
format migration during Web delivery of images. (See D-Lib Magazine, Jan
2005: http://www.dlib.org/dlib/january05/rosenthal/01rosenthal.html ). 

Solutions will probably be found using combinations of these approaches.

Regards,
Gerard Clifton

Digital and Audio Preservation Resources
National Library of Australia
tel. 02 6262 1366  fax 02 6262 1653
email gclifton@nla.gov.au