Electronic records management
I’ve been thinking about what a free (libre) electronic record management system (ERMS) would look like, whose needs it would address and how it would be put together.
Record management is a complex area, but a certain amount of the complexity of electronic systems is in my view unnecessary – it’s a by-product of attempting to transpose the methodology of paper-based record management into an electronic software product.
There is no need, for example, to have every record located within a monolithic hierarchical file plan. Electronic records still need to be organised and classified, in quite a variety of ways, but as they no longer have a physical location as such, it’s no longer necessary for that classification to encode information about where you have to go in order to find them.
Already in ERMSes that use a hierarchical fileplan it’s quite common to have a record “located” in several places at once (or cross-referenced between multiple locations). Something like Google mail’s system of applying tags to records, rather than “filing them away” in “folders”, seems ultimately more appropriate to this reality. A sufficiently flexible query engine would allow the user to traverse any of several overlapping classificatory schemes in search of related material, providing multiple views (some of which might indeed be hierarchical) onto a multidimensional grid of associations.
A user searching for related information should be able to pull on any of several threads criss-crossing through a particular record, and see what else is strung on each thread or where any particular pair of threads cross again elsewhere.
A user looking for a group of records with something in common should be able to identify that something unambiguously without having to refer to a monolithic hierarchical fileplan that creates ambiguities whenever some new piece of information has to be slotted into it that doesn’t quite fit into any of its little boxes – does it belong over here, or over here…?
A user capturing some new piece of information and creating a new record should be able to give it a partial classification that captures some, but not all, of its salient characteristics, such that this classification can be extended over time as other users discover new contexts in which that record can be seen as pertinent.
Oh, and all of this has to be realised without creating a hideously complex and completely unmanageable mess…