Imagining URLS
As a glance at any URL will tell you, unless you’re a computer with internet access, a “Universal Resource Locator” is nearly meaningless. Further, as IT standards continue to evolve no doubt even this seemingly bedrock spec will probably change. Further, far from “Universal”, URLs are only meaningful in specific, limited contexts: there are many resources that cannot be located with a URL.
This leads me to ponder a truly Universal Resource Locator. What form would it take? What attributes would it have? How would you use it?
The question stems from a practical problem in my current work with the James Madison Carpenter Folklore Collection edition. In the final printed volumes we wish to reference unique original documents held by archives, their digital counterparts and some digital-only resources in a way that is comprehensible to an ordinary human (as much XML can be, for example), not tied to a particular technology which will no doubt become obsolete or superseded, and which can describe many kinds of things: documents, sound recordings, photographs and so on.
Bibliographic references offer an analog: they have evolved over the years to reference different kinds of print materials in particular contexts. Sadly, there is no universal standard. Author-date, MLA, APA, etc. are all variations on this theme. But these standards have evolved because they’re useful: they provide a consistent and clear way to help you locate a book, article or other document or portion therof.
So that’s the challenge I’d like to pose to anyone who’s interested: help me imagine a technology-independent, obsolescence-resistant and truly Universal resource locator.
Thoughts? Ideas?
Very interesting question. yThe kernel of the question to me , is what is the atomic unit of a “resource?” The standard library catalog systems use a published work as a unit, in general. URLs… it’s more variable, and maybe there isn’t a good word for what the unit is… a single web-servable page or similar entity? URLs can be nested, as images on an HTML page. Would you want a similarly nestable system, where you can catalog a volume as well as the contents? Similar to cataloging a serial and the issues within it in a library.
Bob – I’ll write you a more studied response in the next few days, but fear you are doomed to fail – if only for the reason that it is very difficult to write a solution which covers the eventualities which you haven’t yet thought about. Certainly a hierarchical solution is dead in the water – no two people will ever agree on its structure – the question being in it’s most simplistic form – do you file by author then subject – or subject then author? So we need a flat structure with multiple (infinite) index fields and rules for populating each field – perhaps each index field is itself a complex object defined to cover a resource type or family – eg a sound record characterised by the medium it is recorded on, the date it was recorded , the length of the recording, the type of content. I have had a view for many years that the “Killer App” will be a tool which lets you display and navigate in two (or possibly three) dimensional space a multi-dimensional data set.
This is a very knotty problem, and I am not hopeful that anyone can devise a complete solution. As Tim pointed out, a truly universal solution has to be open-ended, to accommodate in the future what we could not imagine today. And, as is clear in Bowker & Star’s book, categorization requires approximation and compromise.
I will suggest the same two book resources that I think I mentioned to someone else locally (Nat?) a year or two ago :
Data and Reality, by William Kent.
Sorting Things Out, by Geoffrey Bowker and Susan Leigh Star.
The best class I took in CSCI graduate school was Logical Data Structures, taught by John V. Carlis at the U of M, and I heard of Kent’s book from him. I think John is still at the U and he might find your challenge very interesting.