Andrea posed very eloquently the challenge ahead of us as we prepare our exhibit for Friday: why should others at DHSI care about what we have discussed this week? What is their take away?
For me, the theme running through all of this week’s courses (including ours) is that of learning how to build models: representations of one thing seen through some other lens.
Inevitably, this involves a series of decisions about what information to emphasize, and what to let slip away–either because you’ve excluded it all together, or you simply haven’t marked or emphasized it.
The core of TEI is the task of creating a model of a text: you choose how to mark things, and what to mark; you represent some things and not other things.
A database is a model, too: you define relationships and put things into tables according to labels and categories that you define–someone else might do it differently. Defining things in one way may mean that they can’t be simultaneously defined in some other way. The structure you choose has implications.
Visualizations let us see, for example, all the places mentioned in a text plotted on a map, precisely by hiding/omitting all the other non-place words. A model is as effective in what it leaves out as well as what it leaves in.
As Helene noted, this is also what catalogers do: the bibliographical record is a model of the book: it usefully tells us some information quickly, by ignoring lots of other information.
I think we might do well to imagine digitization of books, by which I mean, mostly, image-based representations of books online, as models of objects, in this same way, as a model.
This perspective sheds light on some of our challenges. We are anxious about what is lost with digitization, but when we talk about models, it seems easy to acknowledge that they *only work* when they privilege some information over other information. Inevitable loss, then, can be used to our advantage.
Likewise, a model is devised to serve a certain purpose, and most of the time, to serve only one purpose at a time. When you try to show too many things on a graph at once, it quickly becomes insensible and serves no one’s needs. An attempt to do everything at once with a digitization will inevitably go too far and simultaneously fall short.
When we look at a graph, we understand that it is not the entirety of the data itself–it is a refined selection, chosen to make one argument.
Taken in this way, the limitations of digitization don’t make me so anxious at all.
I think this idea of the model might resonate with others at DHSI–they have probably been wrestling with it themselves all week.