Evaluating Digital Work in the Humanities

https://i0.wp.com/projects.chass.utoronto.ca/amphoras/ills/sym99-f2.gifLeave it to an archaeologist, but when I heard the CFP from Digital Humanities Now on ‘evaluating’ digital work, I immediately started thinking about typologies, about categorizing. If it is desirable to have criteria for evaluating DH work, then we should know roughly the different kinds of DH work, right? The criteria for determining ‘good’ or ‘relevant’, or other indications of value will probably be different, for different kinds of work.

In which case, I think there are at least two dimensions, though likely more, for creating typologies of DH work. The first – let’s call it the Owens dimension, in honour of Trevor’s post on the matter- extends along a continuum we could call ‘purpose’, from ‘discovery’ through to ‘justification’. In that vein I was mulling over the different kinds of digital archaeological work a few days ago. I decided that the closer to ‘discovery’ the work was, the more it fell within the worldview of the digital humanities.

The other dimension concerns computing skill/knowledge, and its explication. There are lots of level of skill in the digital humanities. Me, I can barely work Git or other command-line interventions, though I’m fairly useful at agent simulation in Netlogo. It’s not the kinds of skills here I am thinking about, but rather how well we fill in the blanks for others. There is so much tacit knowledge in the digital world. Read any tutorial, and there’s always some little bit that the author has left out because, well, isn’t that obvious? Do I really need to tell you that? I’m afraid the answer is yes. “Good” work on this dimension is work that provides an abundance of detail about how the work was done so that a complete neophyte can replicate it. This doesn’t mean that it has to be right there in the main body of the work – it could be in a detailed FAQ, a blog post, a stand alone site, a post at Digital Humanities Q&A, whatever.

For instance, I’ve recently decided to start a project that uses Neatline. Having put together a couple of Omeka sites before, and having played around with adding plugins, I found that (for me) the documentation supporting Neatline is quite robust. Nevertheless, I became (am still) stumped on the problem of the geoserver to serve up my georectified historical maps. Over the course of a few days, I discovered that since Geoserver is java-based, most website hosting companies charge a premium or monthly charge to host it. Not only that, it needs Apache Tomcat installed on the server first, to act as a ‘container’. I eventually found a site – Openshift – that would host all of this for free (! cost always being an issue for the one-man-band digital humanist), but this required me to install Ruby and Git on my machine, then to clone the repository to my own computer, then to drop a WAR file (as nasty as it sounds) into the webapps folder (but what is this? There are two separate webapp folders!) , then ‘commit, push’ everything back to openshift. Then I found some tutorials that were explicitly about putting Geoserver on Openshift, so I followed them to the letter…. turns out they’re out of date and a lot can change online quite quickly.

If you saw any of my tweets on Friday, you’ll appreciate how much time all of this took…. and at the end of the day, still nothing to show for it (though I did manage to delete the default html). Incidentally, Steve from Openshift saw my tweets and is coaching me through things, but still…

So: an importance axis for evaluating work in the digital humanities is explication. Since so much of what we do consists of linking together lots of disparate parts, we need to spell out how all the different bits fit together and what the neophyte needs to do to replicate what we’ve just done. (Incidentally, I’m not slagging the Neatline or Omeka folks; Wayne Graham and James Smithies have been brilliant in helping me out – thank you gentlemen!).  The Programming Historian has an interesting workflow in this regard. The piece that Scott, Ian, and I put together on topic modelling was reviewed by folks who were definitely in the digital humanities world, but not necessarily well-versed in the skills that topic modeling requires. Their reviews, going over our step by step instructions, pointed out the many, many, places where we were blind to our assumptions about the target audience. If that tutorial has been useful to anyone, it’s entirely thanks to the reviewers, John Fink, Alan MacEachern, and Adam Crymble.

So, it’s late. But measure digital humanities work along these two axes, and I think you’ll have useful clustering in order to further ‘evaluate’ the work.