One of the thing sI like most is when a data visualization tool encourages play, the more intuitively, the better!If all the world’s a stage, with our parts, entrances, and exits, should “digital humanities” not, by the very nature of its name, promote digital methods to which we see the complicated storylines, character arcs, and acting methods? I believe the best ones do. There are, however, static aspects of a theatrical play, such as the tableau, a still frome easily photographed, pleasing visually and telling of the story in some cases. The Tableau software we worked with this week accomplished just that, but for data sets. And, while I believe Tableau does well enough at accomplishing its purpose (a still visualization of a complex, real life system), we should not get bogged down too much in analyzing it. If we do, we run the risk of losing sight of the intricacies of the human system – or stage, in our metaphor – the nervousness, the adrenaline, the facade, the spectacle, the introspection, the emotion… the part that makes it as much a living, human thing as anything else. The part “the humanities” clings to, quite desperately.
Now, I don’t mean for this to be a heavy critique of Tableau by griping about what it doesn’t do. I would rather engage with it on its own terms. It is moreso the attitude around digital humanities critique that I would like to, well, critique. To start our show, though, let’s just walk through Tableau.
Tableau
My experience with Tableau was not remarkable compared to other data analysis methods I have used in the past. I am impressed with the ease of operations in tableau and the ability to switch between different modes of visualizations, such as polygon map, point map, and basic chart. I also was impressed with the options for topographic map layers supplied by Tableau – I used one above, called “Percentage of Population, Black/African American.” This was an interesting map layer to put under my layer of pin drops showing the frequency of “negro ballads” recorded by John and Ruby Terrill Lomax on their 1939 Southern Field Recording Trip. Notably, it is particularly helpful that in the search for “negro ballads,” the Lomaxes left out a state with a considerably-sized black population. The only issue for these topographic layers supplied by Tableau, though, is that they don’t provide the source or the specific numbers – so it is difficult to say exactly what the statistic is or from what year it was taken. Since Georgia had a sizeable black population in 1939, I still find it a helpful visualization for my purposes, but also would need to be more heavily researched.
The Tableau maps and charts, while fairly quick to make and easy to modify, are not interactive. While the goal of Tableau is not to make interactive maps and visualizations, I will not fault them for this. That would be unfair. I will, however, note that there are other mapping platforms that perform similar functions and let the user (not just the creator) play with map layers, pin-drop pop ups, and topographical layers, such as ArcGIS, CartoDB, and Omeka. While they are less intuitive than Tableau in the map-making process, Tableau is great for creating maps and intentionally thinking about the data in charts, as opposed to batch uploads which magically produce drops on a map.
I hope that in my comparison of these mapping tools I have emphasized a form of writing most scholars are unnused to – compare and contrast, with no objective “best” form of software that does everything. There is no software that does everything – so based on your needs, different software will be best for you. That doesn’t mean that something else isn’t extremely useful to someone else. That being said, there are some forms of digital tool critique that I find quite helpful and illuminating. The first is accessibility and readability.
Especially when concerned with data visualization, we have to keep our reader’s visual literacy in mind. Nessa’s “Visual Literacy in an age of data” offers a nice critique of the ways in which increasingly complicated ways of depicting data actually create more distance between the researcher and the audience and makes you lose readers by making your work visually inaccessible. Practical solutions included opting for bar graphs instead of pie charts; clean, simple text that explains even complex arguments with as much layman’s terminology as possible; and omitting arbitrary visualizations.
We also have to be careful about the way we frame the possibilities and practicalities of digital tools. Physics arXiv Blog’s article “When a Machine Learning Algorithm Studied Fine Art Paintings, It Saw Things Art Historians Had Never Noticed,” makes it sound like the AI potential in their software is capable of things that human art historians were unable to do – this is untrue. They were able to make one connection in particular between two unrelated artworks that hadn’t been made before (presumably whoever noticed this has read all of the literature possible on these artworks and therefore can confidently say that connection had not been previously made). The title of the post is misleading, as even in the article they say that the AI in no way could replace a real, human art historian. The flashy headline to catch readers’ attention to a DH project ended up negatively affecting the overall tone of the article and came across as very off-putting to every art historian I’ve talked about it with.