Isochrone maps are well-known enough to have their own Wikipedia page and Google Maps API tutorial. These represent the time to get to or from locations as a gradient and/or contour. Their usefulness in representing historical movement is obvious, as evidenced by these two isochrone maps of travel to and from Rome in July:
Isochrone map showing time in days to travel to Rome in July
Isochrone map showing time in days to travel from Rome in July
There is not, as far as I’m aware, a similar term for maps that display the expense of transporting goods to and from one or more locations on a map. Given isochrone (from isos chronos or “equal time”) there should be an equivalent ancient Greek word for expense to transport. It turns out, according to The Perseus Project, that there are at least three words for the expense related to transporting goods. Fortunately, since my ancient Greek is nonexistent, I can rely on Walter Scheidel and Dan-el Padilla in our Classics department to select the right one. Thus, I can now properly label my isophoretric maps, or maps that show “equal expense to transport goods”:
Isophoretric map showing the cost in denarii to transport a kilogram of grain to Rome in January
Isophoretric map showing the cost to transport grain to Rome in July
Naturally, these are drafts, and as such should not be utilized for navigational or capitalist purposes by time travelers. For instance, you’ll notice in the January isophoretric map that I’ve manually masked off the contour lines before the exaggerated (and simply wrong) interpolated values stretch into the regions for which we do not have data.
When Ian Bogost and Mike Migurski both mention the same term in close chronological proximity, I feel the need to pay attention. Of course, the one thinks it’s more fodder for taking seriously the personhood of objects (so much so that my use of ‘personhood’ in describing this would likely result in claims that I’m being personist) while the other sees it as a frame for the ever-increasing beauty and sophistication of digital maps. Regardless of whether one takes either of these positions, we have to recognize that there is a new aesthetic movement and that it bodes well for digital humanities scholarship.
Fortunately, you need not be terribly aesthetically aware to benefit from this in the practical terms of integrating these new principles into the creation, representation and evaluation of digital objects. There is a maturity of praxis in the digital creative fields, such as computer gaming, digital painting, citizen GIS, and too many other digital flowers for Mao to count. And this folding in of new digital movements (artistic and social) into academic research, archiving and publishing is an important process for successful digital humanities work
For instance, what Eric Fischer did in mapping the different patterns of geo-tagged tweets and Flickr photos is not only interesting aesthetically but a good methodology that should be explored and built out. It’s notable that Linna Li and Mike Goodchild’s exploration of the same topic took place nearly a year after Fischer had done this. Likewise, my own mapping of DBPedia was inspired by Fischer’s project and is currently being developed to deal with a host of big linked geodata resources in support of the study of natural environments in urban areas. This great vibrancy of digital creativity–whether it is in cultural critique, representation of society, radical geography or artistic procedurality–is fertile ground for scholars of the humanities to draw inspiration (and, dare I suggest it, collaboration) for their own work.
How do we do this? First, by embracing design in the development of digital scholarly work. I’ve been late to the game when it comes to understanding the importance of UI/UX and other design principles in finished products, thinking that an opaque data visualization is fine as long as the few people who are highly invested in it can access it. My concept of visual literacy, for a long time, placed the impetus on the reader. Like the spur to algorithmic literacy embodied by such chastisements as “Program or be programmed”, I felt that it was incumbent on the part of the reader to understand the importance of being literate (if not fluent) in how information visualization functioned, with creators being not only less responsible for the poor communication going on but also acknowledged as less likely to change since they held so much power. But a literate society benefits both groups, and making algorithms and data more accessible is as much the responsibility of the creator as the user.
This last point is the key, I think, to the New Aesthetic, in that it embodies the fact that all are creators (and hence programmers) and all are users (and hence programmed). While those of us ensconced in our ivory silos may be tempted to think that lesser persons who cannot code make up the new underclass, there are very few people who now are not digital content creators using visual toolkits to build complex digital objects. Creating something with the richness and reach of a WordPress blog is now so easy that the digerati mock it as not being real digital creation, rather than acknowledging the growing ease and sophistication of digital work. Creating complex data-driven maps is so simple now that you need not ever have taken a GIS course, much less become a GIS professional as was necessary a decade ago, and the result is a growing sense of disdain for all the new cartographers who can produce an amazing map without knowing what a datum is.1
We have already reached the point where hoarding data is more likely to result in missed opportunities than it is to result in benefiting research. If we try to hoard processes, or simply fail to acknowledge this ripe field of interesting methods being developed by the New Aesthetic movement, then the same thing will happen with techniques and tools.
1I think the opening up of GIS to people who cannot afford ArcGIS licenses or the time and money necessary for quantitative geography courses and GIS tutorials is very similar in tone to the opening up of Instagram.
A map of expense contours developed from the ORBIS model. Each contour line denotes the cost to deliver goods to Rome.
This event is open to the public.
On May 2nd, I’ll be presenting a model of the Roman World, which I created with Walter Scheidel over the course of the last eight months. Data-driven models take many forms in their use and presentation, and in this case we’ve created a website that acts not only as an interface into that model but as a representation of the model itself. It’s not the entire Roman World, and by that I mean not that it doesn’t cover the pieces of Eurasia and Africa associated with Rome, but rather that it only models the transportation of individuals and goods. I find this to be an important distinction to make about representation and the “world”, which tends to have a very fluid definition in traditional humanities scholarship but can quickly devolve into a purely geographic term in digital humanities scholarship.
I’m not the principle investigator but rather technical lead on the project and also as a member of the Stanford Library. On one hand, this means that I can’t speak with authority to the historical principles that underlie the model. So, while I will be capable of demonstrating the variation in time and expense for travel from Rome to Britain over the course of a year, I cannot speak to the decisions as to how to rank particular sites or how to evaluate the cost of particular routes. That said, the representation of the structure of the model I alluded to earlier includes the technical elements that went into building the data and functions as well as the historical research required to design the relationships within the model.
The integration of digital resources with humanities scholarship is nothing new. GIS and spatial analysis have increasingly been used by historians as part of their craft, and the Spatial History Project has been actively making this possible at Stanford for over half a decade. It’s gotten to the point that the development of common resources and applications for the storage and presentation of geodata is a common theme both in academia, such as Harvard’s WorldMap or Geodata@Berkeley, and in the public sphere with projects like GeoCommons. Such broad penetration of spatial data affords us the opportunity to look ahead toward even more sophisticated uses of digital methods and objects.
One way that sophisticated representation of scholarly arguments can occur is to embed the scholarly argument within the structure of the digital object. Rather than creating a separate collection of datasets that are alternately analyzed and visualized, humanities scholars can begin to define how these points of data cause and are affected by each other. Interrelation, not correlation, allows for the scholar to situate their argument within the digital object. This is the definition of a model–a collection of data coupled with functions that define how the data interact with each other.
Compare this to a typical digital map, whether static, interactive or dynamic. Data visualization is just a new word for illustration and figure, and just because an illustration is a geographic one does not make it particularly different. Static and animated maps are typically offered as a generic tool (usually an index) to be used to look for spatial correlation (spatio-temporal correlation in the case of maps that are animated by time) by an expert observer. In the case of maps of historical phenomena, this means a historian looking for structural elements, anchored in space (and therefore amenable to spatial principles like Tobler’s First Law), that can be associated with historical phenomena not represented on the map but represented alongside or otherwise divorced from it in linear text. This “other phenomena” includes political movement, religious tension, ambition, administration, colonization, extraction, oppression and every other active property that makes up history.
It is this “other phenomena” that needs to be represented in digital form. Until then, we are only displaying arguments of correlation and not of causation. A map that shows the Angkor empire grow and recede can be disputed based on the quality of data or decision to display shaded regions or even the choice of using tweens in the map. But, since it does not embed any procedural claim that one entity affected another, or that the actions of members of one group somehow influenced another, then the only engagement with such claims can occur within the realm of the narratives that surround such a map in print.
If, however, the map were not simply an animation of spatial change over time but rather a visual representation of some model that indicated how a process of Khmer Imperialism would function and how it might create and then be dismantled by another process, then it is a scholarly argument that can be meaningfully engaged with.
Narrative text would necessarily surround such an argument, especially in the early phases of production of such digital scholarly work because it would be so unfamiliar, but engagement with it would be with causal claims and not simple correlation. Like traditional scholarly works, it is a tool–a device to be used in other research. Good research serves as the building blocks of later, better research. But unlike a pure tool like Voyeur, and just like a traditional scholarly work, a model can be contested, amended and extended. The ORBIS model makes specific claims about the shape and size of the Roman World and how that shape changed based on goals, means and even the time of year of the act of participation within it. These claims are not simply correlative, and can be disputed or enhanced by later scholarship within the same architecture of models. This may not sound like the most interesting claim, especially when compared to the lofty list of “other phenomena” presented earlier, but it is only through grappling with the comparatively simple processes, such as transportation, that we can lay the groundwork to deal with extremely complex ones, like religious diffusion.1
Models are traditionally something that exists elsewhere, as part of earlier research or in a sort of semi-religious state to be consulted like the Oracles of old. Instead, ORBIS presents the model as a product of digital humanities research, to be contested or expanded by later work. As such, exposing the structure of the model in as many manners as possible is paramount. The data will be available via API, and the functions will all be published so that they can be reviewed for validity and integrated into other research. The creation of a transportation network model of the classical world would likely require a large grant and several interested working groups, as well as developing methodologies for mapping networks in less historically attested regions. This is one possible direction that later researchers could take. Another is to focus on and refine the model of Rome or develop a component that deals with the production of goods and raw materials, or the transmission of information. All of these would require that historians formally define historical processes for computational representation, allowing for the expression of such processes in a form other than linear, narrative text.
There is another reason to move away from data and toward models: data does not scale. ORBIS provides route data for travel between 700 sites in the Roman World, with variation based on month, type of travel and priority. If we were to simply generate individual polylines for each route, this would be 211,680,000 separate lines.2 If the epidemic of big data in the modern age has taught me anything, it’s that we should focus on describing processes and not building great archives of data. Nick Montfort suggested that we need a Digging Into Code grant instead of a Digging Into Data grant, to shift the focus of digital humanities research toward process. The ORBIS dataset is puny in comparison to many high-profile digital humanities projects, with only a few thousand pieces of data, but the code that runs the multimodal, temporally dynamic network, as well as the code that provides users with an interface into that network, is very sophisticated.
The academic publishing industry continues to grapple with how it can continue to support traditional scholarly works while providing the opportunity for scholars to create rich digital scholarly works. While many of the methods and technologies needed to create digital scholarly works are mature, they still require a level of creativity and administration far above that necessary for a traditional narrative. It is, to put it crudely, far easier within the academic sector to format, typeset and edit a linear narrative than it is to deal with even interactive and dynamic maps, much less true models. That’s why so much of the innovation in digital publishing is still confined to narrative representation of arguments, such as that found in a typical digital journal or a commons-based peer collaborative or social media-based system. Given the high cost and specialized skills needed to create and publish digital scholarly work, it seems the best model, if you’ll pardon the pun, is for the university library to provide both research support, avenue to publish and resources to archive and maintain this new genre.
1This is not an idle example. Lew Lancaster suggested in his keynote address, “Crossing a Boundary: Where, When, How,” (UVa, Feb 2009) that the origin of Mahāyāna Buddhism would be better understood if we had a functioning network model of the sea and land trade routes typically known as the Silk Road.
1This is very much a back-of-the-envelope calculation. Many of these routes would overlap, reducing the high value, but this does not take into account the ability to constrain modes or use different categories of sea or river transportation, which would increase the number.
When I suggested an animated encyclopedia of verbs and processes, I naively assumed that nouns were taken care of. Not so, as The Noun Project demonstrates. The site, which began with the question, “what if I had a sketch for every single object in the world?”, is a repository of simple SVG icons typically found on maps and signs, as well as a few not-so-common icons. It is a sign of the ubiquity of information visualization that these sorts of projects are beginning to appear to complement the growing data visualization libraries and tools.
I was disheartened to see that, aside from a single soldier, there were no icons for Ancient Rome–the icon for a “city” is of the kind that comes with clock towers and skyscrapers. But it makes me think we should have a Digital Humanities Icon Camp to try to work out a few icon primitives for nouns in the humanities. This might all sound very silly to some, but as we begin to define vocabularies (or ontologies or data models) as one of the products of digital humanities work, how we represent individual elements of those vocabularies in data visualization should not default to whatever icons are available in the standard ArcGIS or Tableau libraries. If we are to use discrete, abstract representations of nouns in our data visualization in the humanities, which seems to be the case, then should we not formally examine the process by which these discrete representations are created, defined and implemented?
One of the worries about using and representing data in the humanities is the implied exactitude of it all. So, it’s enlightening to browse through the submissions and see the number of different ways that bicycles, mopeds and guitars can be displayed using simple black-and-white vector graphics. If we can have thirteen different icons for a motorcycle, each emphasizing different legitimate views of what constitutes the formal object, then maybe we can also have a few different icons for an aristocrat or a revolutionary or a poet. Maybe such an icon can even change over time to indicate the changing definition of such a role within a society. “Ambitious temporally-aware dynamic icons for digital humanities information visualization” sounds like a good title for a grant.
The Noun Project also maintains a blog that’s a great resource for design thinking, with insightful posts on symbology and effectiveness. It also provides some helpful advice for those of us who are not Scott Weingart.
I’ve always felt that the cartoonish nature of icons and symbols that typify infographics undercut the perceived sophistication of visual display of knowledge. Cartoons, after all, are for kids. But as video games grow up and cartoons grow up, perhaps an investment of intellectual capital into the production of suitable icons for sophisticated knowledge representation can become a part of the production of new digital scholarly work.
The inaugural edition of the Journal of the Digital Humanities is now online, and includes an article by me in the Conversations section titled Digital Humanities as Thunderdome.
I will resist posting a photo of Tina Turner and Mel Gibson so as to maintain my newfound (slight) credibility.
Stamen Design, who has done too many beautiful geospatial things for me to count, has created something I’ve been wanting for years. It’s a web basemap in watercolor.
Incredible.
More information on how they were made can be found here and here.
I took part in the Day of DH yesterday, though not as vigorously as I did last year. Here is the Day of Elijah Meeks, in case you’re really, really interested. The post of mine I thought the most valuable is reprinted below, but here with an illustration.
This is my favorite visual metaphor for collaboration. On some days, I think of the troll as an interesting, sophisticated digital humanities research problem, that can be solved by working together. On other days, I think that the troll is the one poor DH schlep actually trying to do something, and that the wolves are his "collaborators", a view made more terrible when you think of how many of the wolves are just howling.
Collaboration is an interesting bird. I’ve spent the better part of my day, when not trying to coax little shapes or bits into attractive arrangements, collaborating. But I think we should all concern ourselves that collaboration has two meaning:
The first, productive and good, is an emphatic, visible, proclaimed part of the digital humanities. It involves bringing together disparate scholars and experts into a single project to develop an amazing piece of synthetic scholarship that will revolutionize (and perhaps create) multiple fields of scholarship. I met with Glen Worthey, the Digital Humanities Lorax of Evidenciary Bias, and discussed how we could work together to make digital humanities support at Stanford even more successful and amazing and revolutionary. Collaboration revolutionary! Then I spoke with Ursula Heise about a paper presenting our work analyzing the cultural content of the IUCN Red List database, and we discussed how we could bring even more amazing digital and environmental humanities-oriented scholarship to the fore. Collaboration amazing! And then I met with Karl Grossner and discussed the final, perfect pieces of our nearly complete and triumphant collaborative endeavor with Walter Scheidel (and a cast of thousands!). Collaboration triumphant!
The second meaning of collaboration is most commonly associated with the Nazi occupation of France. There, collaborators were to be resisted and collaboration was the opposite of amazing, revolutionary triumph. I have no stories to describe the darker side of collaboration in the digital humanities, because we don’t talk about how we accede to demands to grant PI status to recognizable names who contribute nothing to a project just so that the grant goes through, or how we speak highly of work done that isn’t really that impressive because we don’t want to harm our colleagues or damage our field, nor do we criticize the review processes of our most visible conferences and grants, lest we damage our own reputations within this very, very small community. Oh, it’s collaboration, alright.
Presenting two sides like that implies that it’s a 50/50 split at best, and a pessimist would see it as weighted toward the seemier side, which appeals to the romantic in me, though I have to admit to a rather Heidi-esque experience in the digital humanities. My position isn’t grant-driven, so grant horror stories are all things that I’ve heard and not experienced, which may or may not be a good way to get a sense of the phenomenon. I’ve also been purposefully curmudgeonly in my call for peer review of digital scholarship, and that’s grown more and more a visible topic over the years. On top of that, embedded in a community like Stanford where there are enough people doing significant DH work, I’m privy to the informal channels that allow for serious critical engagement with my methods and output, which likely is going on throughout DH. Still, I can’t shake the feeling that the less utopian form of collaboration is too prominent (whether in DH in particular, or just the modern university, I can’t tell) and it would be good for us to consider some form of La Résistance to face it.
This morning I gave a presentation on the role of data visualization in DH work. The annotated slides can be found on Google Docs here.
An attempt to illustrate something that may be better expressed through a less traditional visualization.
