Is digital archaeology part of the digital humanities?

This isn’t to get into another who’s in/who’s out conversation. Rather, I was thinking about the ways archaeologists use computing in archaeology, and to what ends. The Computer Applications in Archaeology Conference has been publishing proceedings since 1973, or longer than I’ve been on this earth. Archaeologists have been running simulations, doing spatial analysis, clustering, imaging, geophysicing, 3d modeling, neutron activation analyzing, x-tent modeling , etc, for what seems like ages.

Surely, then, digital archaeologists are digital humanists too? Trevor Owens has a recent post that sheds useful light on the matter. Trevor draws attention to the purpose behind one’s use of computational power – generative discovery versus justification of an hypothesis. For Trevor, if we are using computational power to deform our texts, we are trying to see things in a new light, new juxtapositions, to spark new insight. Ramsay talks about this too in Reading Machines (2011: 33), discussing the work of Jerome McGann and Lisa Samuels. “Reading a poem backward is like viewing the face of a watch sideways – a way of unleashing the potentialities that altered perspectives may reveal”. This kind of reading of data (especially, but not necessarily, through digital manipulation), does not happen very much at all in archaeology. If ‘deformance’ is a key sign of the digital humanities, then digital archaeologists are not digital humanists. Trevor’s point isn’t to signal who’s in or who’s out, but rather to draw attention to the fact that:

When we separate out the the context of discovery and exploration from the context of justification we end up clarifying the terms of our conversation. There is a huge difference between “here is an interesting way of thinking about this” and “This evidence supports this claim.”

This, I think, is important in the wider conversation concerning how we evaluate digital scholarship. We’ve used computers in archaeology for decades to try to justify or otherwise connect our leaps of logic and faith, spanning the gap between our data and the stories we’d like to tell. A digital archaeology that sat within the digital humanities would worry less about that, and concentrate more on discovery and generation, of ‘interesting way[s] of thinking about this’.

In a paper on Roman social networks and the hinterland of the city of Rome, I once argued (long before I’d ever heard the term digital humanities) that we should stop using GIS displaying North at the top of the map, that this was hindering our ability to see patterns in our data. I turned the map sideways – and it sent a murmur through the conference room as east-west patterns, previously not apparent, became evident. This, I suppose, is an example of deformation. Hey! I’m a digital humanist! But other digital work that I’ve been doing does not fall under this rubric of ‘deformation’.

My Travellersim simulation for instance uses agent based modeling to generate territories, and predict likely interaction spheres, from distributions of survey data. In essence, I’m not exploring but trying to argue that the model accounts for patterns in the data. This is more in line with what digital archaeology often does.

Bill Caraher, I suspect, has been reading many of the same things I have been lately, and has been thinking along similar lines. In a post on archaeological glitch art Bill has been changing file extensions to fiddle about in the insides of images of archaeological maps, then looking at them again as images:

“The idea of these last three images is to combine computer code and human codes to transform our computer mediate image of archaeological reality in unpredictable ways. The process is remarkably similar to analyzing the site via the GIS where we take the “natural” landscape and transform it into a series of symbols, lines, and text. By manipulating the code that produces these images in both random and patterned ways, we manipulate the meaning of the image and the way in which these images communicate information to the viewer. We problematize the process and manifestation of mediating between the experienced landscape and its representation as archaeological data.”

In the same way, Trevor uses augmented reality smartphone translation apps set to translate Spanish text into English, but pointed at non Spanish texts. It’s a bit like Mark Sample’s Hacking the Accident, where he uses an automatic dictionary substitution scheme (n+7, a favorite of the Oulipo group) to throw up interesting juxtapositions. A deformative digital archaeology could follow these examples. Accordingly, here’s my latest experiment along these lines.

Let’s say we’re interested in the evolution of amphorae types in the Greco-Roman world. Let’s go to the Netlogo models library, and instead of building the ‘perfect’ archaeological model, let’s select one of their evolutionary models – Wilensky’s ‘Mimicry‘ model, which is about the evolution of Monarch and Viceroy butterflies swapping in ‘amphora’ for ‘moth’ everywhere in the code and supporting documentation, and ‘Greeks’ for ‘birds’.

In the original model code, we are told:

“Batesian mimicry is an evolutionary relationship in which a harmless species (the mimic) has evolved so that it looks very similar to a completely different species that isn’t harmless (the model). A classic example of Batesian mimicry is the similar appearance of monarch butterfly and viceroy moths. Monarchs and viceroys are unrelated species that are both colored similarly — bright orange with black patterns. Their colorations are so similar, in fact, that the two species are virtually indistinguishable from one another.

The classic explanation for this phenomenon is that monarchs taste desireable. Because monarchs eat milkweed, a plant full of toxins, they become essentially inedible to butterflies. Researchers have documented butterflies vomiting within minutes of eating monarch butterflies. The birds then remember the experience and avoid brightly colored orange butterfly/moth species. Viceroys, although perfectly edible, avoid predation if they are colored bright orange because birds can’t tell the difference.

This is what you get:

We have two types of amphorae here, which we are calling the ‘monarch’ type (type 1) and the ‘viceroy’ type (type 2).

This model simulates the evolution of monarchs and viceroys from distinguishable, differently colored types to indistinguishable mimics and models. At the simulation’s beginning there are 450 type 1s and type 2s distributed randomly across the world. The type 1s are all colored red, while the type 2s are all colored blue. They are also distinguishable (to the human observer only) by their shape: the letter “x” represents type 1s while the letter “o” represents type 2s. Seventy-five Greeks are also randomly distributed across the world.

When the model runs, the Greeks and amphorae move randomly across the world. When a Greek encounters a amphora it rejects the amphora, unless it has a memory that the amphora’s color is “desireable.” If a Greek consumes a monarch, it acquires a memory of the amphora’s color as desirable.

As amphorae are consumed, they are regenerated. Each turn, every amphora must pass two “tests” in order to reproduce. The first test is based on how many amphorae of that species already exist in the world. The carrying capacity of the world for each species is 225. The chances of regenerating are smaller the closer to 225 each population gets. The second test is simply a random test to keep regeneration in check (set to a 4% chance in this model). When a amphora does regenerate it either creates an offspring identical to itself or it creates a mutant. Mutant offspring are the same species but have a random color between blue and red, but ending in five (e.g. color equals 15, 25, 35, 45, 55, 65, 75, 85, 95, 105). Both monarchs and Viceroys have equal opportunities to regenerate mutants.

Greeks can remember up to MEMORY-SIZE desireable colors at a time. The default value is three. If a Greek has memories of three desireable colors and it encounters a monarch with a new desireable color, the Greek “forgets” its oldest memory and replaces it with the new one. Greeks also forget desireable colors after a certain amount of time.

And when we run the simulation? Well, we’ve decided that one kind of amphora is desireable, another kind is undesireable. The undesireable ones respond to (human) consumer pressure and change their color; over time they evolve to the same color. Obviously, we’re talking as if the amphorae themselves have agency. But why not? (and see Godsen, ‘What do objects want?’) That’s one interesting side effect of this deformation.

As I haven’t changed the code, so much as the labels, the original creator’s conclusions still seem apt:

Initially, the Greeks don’t have any memory, so both type 1 and type 2 are consumed equally. However, soon the Greeks “learn” that red is a desireable color and this protects most of the type 1s. As a result, the type 1 population makes a comeback toward carrying capacity while the type 2 population continues to decline. Notice also that as reproduction begins to replace consumed amphorae, some of the replacements are mutants and therefore randomly colored.

As the simulation progresses, Greeks continue to consume mostly amphorae that aren’t red. Occasionally, of course, a Greek “forgets” that red is desireable, but a forgetful Greek is immediately reminded when it consumes another red type 1. For the unlucky type 1 that did the reminding, being red was no advantage, but every other red amphora is safe from that Greek for a while longer. Type 1 (non-red) mutants are therefore apt to be consumed. Notice that throughout the simulation the average color of type 1 continues to be very close to its original value of 15. A few mutant type 1s are always being born with random colors, but they never become dominant, as they and their offspring have a slim chance for survival.

Meanwhile, as the simulation continues, type 2s continue to be consumed, but as enough time passes, the chances are good that some type 2s will give birth to red mutants. These amphorae and their offspring are likely to survive longer because they resemble the red type 1s. With a mutation rate of 5%, it is likely that their offspring will be red too. Soon most of the type 2 population is red. With its protected coloration, the type 2 population will return to carrying capacity.

The swapping of words makes for some interesting juxtapositions. ‘Protects’, from ‘consumption’? This kind of playful swapping is where the true potential of agent based modeling might lie, in its deformative capacity to make us look at our materials differently. Trying to simulate the past through ever more complicated models is a fool’s errand. A digital archaeology that sat in the digital humanities would use our computational power to force us to look at the materials differently, to think about them playfully, and to explore what these sometimes jarring deformations could mean.

—–

Godsen, Chris. 2005. ‘What do objects want?’ Journal of Archaeological Method and Theory 12.3 DOI: 10.1007/s10816-005-6928-x

Ramsay, Stephen. 2011. Reading Machines. Towards An Algorithmic Criticism. U of Illinois Press.

Wilensky, U. (1997). NetLogo Mimicry model. http://ccl.northwestern.edu/netlogo/models/Mimicry. Center for Connected Learning and Computer-Based Modeling, Northwestern University, Evanston, IL.

Wilensky, U. (1999). NetLogo. http://ccl.northwestern.edu/netlogo/. Center for Connected Learning and Computer-Based Modeling, Northwestern University, Evanston, IL.