Reconstruction, visualization and frontier archaeology

Recently on holiday in the North East, I took in two Roman forts of the frontier of Hadrian’s Wall, Segedunum and Arbeia. Both have stories to tell, narratives, about the Roman occupation of Britain, and in the current period both have been curated in various ways. At both, the curating authorities (Tyne and Wear Museums), with ongoing archaeological research being undertaken by the fantastic WallQuest community archaeology project.

The public walkthrough reconstructions of what the buildings and the contents might have been like at both sites pose some interesting questions about the nature of historical/archaeological narratives, and how they can be elaborated. At Segedunum, there is a reconstruction of a bath house. Although the fort itself had such a structure, modern development means that it is not in the same place, nor does the foundations of the reconstruction relate directly to archaeological evidence. The features of the bath house are drawn from composite analysis of bath houses from throughout the Roman Empire. So what we have here is a narrative, but it is a generic narrative: it is stitched together, generalized, a mosaic of hundreds of disparate narratives, but it can only be very loosely constrained by time (a bath house such as that at Segedunum would have had a lifespan of 250-300 years), and not to any one individual. we cannot tell the story of any one Roman officer or auxiliary solider who used it.

Image
Reconstructed bath house at Segedunum

On the other hand at Arbeia, there are three sets of granaries, the visible foundations all nicely curated and accessible to the public. You can see the stone piers and columns that the granary floors were mounted on, to allow air movement to stop the grain rotting. Why three granaries for a fort of no more than 600 occupants? Because in the third century, the Emperor Severus wanted to conquer the nearby Caledonii; and for his push up into Scotland we needed a secure supply base with plenty of grain.

Image
Granaries at Arbeia, reconstructed West Gatehouse in the background

This is an absolute narrative. It is constrained by actual events which are historical and documented. At the same fort is a reconstructed gateway, which is this time situated on actual foundations. This is an inferential narrative, with some of the gateway’s features being reconstructed again from composite evidence from elsewhere (did it have two or three stories? A shingled roof? We don’t know, but we infer). These narratives are supported by annotated scale models in the gateway structure which we, they paying public (actually Arbeia is free), can view and review at our leisure. This speaks to the nature of empirical, inferential and conjectural reconstruction detailed in a forthcoming book chapter by myself and Kirk Woolford (of contributions to the EVA conference, published by Springer).

Narratives are personal, but the can also be generic. In some ways this speaks back to the concept of the Deep Map (see older posts). The walkthrough reconstruction constitutes, I think, half a Deep Map. It provides a full sensory environment, but is not ‘scholarly’ in that it does not elucidate what it would have been like for a first or second century Roman, or auxiliary soldier to experience the environment. Maybe the future of 3D visualization should be to integrate modelling, reconstruction, remediation, and interpretation to bring available (and reputable) knowledge from whatever source about what that original sensory experience would have been – texts, inscriptions, writing tablets, environmental archaeology, experimental archaeology etc. In other words, visualization should no longer be seen as a means of making hypothetical visual representations of what the past might of been, but of integrating knowledge about the experience of the environment derived from all five senses, but using vision as the medium.  It can never be a total representation incorporating all possible experiences under all possible environmental conditions, but then a map can never be a total representation of geography (except, possibly, in the world of Borges’s On the Exactitude of Science).

Last day in Indiana

It’s my last day in Indianapolis. It’s been hard work and I’ve met some great people. I’ve experienced Indianapolis’s hottest day since 1954, and *really* learned to appreciate good air conditioning. Have we, in the last two weeks, defined what a deep map actually is? In a sense we did, but more importantly than the semantic definition, I reckon we managed to form a set of shared understandings, some fairly intuitive, which articulate (for me at least) how deep mapping differs from other kinds of mapping. It must integrate, and at least some of this integration must involve the linear concepts of what, when and where (but see below). It must reflect experience at the local level as well as data at the macro level, and it must provide a means of scaling between them. It must allow the reader (I hereby renounce the word ‘user’ in relation to deep maps) to navigate the data and derive their own conclusions. Unlike a GIS – ‘so far so Arc’ is a phrase I have co-coined this week – it cannot, and should not attempt to, actualize every possible connection in the data, either implicitly or explicitly. Above all, a deep map must have a topology that enables all these things, and if, in the next six months, the Polis Center can move us towards  a schema underlying that topology, then I think our efforts, and theirs, will have been well rewarded.

The bigger questions for me are what does this really mean for the ‘spatial humanities’; and what the devil are the spatial humanities anyway. They have no Wikipedia entry (so how can they possibly exist?). I have never particularly liked the term ‘spatial turn’, as it implies a setting apart, which I do not think the spatial humanities should be about. The spatial humanities mean nothing if they do not communicate with the rest of the humanities, and beyond. Perhaps – and this is the landscape historian in me talking – it is about the kind of topology that you can extract from objects in the landscape itself. Our group in Week 2 spent a great deal of time thinking about the local and the experiential, and how the latter can be mapped on to the former, in the context of a particular Unitarian ministry in Indianapolis. What are the stories you can get from the landscape, not just tell about it.

Allow me to illustrate the point with war memorials. The city’s primary visitor information site, visitindy.com, states that Indianapolis has more war memorials than any city apart from Washington D.C.. Last Saturday, a crew of us hired a car and visited Columbus IN, an hour and a half’s drive away. In Columbus there is a memorial to most of America’s wars: eight by six Indiana limestone columns, arranged in a close grid formation with free public access from the outside. Engraved on all sides of the columns around the outside, except the outer facing edges, are names of the fallen, their dates, and the war in which they served. On the inner columns– further in, where you have to explore to find them, giving them the mystique of the inner sanctum – are inscribed the full texts of letters written home by fallen servicemen. In most cases, they seem to have been written just days before the dates of death.  The deeply personal natures of these letters provide an emotional connection, and combined with the spatiality of the columns, this connection forms a very specific, and very deliberately told, spatial narrative. It was also a deeply moving experience.

Today, in Indianapolis itself, I was exploring the very lovely canal area, and came across the memorial to the USS Indianapolis. The Indianapolis was a cruiser of the US Navy sunk by Japanese torpedoes in 1945, with heavy loss of life. Particular poignancy is given to the memorial by a narrative of the ship’s history, and the unfolding events leading up to the sinking, inscribed in prose on the monument’s pedestal. I stood there and read it, totally engrossed and as moved by the story as I was by the Columbus memorial.

Image
USS Indianapolis memorial

The point for deep maps: American war memorials tell stories in a very deliberate, designed and methodical way, to deeply powerful effect in the two examples I saw. British war memorials tend not to do this. You get a monument, lists of names of the fallen and the war in question, and perhaps a motto of some sort. An explicit story is not told. This does not make the experience any less moving, but it is based on a shared and implicit communal memory, whose origins are not made explicit in the fabric of the monument. It reflects a subtle difference in how servicemen and women are memorialized, in the formation of the inherently spatial stories that are told in order to remember them.

This is merely one example of subtle differences which run through any built environment of any period in any place, and they become less subtle as you scale in more and more cultures with progressively weaker ties. Britain and America. Europe, Britain and America. Europe, America and Africa, and so on. We scale out and out, and then we get to the point where the approaches to ‘what’ ‘when’ and ‘where’ – the approaches that we worked on in our group – must be recognised not as universal ways of looking at the world, but as products of our British/American/Australian backgrounds, educations and cultural memories. Thus it will be with any deep map.

How do we explain to the shade of Edward Said that by mapping these narratives we are not automatically claiming ownership of them, however much we might want or try not to? How deep will these deep maps need to go…?

Deep maps in Indy

I am here in a very hot and sunny Indianapolis trying to figure out what is meant by deep mapping, with an NEH Summer Institute at UIPUI hosted by the Polis Center here. There follows a very high-level attempt to synthesize some thoughts from the first week.

Deep mapping – we think, although we’ll all probably have changed our minds by next Friday, if not well before  – is about representing (or, as I am increasingly preferring to think, remediating) the things that Ordnance Survey would, quite rightly, run a perfectly projected and triangulated mile from mapping at all. Fuzziness. Experience. Emotion. What it means to move through a landscape at a particular time in a particular way. Or, as Ingold might say, to negotiate a taskscape. Communicating these things meaningfully as stories or arguments. There has been lots of fascinating back and forth about this all week, although – and this is the idea at least – next week we move a beyond the purely abstract and grapple with what it means to actually design one.

If we’re to define the meaning we’re hoping to build here, it’s clear that we need to rethink some pretty basic terms. E.g. we talk instinctively about ‘reading’ maps, but I have always wondered how well that noun and that verb really go together. We assume that ‘deep mapping’ for the humanities – a concept which we assume will be at least partly online – has to stem from GIS, and that a ‘deep map, whatever we might end up calling that, will be some kind of paradigm shift beyond ‘conventional’ computer mapping. But the ’depth’ of a map is surely a function of how much knowledge – knowledge rather than information – is added to the base layer, where that information comes from, and how it is structured. The amazing HGIS projects we’ve seen this week give us the framework we need to think in, but the concept of ‘information’ therein should surely be seen as a starting point. The lack of very basic kinds of such information in popular mapping applications has been highlighted, and perhaps serves to illustrate this point. In 2008, Mary Spence, President of the British Cartographic Society, argued in a lecture:

Corporate cartographers are demolishing thousands of years of history, not to mention Britain’s remarkable geography, at a stroke by not including them on [GPS] maps which millions of us now use every day. We’re in danger of losing what makes maps so unique, giving us a feel for a place even if we’ve never been there.

To put it another way, are ‘thin maps’ really all that ‘thin’, when they are produced and curated properly according to accepted technical and scholarly standards? Maps are objects of emotion, in a way that texts are not (which is not to deny the emotional power of text, it is simply to recognize that it is a different kind of power). Read Mike Parker’s 2009 Map Addict for an affectionate and quirky tour of the emotional power of OS maps (although anyone with archaeological tendencies will have to grit their teeth when he burbles about the mystic power of ley lines and the cosmic significance of the layout of Milton Keynes). According to Spence, a map of somewhere we have never been ties together our own experiences of place, whether absolute (i.e. georeferenced) or abstract, along with our expectations and our needs. If this is true for the lay audiences of, say the Ordnance Survey, isn’t the vision of a deep map articulated this past week some sort of scholarly equivalent? We can use an OS map to make a guess, an inference or an interpretation (much discussion this week has, directly or indirectly, focused on these three things and their role in scholarly approaches). What we cannot do with an OS map is annotate or embed it with any of these. The defining function of a deep map, for me, is an ability to do this, as well as the ability to structure the outputs in a formal way (RDF is looking really quite promising, I think – if you treat different mapped objects in the object-subject-predicate framework, that overcomes a lot of the problems of linearity and scale that we’ve battled with this week). The different levels of ephemerality that this would mean categorising (or, heaven help us, quantifying) is probably a story for another post, but a deep map should be able to convey experience of moving through the landscape being described.

There are other questions which bringing such a map into the unforgiving world of scholarly publication would undoubtedly entail. Must a map be replicable? Must someone else be able to come along and map the same thing in the same way, or at least according to their own subjective experience(s)?  In a live link up the UCLA team behind the Roman Forum project demonstrated their stuff, and argued that the visual is replicable and –relatively easily – publishable, but of course other sensory experiences are not.  We saw, for example, a visualisation of how far an orator’s voice could carry. The visualisation looks wonderful, and the quantitative methodology even more so, but to be meaningful as an instrument in the history of Roman oratory, one would have to consider so many subjective variables – the volume of the orator’s voice (of course), the ambient noise and local weather conditions (especially wind). There are even less knowable functions, such as how well individuals in the crowd could hear, whether they had any hearing impairments etc. This is not to carp –after all, we made (or tried to make) a virtue of addressing and constraining such evidential parameters in the MiPP project, and our outputs certainly looked nothing like as spectacular as UCLA’s virtual Rome – but a deep map must be able to cope with those constraints.

To stand any chance of mapping them, we need to treat such ephemera as objects, and object-orientation seemed to be where our – or at least my – thinking was going last week. And then roll out the RDF…

Digital Ghosts

Here’s a preview of my upcoming talk at the Turing Festival in Edinburgh.

Credit: Motion in Place Platform Project

3D imaging is prevalent in archaeology and cultural heritage. From the Roman forum to Cape Town harbour, from the crypts of Black Sea churches to the castles of Aberdeenshire, 3D computer graphic models of ancient buildings and ancient spaces can be explored, manipulated and flown-through from our desktops. At the same time however, it is a basic fact of archaeological practice that understanding human movement constitutes a fundamental part of the interpretive process, and of any interpretation of a site’s use in the past. Yet most of these digital reconstructions, and the ones we see in archaeological TV programmes, in museums, in cultural heritage sites and even in Hollywood movies, tend to focus on the architecture, the features, and the physical surroundings. It is almost paradoxical that the major thing missing from many of our attempts to reconstruct the human past digitally are humans. This can be traced to obvious factors of preservation and interpretation: buildings survive, people don’t. However, this has not stopped advances in 3D modelling, computer graphics and web services to support 3D images from drawing archaeologists and custodians of cultural heritage further and further into the 3D world, and reconstructing ancient 3D environments in greater and greater detail. But the people are still left behind. This talk will reflect on the Motion in Place Platform (MiPP) project, which seeks to use motion capture hardware and data to test human responses and actions within VR environments, and their real-world equivalents. Using as a case study domestic spaces – roundhouses of the Southern British Iron Age – it used motion capture to compare human reaction and perception in buildings reconstructed at 1:1 scale, with ‘virtual’ buildings projected on to screens. This talk will outline the experiment, what might be learned from it, and how populating our 3D views of the past with ‘digital ghosts’ can also inform them, and make them more useful for drawing inferences about the past.

End of project MiPP workshop

At the closing MiPP project in Sussex last week. Due to a concatenation of various cirumstances, I had to take a large broomstick, which will be used in next week’s motion capture exercises at Butser Farm and in Sussex, on a set of trains from Reading, via the EVA London 2011 conference in Central London, to the workshop in Falmer, Sussex. Given this thing is six feet tall and required its own train seat (see picture), I got a variety of looks from my fellow passengers, especially on the Underground, ranging from suspicion to pity to humour. Imagined how one might have handled a conversation: ‘There’s a logical explanation. Yes, it’s going to be used as a prop in an experiment to test the environment of Iron Age round houses in cyberspace versus the real thing in the present day.’ ‘Oh yes? And that’s your idea of a logical explanation is it?’

Of course I could have really freaked people out be getting off the train at Gatwick Airport and wandering around the terminal, asking for directions to the runway.

As with the entire MiPP project, the workshop was highly interdisciplinary. A varied set of presentations included ones from Bernard Frisher of the University of Virginia, on digital representation of sculpture, and from colleagues at Southampton on the fantastic PATINA project. All of which coalesced  around questions of process, and how we represent it. Tom Frankland’s presentation on studying archaeological processes, including such offsite considerations as the difference between note taking in the lab and in the field, filled in numerous gaps of documentation that our work at Silchester last summer left.

When I got to my feet on day to two present, I veered slightly off my promised topic (as with most presentations I have ever given) and elected instead to reflect on the nature remediated archaeological objects. I would suggest that there is a three-way continuum on which any digital object representing an archaeological artefact or process may be plotted: the empirical, the interpretive and the conjectural. An empirical statement, such as Dr. Peter Reynolds, the founder of Butser Farm would have approved, might state that ‘the inner ring of this round house comprised of twelve upright posts, because we can discern twelve post holes in ring formation’.  An interpretative conclusion might be built on top of this stating that, because ceramic sherds were found in the post hole, cooking and/or eating took place near to this inner ring. This could in turn lead to a conjecture that a particular kind of meat was cooked in a particular way at this location, based not on interpretation or empirical evidence immediately to hand, but on the general context of the environment, and on what is known more broadly about Iron Age domestic practice.

More on all this next week, after capture sessions at Butser.

MiPP: Forming questions

The question about our MiPP project which I’m most often asked is ‘why?’ In fact that this is the whole project’s fundamental research question. As motion capture technologies become cheaper, more widely available, less dependent on equipment in fixed locations such as studios, and less dependent on highly specialist technical expertise to set them up and use them, what benefits can these technologies bring outside their traditional application areas such as performance and medical practice? What new research can they support? In such a fundamentally interdisciplinary project, there are inevitably several ‘whys’, but as someone who is, or at least once was, an archaeologist, archaeology is the ‘why’ that I keep coming back to. Matters became a lot clearer, I think, in a meeting we had yesterday with some of the Silchester archaeological team.

As I noted in my TAG presentation before Christmas, archaeology is really all about the material record: tracing what has survived in the soil, and building theories top of that. Many of these theories concern what people did, and where and how they moved while they were doing them. During a capture session in Bedford last week (which alas I couldn’t attend), the team tried out various scenarios in the Animazoo mocap suits, using the 3D Silchester Round House created by Leon, Martin and others as a backdrop. They reconstructed in a practical way how certain every day tasks might have been accomplished by the Iron Age inhabitants. As Mike Fulford pointed out yesterday, such reconstructions – which are not reconstructions in the normally accepted sense in archaeology, where the focus is usually on the visual, architectural and formal remediation of buildings (as excellently done already by the Silchester project) – themselves can be powerful stimuli for archaeological research questions. He cited a scene in Kevin Macdonald’s The Eagle, where soldiers are preparing for battle. This scene prompted the reflection that a Roman soldier would have found putting on his battle dress a time consuming and laborious process, a fact which could in turn be pivotal to the interpretation of events surrounding various aspects of Roman battles.

One aim of MiPP is to conceptualize theoretical scenarios such as this as visual data comprising digital motion traces. The e-research interest in this is that those traces cannot really be called ‘data’, and cannot be useful in the particular application area of reconstructive archaeology, if their provenance is not described, or if they are not tagged systematically and stored as retrievable information objects. What we are talking about, in other words, is the mark-up of motion traces in a way that makes them reusable. Our colleagues in the digital humanities have been marking up texts for decades. The TEI has spawned several subsets for specific areas, such as EpiDoc for marking up epigraphic data, and mark-up languages for 3D modelling (e.g. VRML) are well developed. Why then should there not be a similar schema for motion traces? Especially against the background of a field such as archaeology, where there are already highly developed information recording and presentation conventions, marking up quantitative representations of immaterial events should be easy. One example might be to assign levels of certainty to various activities, in much the same way that textual mark-up allows editors to grade the scribal or editorial certainty of sections of text. We could then say, for example, that ‘we have 100% certainty that there were activities to do with fire in this room because there is a hearth and charring, but only 50% certainty that the fire was used for ritual activity’. We could also develop a system for citing archaeological contexts in support of particular types of activity; in much the same way that the LEAP project cited Silchester’s data in support of a scholarly publication. It boils down to the fundamental principle of information science, that an information object can only be useful when its provenance is known and documented. How this can be approached for motion traces of what might have happened at Silchester in the first century AD promises to be a fascinating case study.