Refractions

In 2006, a camera installed in the atrium of Louisiana State University\'s College of Art and Design to monitor a high-contrast mural performed continuous blob detection, identifying regions of contrast difference between the mural and pedestrians moving through the space. Custom software converted those differences into real-time isolines displayed on an adjacent monitor. As people moved through the atrium, they became topographic events and their movement generated ephemeral landforms.

The installation was called Thresholds and what it revealed was not what it was designed to demonstrate. The contour line, the foundational notation of topographic representation, is not a static notation, and in the exhibition it was a reading between a sensing instrument and a calibrated context. Without the mural as datum, the isolines would have been noise. Change the datum and a different landscape emerges and controlling the datum, decides in advance, what the landscape will reveal.

Practice-Based Research

This dissertation follows practice-based methods of research as that is how this form of knowledge is produced, not by applying theory to practice, but by building, deploying, and interrogating propositions in material and spatial form.

The work develops a theoretical framework for landscape architecture operating at the territorial scale, using my creative practice and scholarship not as case studies of existing theory, but instead as a foundation to test, generate, and evolve a comprehensive theory on adaptive management that integrates ecological thinking and computational methods. The theoretical chapters, case studies, and reflections that are developed within the dissertation aim to make the practice itself legible as research. In that sense, the dissertation conforms to a recognized structure of a practice-based PhD, which is a substantial body of scholarship, creative practice, and technical prototypes paired with an expanded critical interpretation, where the written component functions as an epistemic bridge between my practice and publicly verifiable knowledge (Frayling 1993; MacLeod 2000). Together, these methods generate five interlocking frameworks: adaptive epistemology, technogeographies of sensing, the cultivant, synthetic ground, and reflexive stewardship that are developed across the theoretical and project chapters that follow.

This methodological frame sits within a broader conversation on practice-based and practice-led research in art, design, and architecture. As a starting point, Frayling\'s distinction between research \'into,\' \'for,\' and \'through\' art and design provides a useful orientation as research into design analyzes a practice from the outside, research for design develops tools or knowledge that support practice, and research through design uses design itself as the primary medium of inquiry (Frayling 1993). I am extending this argument by positing that, in practice-led research, creative practice is both the field of inquiry and the means of communicating the research, rather than a secondary illustration (Rust, Mottram, and Till 2007).

The dissertation aligns most strongly with research through practice. Projects, models, interfaces, and collaborations are not appendices to the research, they are where the research happens. The critical interpretation then situates that practice within the discourses of landscape architecture, engineering, and environmental design (Deming and Swaffield 2011; Swaffield and Deming 2011). Therefore, the dissertation works to demonstrate how the practice functions as research, and to show that the reflections and theoretical scaffold satisfies the expectations of rigor, relation, and contribution that apply to doctoral level work (MacLeod 2000; Rust, Mottram, and Till 2007).

Several projects appear in more than one chapter. Algorithmic Cultivation, the NEOM consultation, the UVA Geomorphology Lab, Designing Autonomy, Prototyping the Bay, and Indeterminate Futures each resurface across chapters because the questions this dissertation asks are not answered by any single project read from a single angle. Each chapter encounters the same practice from a different vantage point and what adaptive management reveals about the geomorphology table differs from what co-authorship reveals about it, which differs from what the politics of sensing reveals about it. This is not redundancy, it is purposeful and it is the methodology.

Gilbert Ryle called it \"systematic restatement\" (Ryle 1954) where the same object that is described in different vocabularies, with each vocabulary revealing what the others obscure. The practice operates the same way. The reader who recognizes a project from an earlier chapter is not re-reading. They are seeing it refracted with the same facets, but with different light.

Refraction is the primary methodological contribution of this dissertation. The term names a systematic practice of retelling, passing the same body of work through different narrative media and reading each project from vantage points that its instrumental contexts never demanded. Grant narratives, book chapters, competition briefs, and consulting deliverables each required a specific framing of the work, what it does, why it matters, why it should be funded or built or published. Those framings were true but partial. They served the institutional contexts that made the work possible and in doing so they constrained what could be seen.

Refraction does not replace those framings. It holds them alongside new ones, and finds, in the gap between the instrumental account and the doctoral retelling, properties of the work that were always present but traveling invisibly within the original framing. This is not a method of critique. It is a method of recovery.

The Practice Research Symposia, six biannual sessions held between 2020 and 2023 with a seventh supplementary session, were the mechanism through which refraction first became systematic and named. At each session, the same body of work was presented to the committee and a panel of external respondents. Each time, the story was told slightly differently. The projects changed minimally. The angle of inquiry changed. And from each new angle, properties emerged that no single instrumental context could have made visible. The PRS did not produce the method. It produced the conditions under which the method became legible to the practitioner conducting it.

But refraction does not end with the PRS. The dissertation itself is its continuation. Each chapter that follows encounters the practice from a different theoretical vantage, adaptive management, sensing politics, multi-species authorship, generational autonomy, and each encounter is a refraction. Projects reappear not because the argument requires repetition but because the same body of work, seen from a new angle, yields new knowledge. The reader who recognizes Algorithmic Cultivation from an earlier chapter is not re-reading. They are watching the method work.

Refraction is appropriate to this practice because the knowledge it produces cannot be recovered by adding more data or conducting additional fieldwork. It is already present in the work. What it requires is a sustained, disciplined shift in the narrative medium through which the work is examined. A practice that has been building, deploying, and interrogating propositions across twenty years contains more knowledge than any single framing has been able to name. The doctoral inquiry creates the conditions in which that excess becomes legible.

Richard Blythe\'s written response to PRS 3 named this structure directly. Where previous respondents had engaged the work\'s content, Blythe named its architecture. \"I now see two registers to your PhD,\" he wrote, \"and I think this is also what Claudia and others were pointing at.\" The first register was the experiment, the concrete project with its own internal logic, its accumulated technical complexity, its documentation and demonstrable outcomes. The second was the speculation, the generative matrix of propositions from which the experiment had emerged and which the experiment, in its institutional life, had partially obscured. The instrumental framing, grant, competition, consultation, had served the experiment. It had contained the speculation. Refraction, as a systematic method, works precisely at that boundary, recovering what the instrumental context suppressed not by discarding the framing but by holding it alongside a new one. The tension between the two registers is not a problem to be resolved. As Blythe put it, the task was \"to document both, to point to discoveries in both, and to show how the combination of registers goes beyond existing knowledge in the field.\"

Paul Kelsch\'s response at PRS 4 worked on the same territory from a different angle. \"It\'s a pretty brave thing to do,\" he said, \"to say: here\'s what I\'ve been telling myself and my colleagues in the world about what this work is about. And yet if I\'m honest with myself it\'s only a certain amount truth and a certain amount lies and in the combination of those two is where the interest is.\" The phrase is precise in a way that \"refraction\" itself struggles to be.

The instrumental framings were not wrong. They were partial, and the partiality was functional. A grant narrative that told the full story of what a project was epistemologically producing would not have been funded. A competition brief that foregrounded knowledge production over formal proposition would not have won. A consulting deliverable that led with theoretical uncertainty would not have been commissioned. Each framing performed the work\'s legitimacy within the institutional context that made the work possible. The partial framing is not a failure of honesty but the condition of practice itself.

What refraction does is not expose the lie. It holds the partial truth of the original framing alongside the fuller claim the doctoral inquiry can now make, and it finds, in the gap between them, exactly what Kelsch called \"where the interest is.\" The combination is not a correction. It is the argument.

Marcelo Stamm\'s intervention at PRS 6 named something the other refractions had approached but not quite articulated. \"What is it about origins that is important,\" he asked, \"that all of a sudden potentialities become apparent again, the virtual space of possibilities that sits with the origins, with the inception space. Where there are all sorts of vectors in which this could go.\" The question was not about the finished projects but about the moment before the instrumental framing closes, the inception point where the work could still have gone in any direction. What the PRS process had been doing, across six sessions, was returning to that inception space repeatedly, not to change what the work became but to recover what else it could have been and to find, in those unrealized directions, the epistemological structure the completed projects had made legible. The vectors were always there. The doctoral inquiry named them.

Research-Through-Design / Research-Through-Designing

The methodological backbone of the dissertation is research through design, or, in the specific case of this dissertation, research-through-designing. In their framework for landscape architectural research, Deming and Swaffield (2011) treat design as one of the three core research strategies that function in parallel to empirical inquiry and interpretive analysis and they emphasize that design has the capacity to generate hypotheses in spatial and material form regarding how landscapes work and their evolution. Lenzholzer, Duchhart, and Koh (2013) also define \"research through designing\" as a method that actively employs design with explicit and rigorous procedures to address spatio-physical inquiries.

Within the field of interaction design and human computer interfaces, we can formalize research through design as a process where designers create artifacts specifically to gain knowledge in regard to larger disciplinary concerns and the artifacts can embody theoretical propositions and therefore can be critiqued, compared, and applied across projects (Zimmerman, Stolterman, and Forlizzi 2010). Adapted to territorial and ecological landscape questions, this posits that when the research speculates how a landscape system behaves under a nascent interaction, an alternative dredge logistics, a robotic diversion, or an ecological sensing network, I address this by designing, building, and testing spatial and technical propositions. When the work demands the invention of new interfaces, tools, or arrangements rather than the evaluation of existing ones, the dissertation supposes that the most meaningful knowledge will come from the performance of a proposal in models, simulations, or pilot interventions, rather than from discourse alone.

Therefore, visualization methods, physical models, computational tools, sensing regimes, and territorial design studies are the hypotheses embodied in form and their design, deployment, and evaluation are the primary research moves. In contrast to this, if the problem is primarily descriptive in reconstructing the history of institutions, clarifying a theoretical framework, or synthesizing disciplinary approaches the research relies on textual analysis, archival reading, and argument rather than solely relying on research through design. Therefore, each method is employed selectively where design prototyping is the most direct way to interrogate and propose the reconfiguration of complex eco-technical systems.

The political stakes of that epistemological claim became explicit at the 2016 Venice Architecture Biennale in a collaboration with VGL. The exhibition invited designers to develop branding packages for artificial landmasses being physically constructed in the South China Sea through real disputed territories and real geopolitical stakes. The work was satirical, but what it revealed was not.

An artificial island becomes sovereign territory not when sand is dredged and piled but when it appears on official maps, receives a name, and circulates in media imagery. The sensing apparatus determines which features became legible as landscape. The representational apparatus in the exhibition determined which territorial claims became legible as governance. In both cases, the instrument shapes what can be known and what can be known shapes what can managed, governed, or manipulated, you cannot separate the sediment from the image.

Case Study and Comparative Project Analysis

Case-based research is particularly appropriate when research addresses complex, real-world situations where control over variables is impossible but the development of contextual depth is essential (Deming and Swaffield 2011). In my research, specific territories, models, and projects including river models at REAL and the UVA Geomorphology Lab, pedagogical framings of coastal islands in the Chesapeake Bay, and consulting for ecological infrastructures are the sites in which broader questions about adaptive management, synthetic grounds, and technogeographies are defined and tested.

This method is deployed when understanding how a particular configuration of actors, infrastructures, ecologies, and institutions actually operate is critical to speculating on future landscape conditions. The fluvial modeling chapter and discursive cases throughout are necessary when the critique requires situated descriptions of a context, including regulatory structures, historical trajectories, and socio-cultural factors. Therefore a dialectical reading of project situations, collaborations with research labs, design offices, and public agencies allows the research to formulate how methods and concepts mutate or fail under a range of institutional and territorial conditions. In this sense they help to test if the concepts developed within have explanatory or generative value beyond a single site or collaboration and operate both reflexively, on the practice, and outwardly on a wider field of inputs.

Physical and Computational Modeling as Experimental Practice

A majority of the concepts extrapolated in the practice are embedded in dynamic environmental processes within fluvial morphodynamics, sediment deposition/erosion, biological communities, hydrologic performance, and environmental sensor networks. My research unpacks dynamic processes using geomorphological models, material experiments that are in conversation with computational simulations and time-based visualizations.

It is important to state that visualization and modeling in this context is treated not as a tool for prediction but as a heuristic tool that produces design heuristics and aesthetic outcomes. Situating this within the research through design methods, physical and digital models are employed to propose design space, reveal behaviors, and to provoke reflection and discourse among collaborators (Zimmerman, Forlizzi, and Evenson 2007; Lenzholzer, Duchhart, and Koh 2013). As a landscape practice the models are prototype environments that highlight overlapping feedback loops as opposed to fixed representations of physical world, the models are instead the propositions of a speculative future.

At the Responsive Environments and Artifacts Lab at the Harvard Graduate School of Design, a Kinect depth sensor mounted above a geomorphology table, a basin of synthetic sediment and programmable water flow, failed to resolve thin depositional layers. The instrument could not see the sediment at a scale that mattered. The failure was the finding, the sensing apparatus determined what could be known, and what could be known determined what could be designed. Subsequent development shifted to ultrasonic range finders and image analysis, not because the Kinect was defective, but because its limits revealed the question that needed to be asked differently.

This is what practice-based research produces that literature review cannot, knowledge that emerges from friction with material dynamics. The model is not a prediction engine. It is a site where the research encounters resistance, and resistance generates understanding.

Bart Lootsma\'s observation at PRS 5 carried a particular authority. He opened by locating himself, a Dutch architectural theorist who had grown up with a school magazine full of engineering simulations, and whose uncle had planted 40 percent of the Netherlands with trees according to succession plans. He understood, from both directions, what it meant to work with systems that had their own logic. The geomorphology tables, he said, could be understood as \"material computers,\" devices in which \"the computing is partly, of course, already in there.\" He was naming a tradition in which physical models were understood to perform computation through material behavior, water finding its own solution through sediment and slope. What the digital sensing layer added was not computation but legibility, a way of reading what the material had already worked out. Matias del Campo arrived at the same term from a different direction. Where Lootsma came from modeling cultures, del Campo came from assemblage theory, the idea that \"something like an assemblage is going on here between human intervention and self-assembling.\" It was in that assemblage quality that he recognized the material computer, the convergence of human intention and self-organizing matter into something that neither fully controls. Two respondents, two frameworks, the same term. That convergence was itself a refraction, two angles of inquiry arriving at the same property that the original lab-equipment framing had kept invisible.

Dana Cupkova pressed the modeling argument into territory the project documentation alone had not reached. The issue of aesthetics, she argued, does not arrive at the end of the process as an output or a result. \"It occurs within the model itself early on, the way you set up the model.\" The aesthetic ideology is built into the feedback structure from the beginning, into the choice of what to sense, how to visualize, what counts as a meaningful response. This observation reframes the entire sensing apparatus. The instruments are not neutral recording devices that produce aesthetic results downstream. They are aesthetic arguments before the first reading is taken, decisions about what matters embedded in the model\'s architecture. The design of the instrument is already the design of the knowledge it will produce.

Nicholas de Monchaux noted, almost in passing, that \"the model as an agent is really interesting, the model as an autonomous agent.\" The observation was brief but it named a trajectory the chapter had been tracing without quite stating it directly. The model begins as instrument, a tool for seeing. It becomes a site, a place where knowledge is produced rather than confirmed. And then, as the practice moves through responsivity toward autonomy, it begins to act, to make decisions, to generate outcomes that exceed the designer\'s specification. The wildness creator is the model taken to its logical conclusion, a system that has learned enough to operate without the designer in the loop. De Monchaux saw that endpoint implied in the work before the work had fully arrived there.

The research relies on models to examine temporal behavior under controlled, repeatable conditions and the outcomes are synthesized to propose prototypical dynamic interactions. Primarily the model is a space to test responsive systems and the outcomes question how sediment migrates when robotic spillways choreograph new behaviors or how sediment patterns respond to specific protocols within a designed sensing and visualization interface. Models are necessary to create shared experimental spaces where designers, engineers, and scientists can tangibly see and manipulate the same system, physically or digitally. This produces moments where the practice seeks to unpack the limits, thresholds, and failure of systems and rarely are employed to generate an idealized performance.

My collaborations with engineers and scientists make it clear the importance of conventional model validation within observed data, but the practice\'s contribution lies in highlighting how spatial design hypotheses are shaped by and shape the model. Using a specific model there are specific futures that become imaginable and others that begin to fall away, therefore the research exists to make the shaping visible as well as negotiable. The intention of the models employed in my practice is not to advance modeling techniques per se, rather it is to use modeling as a mechanism that pushes design research that is speculative and propositional.

Kelsch\'s most sustained intervention across PRS 4 and 6 returned repeatedly to place. The models seemed abstract, he observed, and yet they were tied to specific conditions, the Mississippi\'s sediment loads, the Atchafalaya spillway, the hydrology of the LA River. He quoted the work back directly: \"the areas act to nudge and guide the River system,\" and \"the new riverbed is the product of interactions that guide outcomes controlled within a range.\" What struck him in those phrases was not the technical claim but the value decision embedded in it. \"How much wildness can we tolerate?\" he asked. \"The sublime isn\'t sublime when it\'s actual terror.\" The geographic specificity of the modeling work was not a limitation to be overcome but the site where those value decisions become legible, where the interaction between designed intervention and material system reveals what a practice actually believes about indeterminacy, risk, and responsibility. Refraction does not replace that specificity with abstraction. It makes visible the values the specificity was carrying all along.

Communities of Practice and Collaborative Inquiry

The research is embedded in an ecology of collaboration among design academics, professional firms, engineering and science labs, agencies, and communities. To create a rigorous framework for knowledge production within this context I rely on theories of communities of practice, which conceptualizes learning as a process of relational participation in a shared discourse. Communities of practice are characterized by mutual engagement, joint enterprise, and a shared repertoire of tools, theories, and narratives (Lave and Wenger 1991; Wenger 1998). This framework is preferred here over actor-network theory because the research interest is in knowledge production within the community rather than the translation of interests between actors.

This ecology of practice is wielded to analyze particular concepts and methods that are formed around responsive infrastructures, wetware, and interfaces and evolve within specific constellations of mutual collaboration. Developing the work of the practice with geomorphologists in the lab, engineering offices, institutional staff in workshops, or with my peers in design speculation creates nascent framings, plural workflows, and varied standards of evidence that may be simultaneously qualitative and quantitative. The communities of practice concepts provide a lexicon for unpacking these variations and for explaining why a portion of the collaborations are creatively fertile while others remain transactional.

The community of practice framework also incorporates a strategic dimension guiding the deliberate formation and sustenance of communities centered around questions of sediment, river processes, and territorial scale climate adaptation. The dissertation therefore privileges collaborations in which I participate in problem definition, experimentation, and reflection, which is methodologically generative. Understandably it is more selective about transactional engagements where design appears as a method of visualization or at the end of an externally defined brief.

Autoethnography and Reflective Practice

Because many significant methodological decisions are made in the heat of practice, the dissertation employs autoethnography and modes of reflective practice as complementary methods.

My goal is to distance myself from this process and attempt to use autoethnography to document and systematically analyze practical experience to understand larger cultural experience (Ellis, Adams, and Bochner 2011). The literature emphasizes that autoethnography acknowledges subjectivity, emotion, and positionality as legitimate sources of knowledge and concepts of the reflective practitioner foregrounds reflection in action and reflection on action to produce forms of disciplinary knowledge (Adams, Holman Jones, and Ellis 2015; Schön 1983). The rigor of autoethnographic accounts in this dissertation is maintained through triangulation with collaborator testimony, project documentation, and published outputs.

I utilize these methods when the research questions are centered on methods and concepts that emerge over time in my practice. The concept of interface, technogeographies, failure, wetware, and adaptive epistemologies never arrives fully gestated, it evolves across multiple projects, speculations, and collaborations, and to capture this evolution I treat the output of my practice as reflective data points. Autoethnography is also relevant when internal processes of skepticism, discomfort with compromise, or enthusiasm with prototyping are revealing about the resistances and catalysts within practice-centered research.

My reflections are not a memoir, instead they are expressed through project moments, moments of decision, and interactions, and are related within experiments, prototypes, competition entries, drawings, reflections, and essays adhering to the literature on practice-based research in pursuit of the PhD (MacLeod 2000; Rust, Mottram, and Till 2007). The goal of which is to clearly situate the conditions under which the work unfolds and to acknowledge the effects of my context among institutional structures, funding regimes, and disciplinary histories.

Supporting Methods: Literature, Documents, and Policy

The dissertation is expressed within a substantial foundation of literature where the conceptual chapters synthesize work in landscape theory, environmental humanities, science and technology studies, political ecology, and media theory. The writing is an essential method functioning primarily as scaffolding for the practice as a complementary research driver. My central claims on models, infrastructures, sensing systems, adaptation, and wetware are generated through engagement with my projects and collaborations, placing it in conversation with scholarship. Methods concerning comprehensive discourse analysis or historiographic reconstruction are used when they illuminate practice questions directly but they remain in the background to center design and environmental interaction.

The writing adheres to a recognizable structure, using the introduction, literature review, methods, project descriptions, and theoretical chapters, and synoptic overview as a way to ensure that the creative activity can be evaluated using the same criteria of clarity and contribution that apply to all dissertations (MacLeod 2000).

The drawing project Failure (Drawing Codes, Pratt Institute, 2019; Chicago Architecture Biennial, 2020), in collaboration with Emma Mendel, accumulated layers over months, code-generated urban forms overlaid with ink spills, chemical transfers, erasures, collaborative interventions. No layer was treated as permanent. A companion film catalogued two centuries of environmental failures, catastrophic industrial failures levee collapses, and restoration projects abandoned when the ecological behavior they were designed around failed to materialize. These were predictive models that were invalidated by the systems they claimed to describe.

The argument was direct, complexity makes certain prediction impossible. Accepting this does not require abandoning design. It requires abandoning the assumption that design\'s role is to prevent failure and replacing it with the orientation Nassim Taleb calls anti-fragility (Taleb 2012), designing systems that become stronger through disruptions that cannot be predicted.

What these projects share, Thresholds, Branding Islands, the labs, Failure, is a commitment to maintaining the capacity for ongoing learning, not delivering completed knowledge, but establishing the conditions under which knowledge continues to be produced. The following year, the labs geomorphology table documentation was minted incrementally on the Tezos blockchain throughout the Venice Architecture Biennale a distributed archive built on the premise that experiments conducted today may prove relevant to questions not yet formulated.

That reorientation, developed through project-based research across twenty years, is the argument this dissertation advances.

Why Some Methods Are Not Central

This constellation of methods requires me to be explicit about what the dissertation does not try to do. The primary unit of analysis is the project in its milieu, not a statistical population. The work does not pursue laboratory science in the narrow sense of controlled experiments on single variables. It does not present a purely theoretical or historical treatise developed in isolation from practice, concepts are held accountable to their usefulness in the practice of landscape architecture with design offices, engineers, scientists, agencies, and communities. It does not propose a universal toolkit of methods, it offers a repertoire of situated approaches that focus on design research, collaborative modeling, case-based inquiry, communities of practice, and autoethnographic reflection which others may adapt to their own ecologies of practice.

In short, the dissertation adopts methods that keep design, collaboration, and dynamic environments at the center, and intentionally sidelines methods that presuppose stable objects, fully controllable variables, or disembodied observation.

If our representations of landscape are always products of the instruments and frameworks we bring to them … if \"the landscape\" is already a shaped reading … then what kind of territory have we been operating on? What does landscape architecture inherit when it inherits the project of prediction and control in indeterminant systems?

Full project documentation is gathered in Appendix A.

These are the questions the methodology is designed to hold open. The chapters that follow do not resolve them, they demonstrate what it looks like to work inside them, through the communities, tools, and territories where the research was actually made.