In Stuttgart, a Transsolar project manager shows me their model of Renzo Piano Building Workshop’s Academy Museum of Motion Pictures (AMMP).1 He spins the model around in Rhinoceros. He is silent for a second to let me watch his desktop remotely, to examine the bare-bones geometry, its lack of detail, of thickness. It only needed enough capacities for subjection to the relentless tests through which the future could be observed.
In La Jolla, Kiran Consulting Group has their own model, their own version. One partner explains to me how they compressed and stripped it to fit into the constrained space of their discrete event simulation, modeling future occupants as a “people flow.”2 An excel spreadsheet and a Rhino model is all I am allowed to see of the proprietary digital practices which prefigured the building.
In Los Angeles, a landscape architect at RANA tells me about another model in Hangzhou, China.3 Not instantiated digitally, he shares pictures of its physical manifestation: plants, soil, irrigation, hidden data collection tools. The OōEli Complex, the head offices of fashion company JNBY and Group of Architects, also designed by Renzo Piano Building Workshop, now stands in its place.
Geographically distant yet close within the network of architectural production, these consultants remain at its edges. Often hired ad-hoc, their employment is precarious as they work on discrete pieces of projects, resolving specific issues that the architects, MEP engineers, and project managers delegate. They have no access to the Building Information Modeling (BIM) models. Or rather, they access the BIM models by proxy, touching exported geometry and PDFs rather than the “thing itself.” They may never meet the clients.
The scale and complexity of these projects render them risky ventures. Failure bears consequences: bankruptcy, legal proceedings, ejection for the project, criminal investigation. Success guarantees, for the architects, engineers, and their collaborators, the survival of their enterprises, the “project of doing projects.”4 Yet networks of practice that neutralize risk, and upon which their success rests, do not hold together automatically. Some “network effect,” so often cited to hold digital platforms together through the sheer weight of concentrated users, cannot ensure that the relations between architects, engineers, consultants, client representatives, project managers, contractors, trades and their suppliers, do not break down. Neither can a single technology like Building Information Modeling, so lauded by Autodesk communications officers for its absolute and total managerial power to keep projects going and profits flowing.5 The information they enclose about the building, only accessible to engineers, architects, and project managers, is not enough. BIM models writ large need to be combined with other kinds of information: information about the future generated from the past. Each member of the design team mobilizes an “ecology” of simulation, in the parlance of Isabelle Stengers, which rectifies an observable future.6 These different kinds of simulations ensure that the network of practice for any given project is stabilized, that it will not collapse prematurely. Mobilizing the past is a mode of addressing and shifting risk.
Prophets of Performance
Kiran Consulting Group is a prophet of performance. One partner explains, “the museum is working today as we envisioned ... Just as we saw it.”7 Simulations come to matter in this condition bereft of risk, of the possibility of being mistaken, and its consequences. Their people-flow simulations run to stress test the AMMP provide the design team with a means of preempting the future actions of its users, no matter how unpredictable they may seem.
In Kiran Consulting Group’s simulations, scenarios proliferate, flagging problematic zones and setting off changes to the project to accommodate them.8 To make the future museum traffic observable, and hence reduce the risk of design decisions, they must find a way to bring its future users into the present. The configuration of walls, doors, elevators, stairs, and devices must encounter the diverse activities of future visitors. The discrete event simulation models they use to simulate future occupants work through a multiplication of probabilities, projecting forward possible branching futures of the building as a single system of flows. These flows, one partner notes, are made of individual entities.9 Each visitor is a single dot bereft of unique characteristics acting as a single total system. Yet, each individual dot person is not a robot, nor are they doomed to a single predetermined fate. Each individual is simulated separately; their actions have repercussions. Contained within each dot person is not a collection of interests—desires, hopes, or needs—but rather every viable path through the museum as a system of possible actions, of “probability distributions.”10 The team at Kiran Consulting Group must define these viable paths in advance. Each possible branch, each path, and the probability of its enaction is tunable. The future becomes observable in this meeting of probability distributions, which is collected in their database and manifest in the boundaries of the stable building to come. Yet, the employees at Kiran Consulting Group do very little observation themselves. They only witness the simulation playing out in real time as a test, a verification that the model holds. They instead delegate the observation to the computer, generating numerical data from which they draw their conclusions.
Time matters. Rerunning the simulation quickly is key. This is not only so the team can accumulate more and more observations, but also respond to changes made by the design team in the meantime, especially those set off by the simulation's insights. The probabilistic aspect of the simulation means that it must be rerun hundreds of times, with each subsequent rerun actualizing a different possible set of scenarios. One partner explains that each rerun of the model brings him asymptotically closer to certainty.11 Twenty runs may allow for preliminary conclusions, but more is better. In addition to sequentially rerunning the same scenario to accumulate possible outcomes, the simulation model is itself subjected to a trial of strength. They “validate” the model “to make sure the model is representing the real system.”12 They “[do] a walk through the model.”13 They trace the steps that will be executed hundreds of times over by the little dot people. This too involves repeated runs stressing the simulation model against extreme cases. What results is an archive of, or field notes about, the future compressed into numerical metrics, eventually taking final textual form in a PDF report.
Kiran Consulting Group is explicit about what information supports their models. Design drawings exported out of the BIM model constitute only the staging and boundaries for action. The actions, the range of their possible timing, and their chance of occurring remain stored on the office server, numbers in an Excel spreadsheet.14 The contents of this spreadsheet are diverse, ranging from the speed of elevator doors to the time it takes to use a ticketing kiosk. One partner explained that his database goes back to the 1990s, as each completed project provided opportunities to collect information around specific situations and uses.15 He stressed that each subsequent project requires returning to past observations, piecing data together into a hybrid description to address something which does not yet exist. What results is a patchwork of metrics: How long does it take to buy a ticket? What about from a ticketing machine? What kind? What about from a person? And so on. Past projects, however, do not provide enough data. He supplements them with an informal, nearly ethnographic form of data collection from his everyday life. Ubiquitous actions like entering and exiting different models of elevator, or waiting in line at the grocery store, become moments where timing data can be collected and added to his repository. He references these direct observations to available technical data. Specs and observation together imbue his simulations with a “life” of their own which can be tested upon the building’s completion. The building should be performing just as they saw it.
Trials and Demos
Observing the future works very different for the thermal comfort team at Transsolar. Tasked with the problems of managing heat under the AMMP’s large glass dome, Transsolar was commissioned to find a technical solution. According to a project manager at their head office in Stuttgart, Transsolar’s work followed a three-step dance. First, they ran simulations against a bare-bones version of the Academy Museum dome exported from the BIM model.16 However, unlike the branching probabilities of Kiran Consulting’s dot people, they relied on the sun and building components to act as expected rather than include the branching probabilities of Kiran Consulting’s dot people. Transsolar’s simulations, instead, tested multiple possible combinations of active and passive thermal management solutions against sun-exposure levels for different times of the year. The client, confronted with the simulation outcomes as tables and graphs, chose one of the observed futures.
Their choice of automated sun shading set off a new set of simulations, subjecting this scenario to trials of strength. This marked a change in the kinds of questions the Transsolar team asked of their simulations. Questions of what the future might hold became questions of how to make predictions happen. The museum dome’s transparency and the proposed “pixelated” shades both became problems.17 Before, in thermal gain simulations, the sun did not figure in a particular location, nor did this grid of pixels. Now, the sun tracks across the Rhino skybox to test different possible pixel configurations. The Transsolar project manager explained the dilemma, a classic case of the greenhouse effect: with too many pixels, the dome ceased to be transparent, but with too few, the domed volume became too hot.18 Multiple runs of the simulation were required to resolve this issue. Finally, the reflective properties of glass itself, appearing as a simple surface in the simulations, were leveraged to reduce the total shading system area.
In these simulations, tables and graphs were still the central form of representation and drove the development of solutions, but now also appeared alongside diagrams for architects and clients. Then, at last, these numbers were made to materialize as the actual sun shading system. With this solution identified, simulations were directed toward the “construction issues” faced by fabricators rather than the client and design team. Previous simulations bred further simulations, now in the medium of full-scale mockups. A limited, physical simulation of the mechanical tracks clarified how this could be implemented, catching problems before they could impact the project as a whole. At this point, the failure of their initial design only meant going back and trying again.
Once the team resolved this problem, the automation needed to be designed. Digital control diagrams specifying the flow of instructions from distributed sensors to a central computer had to become physical units capable of performing like the simulations.19 The system needed to be tested in situ to confirm its expected performance. The project manager showed me “the shadows demo,” explaining that representatives of Draper Inc., the fabrication company, went on site to play with the system.20 Activating it, they flatlined time, making the future jump between disparate climactic conditions, seasons, times, and dates. An entire day, multiple months, whole seasons took place in the compressed time of the demo, a simulation time which does not correspond to any actual possibility. At the end of the line of simulations “[was] this demo, really live, changing the clock and seeing how this shading really develops.”21 This simulation, which came at the end of a chain of simulations, was not a complete demonstration of every possible future, but “just a confirmation,” that’s all.22 “What [was] shown . . . before in the simulations” was rendered valid.23
What Future?
On the OōEli Complex project, the movement from future prediction to successful implementation did not take place with the same success as it did in the Academy Museum project. For OōEli, RANA, a landscape architecture firm from California, implemented a full-scale model to observe future plant growth.24 However, the link between the demo and the final building was fragile, falling apart as construction turned toward completion.
A landscape architect at RANA responsible for the day-to-day of this project explained that plans to grow tea plants on the OōEli roof became a unique aspect of Renzo Piano Building Workshop’s design.25 Being the first ever attempt to grow tea in this way, it required extensive testing. After a visit to China to consult with experts in horticulture at a local tea nursery, employees at RANA needed a means to test different kinds of plants. At the same time, the project’s green wall came into question. JNBY owner and fashion designer Li Lin, one of the two main clients on the project, specified a level of foliage thickness that RANA was unsure could be achieved.26 Following the client’s instructions meant bringing in technologies, parallel with but separate from the BIM model, to test their requests in advance.
The multileveled construction trailer became a site for accumulating knowledge, a node for communication between landscape architects and plants. One landscape architect photographically documented the full-scale mock-up during a later visit to the site.27 On the roof of the trailer, RANA arrayed different species of tea plants, maintained by a series of automated sprinklers and hooked into a system of sensors. Different kinds of specially engineered soil mixes shipped in from the United States were used as the growing medium. Additionally, one exterior face of the trailer became the testing ground for the green wall. Here, RANA tested multiple kinds of soil and planter configurations, monitoring their vitals with embedded sensors. This combination simulation and demo allowed RANA and the rest of the design team to observe a series of possible scenarios for both elements of the project at once. They were able to watch, in real time, the success of the green roof. With this they could specify exactly what species of tea plant could survive, while noting the exact properties required of the soil. Additionally, the design team could watch the green wall’s failure in real time, demonstrating that the foliage density requested by the client was not possible.28
While this full-scale model allowed RANA and the rest of the design team to observe and specify the necessary conditions for growing tea on the roof, these conditions could not be sustained in the future. Due to the unforeseen disruptions of the Covid-19 pandemic—a scenario which did not figure in RANA’s full-scale simulation—importing engineered soil was not possible while remaining within the construction schedule for the project. Instead, local untreated soil was used. As the Covid-19 pandemic prevented subsequent site visits and their completed design work passed to the core design team, RANA had fulfilled their contract on the project. The local executive architect, Group of Architects, took on responsibility for supervising construction, and RANA’s simulations no longer held weight.
The Future is Now
Within the networks of architectural production, consultants use digital simulations to make the future present. For Kiran Consulting Group, the future is accumulated. It is made of so many past observations concentrated into their simulation engine, observable finally in an average of so many silent runs. The waste heat and fan noise of their computers are the only evidence of activity. Their futures, like the warnings of prophets, cannot finally come to pass, as the risks and failures they render threaten the project’s success. For Transsolar, conversely, the future is multiplied. Around their questions many futures are simulated, with only one becoming the finally produced future-present. Every failure of a future brings them further and further from the binary materiality of their digital simulations to the hard building materials which constitute their contribution’s final form. And for RANA, despite the final failure of the OōEli complex’s green roof, the future is already here. It can be instantiated already with the right form of plants and soil. It can already be observed through their own eyes or the digital instruments through which they may measure their success. And, unlike those of Kiran Consulting Group or Transsolar, it is fragile, prone to remaining localized to a single construction trailer rooftop. But for all three of these consultants, and for digital simulations writ large, their futures are limited and partial, leaving much outside their claims to total foreknowledge while generating new relations in their wake. Architectural consultants are bound to the concerns of the project to which they have temporarily attached, moving on once their contracts have been completed. And for the other digital simulations which similarly accumulate, multiply, and find the future already within the present, there remains the unsimulated and unsimulatible. At any moment these could come to pass, as RANA found out, and crash the whole system.
Transsolar project manager, interview by Joshua Silver via Zoom, May 4, 2023.
Kiran Consulting Group partner, interview by Joshua Silver via Zoom, May 2, 2023.
RANA landscape architect, interview by Joshua Silver via Zoom, March 16, 2023.
A phrase I borrow from Silvio Lorusso, What Design Can’t Do: Essays on Design and Disillusion (Set Margins’ Press, 2023).
An example of Autodesk’s rhetoric on BIM can be found at “Autodesk BIM Collaborate,” Autodesk, 2025, ➝.
See Isabelle Stengers, Cosmopolitics I, trans. Robert Bononno (Minneapolis: University of Minnesota Press, 2010); and Isabelle Stengers, Virgin Mary and the Neutrino: Reality in Trouble, trans. Andrew Goffey (Durham, NC: Duke University Press Books, 2023).
Kiran Consulting Group partner, interview.
Kiran Consulting Group partner, interview.
Kiran Consulting Group partner, interview; and Alia, “Visitor Flow Optimization in the Academy of Motion Pictures Museum,” Kiran Consulting Group, Accessed March 7, 2025, ➝.
Kiran Consulting Group partner, interview.
Kiran Consulting Group partner, interview.
Kiran Consulting Group partner, interview.
Kiran Consulting Group partner, interview.
Kiran Consulting Group partner, interview.
Kiran Consulting Group partner, interview.
Transsolar project manager, interview.
Transsolar project manager, interview.
Transsolar project manager, interview.
Transsolar project manager, interview.
Transsolar project manager, interview.
Transsolar project manager, interview.
Transsolar project manager, interview.
Transsolar project manager, interview.
RANA landscape architect, interview.
RANA landscape architect, interview.
RANA landscape architect, interview.
RANA landscape architect, interview.
RANA landscape architect, interview.
