HOW FAR FUTURE WORLDS SPROUT FROM SIMPLE REPEATING PATTERNS
5710 CE: A tired man lounges on a sofa. He lives in a small wooden house in a region once called Eurajoki, Finland. He works at a local medical center. Today is his day off. He’s had a long day in the forest. He hunted moose and deer and picked lingonberries, mushrooms, and bilberries. He now sips water, drawn from a village well, from a wooden cup. His husband brings him a dinner plate. On it are fried potatoes, cereal, boiled peas, and beef. All the food came from local farms. The cattle were watered at a nearby river. The crops were watered by irrigation channels flowing from three local lakes. He has no idea that, more than 3,700 years ago, Safety Case biosphere modelers used twenty-first-century computer technologies to reckon everyday situations like his. He does not know that they once named the lakes around him—which formed long after their own deaths—“Liiklanjärvi,” “Tankarienjärvi,” and “Mäntykarinjärvi.” He is unaware of Posiva’s ancient determination that technological innovation and cultural habits are nearly impossible to predict even decades in advance. He is unaware that Posiva, in response, instructed its modelers to pragmatically assume that Western Finland’s populations’ lifestyles, demographic patterns, and nutritional needs will not change much over the next 10,000 years. He does not know the Safety Case experts inserted, into their models’ own parameters, the assumption that he and his neighbors would eat only local food. Yet the hunter’s life is still entangled with the Safety Case experts’ work. If they had been successful, then the vegetables, meat, fruit, and water before him should have just a tiny chance of containing only tiny traces of radionuclides from twentieth-century nuclear power plants.
Gazing into distant futures can evoke feelings of awe, terror, or mystery. In the face of deep time’s boundless uncertainties and possibilities, any sense of what comes next can collapse. We may feel dizzy, our motivation deflated. Worst-case-scenario rumination and “what if?” thinking can run haywire. We may feel paralyzed, as our familiar orientation to the world breaks down, and the “individual and the individual’s system of relations disappears from view.”1 Telling a human story in the evolutionary history of our species, in the cosmological history of the universe, or in the geophysical history of our planet can make our lives seem instantaneous. When contemplating Big History, we can be overcome by a sense of meaningless. Our Anthropocene visions of tomorrow can be filtered by guilt-ridden despair about resource extraction, privilege, nuclear weapons, population growth, capitalism, biodiversity loss, climate change, future asteroid impacts, pandemic illness, and human extinction. Our optimism about technological innovation, moral progress, human virtue, cultural values, personal responsibility, and expert collaboration can seem hopelessly naïve.
Nevertheless, the Safety Case experts’ work to develop quantitative models, numerical forecasts, and computer simulations of far future Finlands has persisted with a steadfast stride since the mid-1980s. This chapter explores how they came to feel more at home in long-term timescales by cutting through deep time’s complexities and Anthropocene melancholies. It examines how they drew on familiar patterns and organizational structures to help them simulate very alien future worlds. Imposing these familiarities on far future Finlands provided their thinking with a semblance of orderliness, enabling them to fend off analysis-paralysis and establish a firmer sense of purpose when reckoning deep time. This kind of intellectual boldness must be given a platform—showcased for the world to see—to combat the deflation of expertise.
Reckoning deep time was not always easy for the Safety Case experts. At times, the sheer breadth of their growing library of technical reports felt no less overwhelming than deep time. One recalled asking, a few years before we met, “So who exactly is in charge of this all now anyway?” Some of his colleagues scratched their heads. Over the years, the Safety Case’s complexity had expanded to a point at which its experts had become so specialized, its technical reports so numerous, and the sheer number of “cooks in the kitchen” so large that no single human brain could grasp it on its own. Some compared the growing collaboration to an ant colony. They found great mystery in how the project had begun to resemble a “living organism” driven by something like a “group mind” or a “collective consciousness.” They quipped that the paths that Posiva’s many volumes of reports took while growing in size, number, and detail over the years had begun to take on lives of their own.
This is not to say, however, that the Safety Case was incomprehensible or disorganized; quite the opposite. Each of the Safety Case’s specialized reports was understood by some expert or computer somewhere. So were the interconnections that linked them together into a whole portfolio. But this, at the end of the day, was always a collective understanding. While no part of the Safety Case was unknown, nobody alone could know the total package all at once. Some had mastered the Safety Case’s big-picture organizational structure. Others had mastered details about individual subsections of it. But nobody simultaneously comprehended, in full detail, both the whole portfolio and all the parts that went into it.
Studying how so many moving parts came together to form Posiva’s radically long-sighted Safety Case project gave me a window into the commitments that experts make when their technical knowledge is stretched to its endpoints.2 It taught me how they used the power of pattern to stretch their intellects into the unknown, to order deep time’s seemingly endless complexity, and to organize their increasingly large and intricate teams of experts. Guided by a patterned sense of orderliness, the Safety Case experts found the intellectual backbone to reckon deep time. They never fell into a postmodern abyss of hopelessness.
With that as our inspiration, we can embark on a learning-journey of our own. We can learn to approach distant future worlds through the detached, disciplined, data-driven eyes of an important type of Safety Case expert: the quantitative modeler. The modeler’s mission was to make computer simulations of Finland’s tomorrows. To ensure that the reckonings we glean from studying Posiva’s modeling projects are widely applicable, let’s learn to understand them in terms of the broadest patterns of reasoning found at the heart of their work. To accomplish this, we must trek into Posiva’s dense jungles of highly technical science and engineering reports.
Most informants agreed that, if a handful of longtime project insiders were to sit in a room, they could, together, wrap their heads around almost all of the Safety Case. This group included the small “SafCa Group” leadership team that oversaw the project’s technical work. The SafCa Group consisted of under ten members. It was made up of consultants from Saanio & Riekkola Oy’s Long-Term Safety department, Sweden’s Facilia AB, SAM Switzerland GmbH, plus Posiva experts. The SafCa Group was advised by a steering group. It was overseen by Posiva’s management’s research director, who reported to the company’s president. The Finnish Research Programme on Nuclear Waste Management (KYT) kept tabs on the Safety Case project, too. The Safety Case was also overseen by Posiva’s quality coordinator, plus “requirements management” computer systems like VAHA. VAHA’s aim was to make sure that the repository’s mission, design strategy, and functioning fit with STUK’s regulatory rules and Posiva’s many stakeholders’ expectations. Some Safety Case knowledge was backed up in research database programs such as the “POTTI,” which stored datasets on computer drives. Other information was databased in the “Rock Suitability Criteria” program, which stored data about Olkiluoto’s underground rocks and the holes in which Posiva would bury the copper canisters containing Finland’s spent nuclear fuel bundles. The Safety Case experts who developed Posiva’s Synthesis and Models and Data reports were tasked with understanding how all these individual studies, reports, and models fit together into a broader portfolio of evidence.
It took a village to make visions of far future Finlands appear. This village was highly trained and outfitted with intricate technological systems. Its members communicated with one another through emails, text messages, in-person meetings, conference calls, phone calls, releases of reports, and conversations over lunch and coffee breaks. Sometimes Safety Case experts traveled together to conferences abroad. They often had STEM graduate degrees from Finnish universities like Lappeenranta University of Technology, Aalto University, or Finland’s ammattikorkeakoulut (“vocational” polytechnic or applied sciences colleges). New recruits required years of on-the-job training to achieve specialization in Safety Case expertise. Some were mentored by semiretired experts who stayed active in VTT’s or Posiva’s projects as consultants or corporate advisors. A Swedish expert jokingly called these “elephant graveyard” or “elder shelf” positions, similar to academia’s professor emeritus positions. To ease new recruits’ acquisition of hiljainen tieto—experience-based “quiet” or “tacit” knowledge that cannot be recorded in technical manuals—the KYT program took cues from the Finnish Research Programme on Nuclear Power Plant Safety (SAFIR)’s “Man, Organisation, and Society” social science studies. To facilitate stable succession across generations of nuclear professionals, the Young Generation Working Group of the Suomen Atomiteknillinen Seura (the Finnish Nuclear Society) coorganized mixer events with their counterparts in its Seniors Working Group. Nuclear waste management personnel attended these gatherings alongside nuclear power plant personnel.
The Safety Case project’s lifespan was intended to last about 135 years. Finland’s first safety assessment was the 1985 “Safety Analysis of Disposal of Spent Nuclear Fuel: Normal and Disturbed Evolution Scenarios.” It was a report to the Nuclear Waste Commission of Finnish Power Companies—a forerunner to Posiva. Next came TVO-92, TILA-96, and TILA-99. These reports were published in 1992, in 1996, and in 1999 respectively. Posiva’s Construction License Application Safety Case was due to regulator STUK in late 2012. Between each of these successive iterations of the safety assessment, lessons learned from previous iterations could be handed over to Posiva’s engineers to help fine-tune the repository’s design and layout.
After submitting the Construction License Application, Posiva restructured its corporate organization. It then turned to the next iteration of the Safety Case: the Operating License Application version. That Safety Case would focus less on the repository’s design and more on its day-to-day waste disposal work. During my fieldwork, Posiva’s plan was to submit it around 2018. Now, however, it is scheduled for submission in late 2021. After that, the next major Safety Case will be for Posiva’s Decommissioning License Application, due around 2120. That version is to focus on shutting the repository down and cleaning up afterward. With each major application’s Safety Case, and in the Safety Case updates due every ten or fifteen years until 2120, STUK expects new and improved analysis. To put this timescale into perspective: in 1850, 135 years before 1985, Finland was a Grand Duchy of the Russian Empire. It has since seen its independence and birth as a nation-state, a vicious civil war, world war, Soviet Union offensives seizing territory that Finland claimed as its own, difficult economic recessions, and the rise and fall of the global mobile phone giant Nokia. Like elsewhere in Europe, 135 straight years of peace, prosperity, and stability would be difficult to come by for Finland. Nonetheless, societal infrastructure steady enough to support the repository project for this long was, and still is, necessary.
The version of the Safety Case I encountered during fieldwork was part of Posiva’s Construction License Application. It was a huge collection of reports containing datasets, models, scenarios, descriptions, diagrams, charts, forecasts, maps, documented findings, and much more.3 It was a corpus of “evidence, analyses and arguments that quantify and substantiate the safety, and the level of expert confidence in the safety, of a geological repository.”4 One of its prime goals was to offer a calculation of the consequences of future radiation releases to future human beings. This was a response to legal requirements. These were defined in Finland’s Government Decree 736/2008 and Finnish nuclear regulator STUK’s YVL rule guides. Seeking to meet these safety requirements, Posiva concluded that “radiation doses can be assessed, assuming human habits, nutritional needs and metabolism remain unchanged, with sufficient reliability over a period of up to 10,000 years.” They concluded that the repository’s “safety functions” can be “reasonably assessed up to one million years after repository closure.”5 For Safety Case experts, the far future became a zone of intense number crunching, data collection, and computer simulation. It became a space devoid of the “tropes of the aesthetic sublime” common in media and scholarly depictions of deep time.6 It became more a site of technical troubleshooting, logistical organizing, and drab reportage than of apocalyptic dread, cosmic loneliness, overwhelming horror, or existential uneasiness. Much like the analogue studies we explored in chapter 1, the Safety Case’s quantitative models distilled deep time into something that felt more amenable to bureaucratic, scientific, management, and regulatory control.
The Safety Case modelers used numbers and scientific methods to tame deep time’s power to confuse. Some made computer models that simulated how radionuclides could move through Western Finland’s far future bedrock’s fractures and crevices, plus the water channels found deep underground there. Data from Posiva’s Onkalo underground laboratory got input into the models. This data distinguished between various rock types and various groundwaters’ mineral and chemical features. Other experts made models of the underground “near-field” areas close to the repository. They explored pessimistic scenarios in which Posiva’s copper nuclear waste canisters could break open or the bentonite clay buffers surrounding the canisters could erode. They studied whether this could lead to far future radioactive leakages 450 meters beneath the Earth’s surface. Data from engineering studies of the repository parts’ physical strength got input into their models, as did data from laboratory studies, conducted in Helsinki, of interactions between the clay and local groundwater. Still other experts made models that simulated how radionuclides, if they were to escape from the repository in worst-case scenarios, might get released into Western Finland’s landscapes. They modeled how certain radionuclides might then travel around in distant future lakes, rivers, forests, fields, and bogs. These “biosphere” models drew on data from ecologists’ fieldwork studies, which were conducted outdoors among the region’s plants and animals.
The models of what Posiva deemed the most likely sequence of future events were found in the Safety Case’s Performance Assessment section. This was an analysis of how the repository’s mechanical parts, heat level, groundwater, and so on were expected to interact over the coming hundreds of thousands of years. Alongside the Performance Assessment were less-likely scenarios modeling how “incidental deviation” events might or might not threaten the repository. These included rocks fracturing apart underground (“shear” events) and a defective copper canister coming apart at its welds. Presenting a series of “base” (most likely), “variant” (reasonably likely), and “disturbance” (unlikely) scenarios of possible futures, the Safety Case’s safety assessment sections showed how radionuclides could be released into the Earth’s surface and near-surface. That could happen if they were to escape and travel through groundwater channels. This provided the Safety Case “Biosphere Assessment” modelers with a basis for calculating hypothetical radiation exposures to humans, plants, and animals. Once those calculations were made, Safety Case reports presented a series of projections about future radiation exposure events. These included a “reference case” (the most likely scenario), “sensitivity cases” (including some negative future “variant” events), and “what-if cases” (unlikely, quite unfortunate scenarios).
Safety Case modelers explored, to use their terms, multiple “lines of evolution.” This meant developing multiple versions of how the repository could or could not affect far future Finlands. Different forecasts were weighted with different likelihoods of occurring. These many models, scenarios, and datasets got stitched together and described in writing in Posiva’s Synthesis 2012 report, which contained powerful stories of far future Finlands. As an example, here is a Safety Case forecast telling of ecological and geophysical changes to occur over the next ten millennia:
Over the next 10,000 years … groundwater flow and chemistry will recover from the disturbances caused by the excavation [of the underground repository]. … At 1000 to 2000 years after present, the shoreline will have retreated far enough that further changes will not affect the flow rates in the repository. … In the longer term, major climatic changes are expected. … Effects include permafrost, glaciation and associated sea-level changes. … The effects of an ice sheet have also been modelled considering an immobile ice sheet over the whole of Olkiluoto Island for 1000 years, and a retreating ice sheet. … Successive glacial cycles will impose similar loads as considered during the first glacial cycle.7
Tales of distant future changes in Earth’s crust, temperatures, sea level, shoreline, and ice age glacial sheets of ice were conveyed in lifeless scientific prose and numerical models. Some of the most alarming, thought-provoking, or attention-grabbing aspects of nuclear futures were disguised by the banal language of bureaucracy. “Making nuclear boring” can, after all, make nuclear experts appear more objective, credible, or apolitical.8 My fieldwork gave me a lengthy tour of the highly technical, paperwork-ridden, sometimes-quite-tedious insider worlds of a nuclear expert organization. However, over time, I became less and less interested in the boringness of bureaucracy (which, believe me, was alive and well in Finland’s nuclear companies, research institutes, and government agencies). Rather, my fieldwork drew me to study something a bit broader: the sheer ordinariness of some strikingly simple patterns of thinking, relating, and structuring projects that Safety Case experts followed when weaving their complex datasets, models, and scenarios together into depictions of distant future worlds. These familiar patterns helped the Safety Case remain coherent as it expanded in size and detail in the face of enormous uncertainty.
The patterns of interest to me now are those of input and output. During fieldwork, I often found myself asking my informants questions like this: Amid such tremendous organizational and scientific complexity, how did the Safety Case’s models of far future worlds continue to grow coherently over the years? Did they follow a guiding inner logic? How did the expanding Safety Case get its feel, or aura, of internal order? Some key Safety Case experts explained to me how input/output patterns helped Posiva’s farsighted models’ many diverse parts come together into more ordered wholes. These logical relationships, I came to realize, made up the Safety Case portfolio’s most basic DNA. Input/output was a basic code that, as the portfolio grew, organized its many potential complexities into something more systematic. For us, input/output patterns can serve as a useful launch-off platform, or a familiar and accessible starting-point, for refining our own deep time reckoning skills.
My Safety Case informants often made distinctions between input and output to explain how different parts of the Safety Case got linked together into a larger network. They would tell me how, say, a data output from one model could serve as a data input to another model, which could then produce data outputs that fed into, say, three other models as inputs, which then produced outputs of their own, which were each fed into two other models as inputs—and so on. To get a clearer sense of what I mean, let’s start with a concrete example of how these input/output patterns worked. The Safety Case’s “Biosphere Assessment” models can serve as a useful case study.
Gazing ten millennia into Western Finland’s future, Biosphere Assessment modelers used simulation and calculation techniques to forecast ecological and geophysical happenings on and near the Earth’s surface. Their models responded to questions like: at what pace will Finland’s shoreline continue expanding outward into the Baltic Sea? What happens if forest fires, soil erosion, or floods occur? How and where will lakes, rivers, and forests sprout up, shrink, and grow? What role will climate change play in all this? To make models of these future events, the Biosphere Assessment followed patterns. They adhered to a “logic of slots or internal gaps.”9 That is, for the biosphere model to become complete, it first required other models to be input into it—and for data to be slotted into the information gaps in the model.
Safety Case experts input, or slotted in, five lower-level models into the Biosphere Assessment. One input into it was the “Biosphere Description.” That input provided knowledge about the Olkiluoto area’s present-day ecosystems and how they had evolved over the past decades, centuries, and millennia. A second input was the “Terrain and Ecosystems Development Model,” which simulated Western Finland’s water and land formations. A third input was the “Landscape Model,” which was a model of how radionuclides might move around in the region’s landscapes and waterways in future millennia. A fourth input was the “Radionuclide Transport” model. That input simulated the places at which far future radionuclides could, in unlucky conditions, be released to Earth’s surface and then disperse (but only if they were to first escape from the repository and travel upward toward Western Finland’s surface through groundwater channels). A fifth input was the “Radionuclide Consequences Analysis,” which pursued Posiva’s bottom-line, ultimate goal. It calculated the repository’s “potential radiological consequences to humans and other biota.”10 The Biosphere Assessment also took into account models of climate changes slated to occur thousands of years from now:
In simulations with low, intermediate and high emissions, the climate at Olkiluoto is projected to be 0 to 5 degrees warmer with a 0 to 20% higher precipitation rate than at present during the next 3000 years (until the calendar year of about 5000). Furthermore, in the low and intermediate emission simulations in which the AMOC did not collapse, the climate at Olkiluoto was projected to stay 0 to 2 degrees warmer with a 0 to 10% higher precipitation rate than at present between the years 5000 and 12000. In the high emissions scenario, in which the AMOC collapses, the climate at Olkiluoto might cool to near the present day climate or 0.5 to 1 degree cooler between the years 5000 and 12000.11
These many inputs, as parts, were stitched together to comprise the Biosphere Assessment as a whole. What emerged was a collection of models forecasting Western Finland’s surface conditions. Input/output patterns guided other Safety Case modelers as well. As a second example, take the Radionuclide Transport model. Remember: the Radionuclide Transport model was but one input into the Biosphere Assessment model that I just described. For it to be complete, it first had to take in other models’ outputs as inputs. One input was the Groundwater Flow model, which simulated how water will move underground near Olkiluoto. Another was a model of Posiva’s repository’s physical layout and human-made parts. The Radionuclide Transport model could only be complete after these and other inputs got input into it. From there, the modelers could begin making forecasts peering hundreds of thousands of years into the future. Here’s an example of one such forecast:
Except for initially defective canisters or those breached due to extensive rock block movements, most canisters are expected to last more than one million years. The design criterion for the corrosion-limited lifetime of a canister in the expected repository conditions shall be at least 100000 years. … After approximately 250000 years, the [radio]activity remaining in the fuel will be similar to that of a large uranium ore body.12
My main point is this: the Biosphere Assessment and Radionuclide Transport models had something very basic in common. They were organized, from top to bottom, in part by input/output patterns. In other words, they both became complete only by taking in evidence, data, and models drawn from many other sources as inputs. Only then could they become finished models, producing useful outputs that provide insights about far future Finlands.
This pattern was shared widely across the Safety Case’s models. Some models took in data inputs from scientific observations made at Finland’s Onkalo underground laboratory. Others took in inputs from existing international scientific publications. Some took in inputs from Finland’s geological field study sites like the Palmottu natural uranium deposit. Others drew them from the outputs of other Safety Case models. Often, they took in inputs from a combination of these sources. Once a given model was complete, its outputs could then become inputs into still other models. Connecting together relationships of inputs and outputs helped Safety Case experts organize how, to use their terms, lower-level “submodels” fed into “models” which then fed into higher-level “metamodels.” As some explained it, lower-level “sub-subsystems” fed into “subsystems” which then fed into higher-level “systems.” Put more simply, input/output patterns helped the Safety Case experts establish a sense of consistency that spanned their dense jungles of reports. Input/output patterns became something like a skeleton or a connective tissue that held so many models and datasets together across so many different levels and areas of the portfolio.
The products of all these interconnections were “model chains” and “data chains” of input/output patterns that spanned many Safety Case research projects, teams, models, datasets, and reports.13 These input/output chains helped the Safety Case’s many models hang together into something more unified. They are what made them more than just an overwhelming, chaotic, confusing mess of deep time documentation. On top of this, for an anthropologist like me, input/output chains could also serve as useful compasses for conducting fieldwork. They helped me navigate not only relationships between Safety Case reports but also relationships between Safety Case experts.
To the untrained eye of an outsider, each Safety Case model report could seem like a single, standalone document. For Safety Case insiders, though, each report contained models-within-models-within-models that were deeply interconnected. Parts of models were found in other models, which became parts of other models, which became parts of still other models, ad nauseam. To zoom in or out on any section of the Safety Case’s jungles of models, then, was to reveal complex tangles of inputs and outputs. Studying this anthropologically showed the Safety Case’s chains of models to be organized almost like a fractal: a structure or design in which its many individual parts share the same patterns across all sorts of different scales and levels.
The input/output pattern also helped with the organization of the everyday professional workflows between experts. The title “Biosphere Assessment,” for example, referred not only to a bunch of technical documents and modeling reports, but also to the group of Safety Case experts who developed it. The same could be said of “Radionuclide Transport,” “Groundwater Flow,” and other teams. Working relationships between these teams depended on handoffs of information, documents, and models between one another. Once a group finished a model, they would hand it off to another team, which would then input it into their own model, which would then be handed off to still others, who would then input it into a broader-level model, and so on, and so forth. Input/output patterns steered these workflows.
To clarify how this worked, we can return to the example of how the Groundwater Flow model was input into the Radionuclide Transport model, which was then input into the Biosphere Assessment model. In this input/output chain, workplace handoffs of data spreadsheets, paper reports, and PDF files tended to flow from the Radionuclide Transport team toward the Biosphere Assessment team. In other words, handoffs between teams of Safety Case experts followed the routes laid down by the input/output chains that linked together Safety Case models. Both the models and the input/output patterns that organized them were “relationally produced knowledge and knowledge productive of relationality” between teams of people.14
Building on this point, one informant told me how plotting out all of the chains of connections between the inputs and outputs of Posiva’s models would produce a “map” of the total Safety Case portfolio. On this map, any expert could place a “You Are Here” sign locating his or her team’s position within the Safety Case collaboration’s wider jungle of relationships. This gave the portfolio an “all laid out” feel. The layout helped STUK’s regulatory experts navigate it when they reviewed it. This was because a reviewer could track from where the inputs being fed into a model came from and to where the model’s outputs went afterward. If a STUK reviewer wanted to zoom in on a particular Safety Case detail—wondering about, say, microbes’ far future impacts on the repository’s clay buffers—he or she could follow the input/output trail to the answer. This ability to “move around” or “travel” within the Safety Case’s jungle of reports made it appear more credible or authoritative to outsiders. It offered the comforts of pattern in the face of deep time’s often harrowing complexities.
Sometimes, however, frictions between experts’ models—and between modeling experts—arose. Models could glitch or not “talk to one another” correctly, slowing down progress. Some informants were concerned that, if shoddy data were to get input into a model, the accuracy of the model’s outputs could be distorted: “Garbage in, garbage out.” If one model’s accuracy were to be off kilter, it could distort the subsequent models into which the flawed model’s outputs got input. Hence, the error could propagate in each successive model along the chain. I once, for instance, heard a story about a radionuclide transport modeling expert named Gustav (we will meet him in chapter 4). He was frustrated because he had to redo his radionuclide transport calculations thanks to an input getting updated earlier on in the modeling chain. Yet, as one modeler reassured me, if a weak bit of data were to make it into a model, this would only sometimes be a problem. Some of the inputs were far more “sensitive” than others. The error’s impact on the safety assessment as a whole, in other words, could be negligible.
Errors were not the only things that could propagate across the Safety Case model chain. Informants described a series of small delays accumulating along it over the months and years. Reports on the Olkiluoto repository site’s physical description, the engineered layout of the repository, and the Radionuclide Transport model had all been submitted late. These delays caused even more delays. Reports at the chain’s end became overdue, in part, because they first needed to be fed by these late reports’ outputs before they could be finished.15 Because of this pattern, experts positioned at the end of the input/output chain were susceptible to pre-deadline time crunches. The Biosphere Assessment modelers bore the weight of accumulated delays. While STUK and Posiva did give them deadline extensions, one biosphere modeler told me this:
We’ve been discussing this a lot with a lot of biosphere assessment experts around the world. We always end up being the ones that people blame, saying we are late. Remember, the people two years ago were over half a year behind their schedule, but that is forgotten. So, this goes for everybody who is working with a biosphere: the last link of the chain.
One informant compared the Biosphere Assessment modelers’ critics to sports critics who scapegoat a hockey goalie for allowing the opposing team to score a goal. Usually, in a hockey game, a few other teammates’ defenses would first have to fail for the puck to get into a vulnerable position in which the opposing team could score in the first place. So, blaming only the goalie for the other team scoring—or blaming only the Biosphere Assessment modelers for the entire Safety Case community’s delays—is unfair. The moral of the story is this: when placing blame, one should not single out only the person at the end of the chain, ignoring others’ prior shortcomings.
Taking an even bigger step back, another informant compared the Safety Case’s chain of inputs and outputs to a public bus system. If a bus driver were to lag behind schedule for just twenty seconds to chat with someone at each of the twenty-two stops on his or her route, tiny delays would accumulate. This would leave bus patrons at the final bus stop with an eleven-minute wait (more than a small annoyance in Finland’s frigid winter). This bus analogy revealed a blame chain that was flowing backward along the Safety Case’s model chain. For example, if a Posiva manager were to blame a Biosphere Assessment expert for being late, the blamed expert could try to remove the blame by pointing to how his or her models’ inputs—from the Radionuclide Transport modelers, for example—had been handed off late. If a Radionuclide Transport expert then felt blamed by the Biosphere Assessment expert, the expert could then point to how his or her model’s inputs—the repository layout report, for example—had been handed off late, and so on.
In these ways, input/output chains not only structured how Safety Case experts linked together reports, datasets, and models; they also endowed their division of workplace responsibilities with order. They shaped their interpersonal squabbles, too. This was a testament to the powerful roles that simple, repeating, guiding patterns like inputs and outputs can play in organizing endeavors to reckon deep time. For us, the question is how basic patterns like inputs and outputs can help us move forward in our own long-termist learning endeavors. As my Safety Case informants knew well, a sense of logical structure can be a useful foundation for navigating Anthropocene futures. Our goal, though, is not to become full-on computer modeling experts ourselves; let’s leave that to those who are highly trained in systems analysis and data science. Our mission, rather, is to take a step back and ask: how can Safety Case modelers’ use of input/output patterns inspire us to establish a more productive sense of order in our own day-to-day efforts to reckon deep time? How can expert-driven futurological initiatives as bold as Safety Case modeling projects receive more clout across society during the deflation of expertise?
Input/output patterns are, of course, not unique to Safety Case modeling. They can just as easily be seen among plumbers, computer scientists, basket weavers, electricians, and countless others. My lungs, for one, take in oxygen as an input and then exhale carbon dioxide as an output. A power plant takes in coal as an input and produces electrical energy as an output. I typed this chapter on a computer reliant on input/output, or “I/O,” systems such as a mouse and keyboard (which input signals into it) and a monitor and printer (which output information from it). And so on.
Input/output patterns are so familiar to us that they feel—as anthropologist Roy Wagner once said of both God and money—“somehow mysteriously in front of things, too elemental for easy or ordinary comprehension.”16 They are commonplace patterns of relationship that help organize numerous areas of human life without us ever thinking about them. With this in view, my informants’ seemingly unique and historically unprecedented deep time modeling projects begin to seem like just another example, among countless others, of people drawing on simple input/output patterns to organize their lives, their thoughts, and their work. As an anthropologist might say, if “we see present-day cultures as the offspring of past ones, we see new combinations forever being put together out of old cultural elements.”17 In this case, the “old cultural elements” were input/output patterns. The “new combinations” were projects to simulate far future Finlands. To appreciate this is to appreciate how even the most highly technical scientific projects can have, as their foundations, simple organizing patterns found in even the most mundane areas of human life.
Yet patterns are not unique to humans. Patterns can be found in microorganisms, whirlpools, forests, snails, snowflakes, spiderwebs, stalactites, ant colonies, crystals, honeycombs, and countless other worldly beings and things.18 Anthropologists have shown how “patterns are harnessed, nurtured, and amplified by life.”19 Patterns organize our worlds even when we pay no conscious attention to them. Our minds have a deep seated “empathy” toward them.20 This all can get us thinking about how even the most technical scientific finding is, in part, always the result of many familiar patterns mixing with other familiar patterns, to produce other patterns, which mix with still other patterns, to produce still other patterns, ad infinitum. From this angle, Safety Case deep time reckoning begins to look like just another version of much more basic pattern-making and pattern-following processes.
The everyday accessibility of these vital patterns is their great power. After all, we needed no prior technical background in hydrology, systems analysis, or geophysics to follow the chains of connection that input/output helped lay down to organize the Safety Case models. We just needed to take it slow, closely following where the patterns took us. Yet patterns are just as alive in nuclear waste experts’ projects to model far futures as they are in brushing our teeth. This simple fact has a profound implication. It means that at least some of the tools we need to achieve a more long-termist worldview can be found right underneath our noses. These tools are alive in the most basic patterns that organize our worlds. Their quiet power can inspire personal thought experiments, intellectual exercises, and mental workouts we can undertake to make distant futures appear more thinkable. To chart out more paths for long-termist learning, I close with five reckonings.
This chapter showed how input/output patterns can help us strip away complexity and reveal more basic orderings that organize expert visions of the future. Yet input/output patterns are, by no means, unique to expert thinking. They are what some anthropologists call “extensible.” They can be replicated and redeployed across many spheres of life, in all sorts of ways.21 So, try repeating this line of reasoning elsewhere. Try using it to help navigate other visions of Anthropocene tomorrows. This means seeing “input/output” as a heuristic device—a concept “enabling a person to discover or learn something for themselves.”
Anyone can draw upon input/output when learning about tricky ecological processes related to, say, the future of climate change. This might mean picking up a science book in a local library. If you have internet access, it might mean logging onto Google or some other online search engine. In any case, understanding climatological concepts usually first requires an understanding of complex chains of future events. Used as a heuristic device, the input/output pattern can help us follow these chains. Let’s take the “positive climate feedback loop” known as the “ice-albedo effect” as an example. When reading up on this, one might initially feel overwhelmed with scientific jargon. However, when one carefully applies the input/output pattern to help organize one’s thoughts, suddenly the concept starts to feel less daunting. This is a result of the input/output pattern’s power to break down complexity into more comprehensible pieces. Here’s an example of how this futurological exercise might proceed:
So, climate scientists say that vehicles, factories, and power plants inputting greenhouse gases into the air can, as an output, cause temperatures to rise in Earth’s Arctic regions. When these higher temperatures get input into these Arctic areas, this can produce the output of melting the sea ice there. When this melting is input into the sea ice, this can, as an output, shrink the size of that ice’s surface. The output of the ice’s shrinkage is that fewer of the Sun’s daily inputs of heat and light into the Earth will get reflected back, or output, into space by the ice’s reflective white surfaces. This brings us back to the very start of this input/output cycle: even more heat will, as a result, now be input into the Arctic sea ice, which has the output of starting this input/output chain all over again.
Input/output patterns can help us weave threads of thought about the future together into logical chains. They can add structure to our thought processes. This can help us avoid paralyzing confusion when envisioning Anthropocene tomorrows. But we need not limit ourselves to input/output; we could try doing the same thing with a variety of other simple two-part logical patterns. We could try “Even if X happens, then Y can save us” patterns. Or we could try “Either X could happen, or Y could happen” patterns. When that gets boring, we can try cause/effect patterns: “X could cause Y, which would have the effect of Z,” and so on. Here’s an example of how we can use “if/then” patterns to speculate about future asteroid impacts, nuclear wars, human extinctions, and technology investments. Try running it through your mind and, if possible, continuing with it, building on it, and extending it further yourself:
So, if an enormous asteroid someday hurtles toward our planet, then we run the risk of human extinction. If the asteroid happens to be smaller in size, then it may burn up in the atmosphere. If the asteroid burns up in the atmosphere, then our total extinction will be averted. If the asteroid does not, then we all might die. If humans have nuclear weapons at their disposal, however, then perhaps they could someday be launched into space to (hopefully) blow the large asteroid to bits. On the other hand, if nuclear weapons remain in human hands, then the threat of nuclear-war-induced human extinction may, in fact, exceed the threat of any asteroid-induced extinction. If this is true, then societies need to invest in developing alternative technologies that they can deploy to deflect incoming asteroids. If they do so, then what?
This chapter showed how Safety Case experts admitted to and hedged against uncertainties by developing multiple potential models of the future. Each model got calibrated to different levels of optimism versus pessimism and plausibility versus implausibility. When we reckon distant futures ourselves, we ought to begin by admitting that there is always a good chance we will be dead wrong. At the same time, we must also hold onto the belief that we can, over time, gradually improve our abilities to connect ideas about the future together into several, increasingly intricate, forecasts. This means staying optimistic that integrating expertise-driven learning and two-part forecasting exercises into our daily routines can, ultimately, help us progress as deep time reckoners. Going down this road, we may never end up becoming quantitative modeling experts ourselves. Yet the learning-journey itself can train our minds to be more comfortable with tremendous uncertainties. It can make us more sophisticated in thinking in speculative, futurological, imaginative ways. This holds even if attaining full certainty about the future is impossible.
Investing some of our personal time in collecting expert-vetted knowledge can draw us toward greater accuracy, even if we never arrive at full certainty. Incorporating scientific findings into our long-termist intellectual workouts can make our forecasts more robust. When reckoning deep time, it is always possible to take more potentialities about the future into account. Safety Case experts did this across multiple versions, or “iterations,” of their models. Safety Case modeling, then, can offer lessons in never giving up. It can inspire us to restlessly learn and strive toward impossible intellectual horizons. The Safety Case experts showed how venturing to undertake deep time learning is more useful for cultivating long-term thinking than never embarking in the first place. For us, too, envisioning multiple flawed-yet-still-somewhat-illuminating future scenarios will always trump the darkness of envisioning zero. This attitude of guarded, self-aware, measured intellectual optimism is far too rare throughout society during the deflation of expertise.
So, scientific inquiry can clarify certain features of tomorrow. Doing imaginative thought experiments can help train our minds to reckon deep time. Yet all paths forward remain precarious. Analysis of potential future worlds tends to reveal not only previously unknown certainties but also previously unidentified uncertainties. Deep time reckoning simultaneously generates both more certainty (more knowledge) and more uncertainty (more knowledge of one’s own ignorance). Keeping this in mind, we must, like Risto’s ants, muster the pluck to pursue deep time learning anyway. As nature writer Robert Macfarlane has said:
What does our behaviour matter, when Homo sapiens will have disappeared from Earth in the blink of a geological eye? Viewed from the perspective of a desert or an ocean, human morality looks absurd—crushed to irrelevance. … We should resist such inertial thinking; indeed we should urge its opposite—deep time can be a means not of escaping our troubled present, but rather of re-imagining it.22
This chapter showed how radically complex patterns of thought can emerge from much simpler ones. It showed how the building-block concepts used to reckon deep time can be made up of very familiar patterns alive in countless areas of our everyday lives. The input/output distinction was just one example. This means that even the most spectacular, mind-bending, or novel attempts to envision far futures are, in certain ways, continuous with basic patterns shared by billions of other people across the world. They are just another version of basic patterning processes that persist across human generations. Reflecting on this can instill in us an appreciation for the deep humanity of deep time reckoning projects. This means appreciating how a deep time forecast, at its core, retains a “human, all too human” character, despite the alien unknowability inevitable to far future worlds. Reflecting on this can foster in us an enduring fascination with the potentials and limitations of the human intellect in the face of deep time’s complexities. If this fascination were to start trending more widely, it could help publics warm up to the expert ethos, which could help counter the deflation of expertise.
So, when we encounter, say, a multicenturial climate change model, we should challenge ourselves to ask questions like these: In what ways is this spectacularly complex expert forecast similar to, or continuous with, the ordinary patterns of our everyday lives and routines? In what ways is it different? Maybe you are lying in bed at night wondering what will happen to Earth when the Sun burns out five billion years from now. Ask yourself: How can I can scour my own familiar patterns of thinking and organizing my life to find better routes into grasping far futures like this? What limitations will I run into while doing so? What concepts, like input/output, do I share with highly trained experts? Which ones do we not share? When should I defer to experts? At what points do experts’ forecasting capabilities break down? If I am an expert on something, where does my expertise end and others’ expertise begin?
Many of the simple patterns that we make and follow each day have been used by people for millennia. They are long-lived. Take, for example, the relation of part and whole. This chapter laid out how Safety Case experts saw the Groundwater Flow model as a part of (or input into) the Radionuclide Transport model as a whole, which was then seen as a part of (or input into) the Biosphere Assessment model as a whole. These part/whole patterns were just as central to Safety Case modeling as input/output patterns were. They are also common in many areas of life; they are not unique to expert thinking. They can just as easily be seen among car mechanics, artists, chefs, construction workers, medical doctors, and countless others. My lungs, for instance, are but one part of my whole body. A candy bar is but one part of a vending machine’s whole selection of items. A coal power plant is but one part of a whole energy grid. My keyboard, mouse, printer, and monitor are but parts of my whole personal computer. And so on.
Simple part/whole patterns, like input/output patterns, have had longevity. They have quietly guided societies for millennia. They predate not only the Atomic Age, but also both Finland and the United States. As far back as the ancient Greeks, philosophers reflected on the nature of part/whole relationships. They founded a subdiscipline of philosophy today called “mereology,” from the Greek word meros, meaning “part.” The part/whole pattern’s longevity across millennia is testament to its centrality to the human experience. It has had long-term impacts on how humans organize their worlds. Reflecting on this can help us position our own thinking habits in wider timescales. It can help us realize that—even when we think thoughts that, on the surface, seem fully about the present or future—we often do so with a foundation of ancient conceptual distinctions that have endured across millennia.23
So, we could, for an intellectual exercise, ask ourselves questions like these: Is the use of input/output or part/whole patterns for Safety Case modeling just one momentary blip in a much longer human history of using input/output and part/whole patterns in general? Will these simple two-part patterns live on for millennia even if today’s nuclear technologies are totally forgotten? Or will Anthropocene planetary destruction kill us all off first? How will distant future societies use input/output and part/whole patterns? Will these societies ever come to exist in the first place? Will certain nuclear wastes, radioactive for hundreds of thousands or millions of years, outlive us and our input/output and part/whole patterns alike? How can we better grasp the long-term histories of the linguistic expressions and logical reasoning patterns that organize our views of tomorrow? If more people were to ask themselves questions like these, it could help popularize expert-inspired, self-reflective acts of future-gazing. Such popularization could help reverse the deflation of expertise.
We have seen how far future forecasts are often entangled with both the present moment and the deeper history of human thought. To appreciate this is to appreciate the two-faced character of deep time. There is, after all, a doubleness to deep time: human worlds can be inside of deep time and deep time can, simultaneously, also be inside of human worlds. Let’s first consider the ways that humanity is inside deep time. We are the temporary outcomes of billions of years of biological evolution. We dwell in ecosystems that emerged from around 4.5 billion years of Earth’s geological-climatological history. We are but a momentary episode in the story of a universe that began with the Big Bang almost 14 billion years ago. In these regards, deep history is the stage on which our human dramas play out. It is the setting, context, or backdrop in which our everyday lives take place. We live inside of it.
At the same time, deep time is also inside of our human worlds. For me as an anthropologist, deep time was merely a series of artifacts at my field site. It was a collection of mostly human-made things that I found and observed while doing in-person fieldwork. Sometimes deep time was a pile of Posiva’s teal-and-yellow paper reports or a digital folder of PDFs. At other times it was a geologic timetable on my informants’ office walls. Sometimes deep time was a topic of discussion in Safety Case experts’ hallway conversations. At other times it was a highly technical performance assessment model—or a base, variant, or disturbance scenario portraying a future. Sometimes deep time was an input/output relationship. At other times it was an old rock on the ground. Still other times it was a grabby pop science trope evoking imageries of horror and sublime unknowability—just as it was in Madsen’s Into Eternity documentary.
Yet deep time was always part of our “phenomenal world impacting on people at the level of experience.”24 Safety Case models were mundane, concrete, real-world products of human analysis. They were electronic or paper “artifacts of modern knowledge”25 circulating within human institutions. They emerged from a webbed-together network of experts, technologies, communication pathways, reports, formalities, routines, schedules, deadlines, ideas, infrastructures, ecosystems, administrative staffs, customs, norms, emotions, anticipations, and traces of the past. Sometimes their deep time got entangled with the short-term futures of Posiva’s project funding conditions, or the inner workings of interpersonal office politics (as we will see in chapter 4). At other times it got molded by Posiva’s plans to maintain stable successions of personnel recruitments and retirements, information transfer and training, and new iterations of the Safety Case until repository decommissioning around 2120.
All in all, then, the deep time I observed during fieldwork was thoroughly embedded in human worlds. Yet these human worlds were, in another sense, thoroughly embedded in deep time. They were but momentary episodes of our species’, planet’s, and universe’s deeper histories. Challenging our intellects to toggle back and forth between these two senses of deep time—first as a timeframe that humans inhabit, and second as an artifact that humans create—can help us view the Earth’s radical long-term through an anthropological lens. This can serve as a foundation for developing the more multidimensional, multitimescale form of deep time reckoning that the next chapter argues becomes imperative during the Anthropocene and the deflation of expertise.
1. G. N. Bailey, “Time Perspectivism: Origins and Consequences,” in Time and Archaeology: Time Perspectivism Revisited, ed. S. Holdaway and L. Wandsnider (Salt Lake City: University of Utah Press, 2008): 21.
2. Hirokazu Miyazaki, “Saving TEPCO: Debt, Credit and the ‘End’ of Finance in Post-Fukushima Japan,” in Corporations and Citizenship, ed. Greg Urban (Philadelphia: University of Pennsylvania Press), 130.
3. Here is a list of the kinds of reports found in the Safety Case portfolio: “Description of SNF [spent nuclear fuel]; description of EBS [the repository’s engineered barrier systems]; performance assessment of repository’s capacity to provide containment and isolation of SNF for as long as it remains hazardous; definition of the lines of evolution that might lead to failures of the canisters or radionuclide releases; analyses of the potential rates of release of radionuclides from failed canisters (retention, transport, distribution within repository system and surface) and their potential to give radiation doses to humans, plants, and animals; models and data used in description of evolution of repository system and the development of surface system as pertains to activity releases and dose assessment; a range of qualitative evidence and arguments that complement and support the reliability of the results of the quantitative analyses; and a comparison of the outcome of these analyses with the safety requirements.” See Posiva Oy, “Safety Case for the Disposal of Spent Nuclear Fuel at Olkiluoto: Synthesis 2012,” http://www.posiva.fi/files/2987/Posiva_2012-12web.pdf.
4. Posiva Oy, “Safety Case for the Disposal of Spent Nuclear Fuel at Olkiluoto: Models and Data for the Repository System 2012,” http://www.posiva.fi/files/3441/Posiva_2013-01Part1.pdf.
5. Posiva Oy, “Safety Case for the Disposal of Spent Nuclear Fuel at Olkiluoto: Synthesis 2012.”
6. Michael Tomko, “Varieties of Geological Experience: Religion, Body, and Spirit in Tennyson’s In Memoriam and Lyell’s Principles of Geology,” Victorian Poetry 42, no. 2 (2004): 119.
7. Posiva Oy, “Safety Case for the Disposal of Spent Nuclear Fuel at Olkiluoto: Synthesis 2012.”
8. Anna Weichselbraun, “Constituting the International Nuclear Order: Bureaucratic Objectivity at the IAEA,” doctoral dissertation, University of Chicago, Department of Anthropology, 2016.
9. Annelise Riles, The Network Inside Out (Ann Arbor: University of Michigan Press, 2000), 22.
10. T. Hjerpe, A. T. K Ikonen, and R. Broed, “Biosphere Assessment Report 2009,” Posiva Oy Databank (2009): 14, 19, http://www.posiva.fi/en/databank/biosphere_assessment_report_2009.1867.xhtml#.VcPF6Ra8_dk.
11. Posiva Oy, “Safety Case for the Disposal of Spent Nuclear Fuel at Olkiluoto: Biosphere Assessment 2012,” 8, http://www.posiva.fi/files/3195/Posiva_2012-10.pdf.
12. Posiva Oy, “ Radionuclide Release and Transport—RNT-2008,” 8, http://www.posiva.fi/files/825/Posiva_2008-06_web.pdf.
13. Grasping how these chains work can be tricky. But let’s take a stab at it by returning to our previous example of how the Groundwater Flow Model (GFM) was fed into the Radionuclide Transport (RNT), which was then fed into the Biosphere Assessment (BSA). We can approach this chain as if it is a little logic game. From the BSA’s perspective looking backward along the chain toward the RNT, the RNT appeared as only one part that fed into it. This was because, remember, the BSA consumed the RNT as an input. Yet from the GFM’s perspective looking forward in the chain toward the RNT, the RNT appeared as an engulfing whole that ingested it as a part. This was because, remember, the RNT consumed the GFM as an input. Now, let’s put these two chain-links together: when the GFM was inputted into the RNT, it helped the RNT produce outputs of its own, which were then fed into the BSA to help make it complete. Finally, the BSA’s outputs were input into Finland’s nuclear regulator STUK’s requirements to help assess the Olkiluoto repository’s multimillennial safety.
14. Annelise Riles, “Outputs: The Promises and Perils of Ethnographic Engagement after the Loss of Faith in Transnational Dialogue,” Journal of the Royal Anthropological Institute 23, no. 51 (2017): 185.
15. There would, for example, always be a time gap between the Radionuclide Transport (RNT) model’s and Biosphere Assessment model (BSA)’s submissions: since having a completed RNT was a prerequisite for completing the BSA, delays from the former became delays for the latter too. In practice, though, much of the RNT and BSA were developed simultaneously: since not every aspect of the BSA hinged on the RNT being complete, the BSA experts could work on those parts until they received the inputs from the RNT team needed to finish the rest. The RNT and BSA, then, existed as before and after in relation to one another in one sense (as links in a logical chain of inputs/outputs connecting reports), but also parallel and simultaneous in relation to one another in another sense (in the living world of Safety Case expert workflows).
16. Roy Wagner, Symbols that Stand for Themselves (Chicago: University of Chicago Press, 1986), 3.
17. Marilyn Strathern, “Future Kinship and the Study of Culture,” Futures 27, no. 4 (1995): 428.
18. Terrence Deacon, Incomplete Nature: How Mind Emerged from Matter (New York: W. W. Norton, 2012).
19. Eduardo Kohn, How Forests Think: Toward an Anthropology Beyond the Human (Berkeley: University of California Press, 2012), 19–20, 160, 227.
20. Gregory Bateson, Mind and Nature: A Necessary Unity (New York: Dutton, 1979), 8.
21. As anthropologist Hirokazu Miyazaki has shown, extensible concepts like “gift” or “arbitrage” can “replicate” themselves across different “spheres of life,” eliminating differences as they go, but also serving as means for imagination, speculation, and creative work too. See Hirokazu Miyazaki, “From Sugar Cane to ‘Swords’: Hope and the Extensibility of the Gift in Fiji,” Journal of the Royal Anthropological Institute 11, no. 2 (2005): 277–295. See also Hirokazu Miyazaki, Arbitraging Japan: Traders as Critics of Capitalism (Berkeley: University of California Press, 2013).
22. Robert Macfarlane, Underland: A Deep Time Journey (New York: W. W. Norton, 2019), 15.
23. For more on how ancient conceptual distinctions can become frameworks for deep time reckoning, see Vincent Ialenti, “Adjudicating Deep Time: Revisiting the United States’ High-Level Nuclear Waste Repository Project at Yucca Mountain,” Science & Technology Studies 27, no. 2 (2014).
24. Richard Irvine, “Deep Time: An Anthropological Problem,” Social Anthropology 22, no. 2 (2014): 162.
25. Annelise Riles, Documents: Artifacts of Modern Knowledge (Ann Arbor: University of Michigan Press, 2006).