Abstract
"What makes a good theory?" is a question philosophers of science usually ask about theories that hold still, evaluated by the people who built them. But a handful of theories, Freud's, Darwin's, and famously nuclear physics, don't hold still. They leave the laboratory, get seized by governments, misread by the public, and end up reorganizing law, politics, and how people understand themselves. This essay asks a different question: not what makes a theory good, but what happens to a theory as it travels, and why some theories travel much farther than others. Using Robert Oppenheimer's story as its central case, the essay tracks nuclear fission across four distinct stages: a specialist curiosity discussed only among physicists, a secret state priority known to a handful of officials years before the public had any idea it existed, a personal reckoning for the man who built the weapon, and finally a Cold War political fact judged by senators rather than scientists. It argues that a theory's reach into public life isn't random. It depends on what kind of thing the theory is actually about, and that reach, once achieved, changes not only who gets to evaluate a theory but what kind of harm it becomes capable of doing. The essay engages two existing bodies of work that study related territory, a recent philosophy of science framework, and four decades of research in science and technology studies, crediting both in detail while arguing that neither quite explains why most scientific ideas stay confined to journals while a rare few end up remaking the world.
I. A Question Behind Guest's Question
Olivia Guest, a philosopher of science, asks a deceptively simple question: what makes a good theory, and how do we make a theory good? Her answer takes the form of an ontology (metaphysical commitment, discursive survival, empirical interface, minimising harm) offered to working scientists as a "metatheoretical calculus" they can build for themselves, article by article, field by field. It is a generous, formally serious, ethically alert piece of work. It is also, I want to argue, a snapshot taken at a single moment in a theory's life, mistaken for a portrait of the whole life.
The question I want to ask instead is not what makes a theory good, but what happens to a theory between the moment it is first proposed and the moment, if it ever arrives, when it has become part of how ordinary people organise their relationships, their institutions, their sense of themselves. Some theories never make that journey. They live and die inside a single subfield, evaluated more or less exactly as Guest describes, by colleagues who read the same journals and attend the same conferences. Other theories, a small, historically decisive minority, leave the seminar room entirely and become woven into the fabric of everyday coordination: into parenting advice, legal doctrine, foreign policy, the stories people tell about who they are. Freud, Darwin, Marx, evolution, relativity by way of the bomb, and, the case this essay will dwell on at length, nuclear fission by way of Robert Oppenheimer.
I will call this journey a theory's recursive career, and the central claim of this essay is that a theory's recursive career is not an incidental sociological fact layered on top of its scientific content. It is a distinct object of inquiry, governed by its own logic, requiring its own vocabulary, and largely invisible to a metatheoretical calculus, however sophisticated, that treats "theory" as a stable unit of publication evaluated by a scientific field. Guest's ontology, I will suggest, is exactly right for the portion of a theory's life it was built to describe. It becomes progressively less adequate, not because its criteria are wrong, but because the kind of object being evaluated has changed, as the theory travels further from its origin. Nolan's Oppenheimer, both the historical episode it dramatises and the film's own narrative architecture, gives us an unusually clean, richly documented case in which to watch that transformation happen in real time, over roughly six years, to a single body of physics and a single physicist. A second body of literature, science and technology studies, has been occupied with a closely related problem for four decades, and this essay engages it directly rather than in passing, because the differences that survive that engagement are the ones worth keeping.
II. Guest's Ontology, Read on Its Own Terms
Before departing from Guest's framework it is worth crediting what it does well, because the departure I want to propose is a relocation, not a refutation.
Guest's ontology organises theoretical virtue and vice into four broad categories. Metaphysical commitment asks what a theory assumes without investigating, what it asserts, and how it is individuated from neighbouring theories and embedded within a field, formalised through a set-theoretic apparatus in which θ denotes an individual theory (typically a unit of publication), Θ a family of related theories, and 𝕋 a field's theoretical positions as a whole, with genealogical descent between successive versions of a theory (θ at t₀ becoming θ at t₁, and potentially fracturing into rival descendants) modelled as a directed graph. A theory, in this account, is always a theory of something, a phenomenon, φ, and the relationship between θ and φ is itself mediated, in her diagrams, by the models a theory permits. Discursive survival asks whether a theory can be understood by informed, good-faith readers and can withstand sustained scrutiny within its intended subfield: clarity of writing, conceptual transparency, consistent use of terms, the difference between a theory that is genuinely pre-formal and one that is merely obfuscated. Empirical interface asks how a theory relates to observation: what formalisms it permits, what methodologies instantiate it, what a given pattern of results should be expected to do to the theory. Minimising harm asks scientists to reckon with a theory's genealogy: whether it silently inherits the assumptions of eugenics or physiognomy, whether it erases the contributions of the people (often women, often people of colour) who actually built it, whether its use has caused or could cause harm in the world.
This is a serious piece of metascience. It refuses the pretence that theory evaluation is or should be reducible to goodness-of-fit, and it insists that scientists take responsibility for the full life of their ideas, including the uncomfortable parts of that life. Guest's discussion of the eugenics tree diagram, the image of eugenics drawing "its materials from many sources" and organising them, tree-like, into a single harmonious discipline spanning anthropology, genetics, psychology, and history, set against the poster demanding an end to forced sterilisation, is not decorative. Neither is her account of Margaret Masterman, whose early contributions to machine translation and computational semantics were absorbed into the field without citation, credited instead, decades later, to the students and successors who rediscovered her ideas independently. These sections are the paper's strongest, and they gesture, more than the formal apparatus quite captures, toward the concern this essay will develop: that a theory's life is not fully described by its internal properties, because what happens to it once it leaves its authors' hands is not fully predictable from those properties either.
III. The Assumption Hiding in the Formalism
Guest is explicit, and this is to her credit, that her set-theoretic apparatus ties "theory" to units of publication rather than to groups of people, precisely so that criticism can be aimed at the work rather than at individuals who may since have changed their minds. This is a defensible choice for the purpose she sets herself. But it also draws a boundary, and the boundary is worth naming plainly: her 𝕋, a field's theoretical positions, is a scientific field. It is populated by scientists, by publications, by citation networks. It is not, and is not meant to be, society: Congress, the press, the courts, the dinner table, the classroom, the ballot box.
For the great majority of theories this boundary is no loss at all, because the great majority of theories never cross it. A paper on linear separability constraints in category learning, a refinement of frontotemporal dementia diagnostic criteria, a proposed formalism for Markov blankets: these live and are judged entirely within 𝕋, exactly as Guest describes, by exactly the kind of interested non-bad actors her framework presupposes. The four criteria fit this majority of scientific life almost perfectly, and this is not a minor achievement.
The trouble appears only for the theories that matter most historically, the small set that eventually acquires what I will call mesocosmic reach, meaning that it becomes woven into the ordinary coordination of human life outside the field that produced it. For those theories, at some point in their career, the evaluating community stops being 𝕋 in Guest's sense at all. It becomes an uncontrolled, heterogeneous, mutually contesting ecology of politicians, journalists, generals, judges, teachers, parents, and propagandists, none of whom are bound by the norms of the originating field, many of whom have never read the paper, and some of whom are actively hostile to it. Discursive survival among "interested non-bad actors" is a coherent demand when the actors in question are, more or less, one's colleagues. It becomes a much stranger demand once the actors include a Senate committee, a tabloid editorial board, and a Cold War security apparatus deciding whether a physicist's associations in the 1930s make him a loyalty risk in 1954. The theory has not changed. The community judging it, and the criteria by which it is being judged, have changed completely, and this is precisely the transition Guest's ontology has no vocabulary for, because it was never designed to track a theory across that threshold. It was designed, admirably, to track a theory up to it.
IV. Guest's Own Near-Miss: Embedding, Reach, and (Mis)use
It would be a disservice to Guest's paper to suggest she has no notion of reach at all. She does: the word appears explicitly, nested as a subsection under metaphysical commitment titled "Embedding, Reach, (Mis)use," and engaging with what she means by it sharpens, rather than dissolves, the distinction I am proposing.
Guest's reach is a question about a theory's disciplinary boundaries: is it explicit where in the sciences a given theory legitimately applies, and what counts as a misuse of it? Her example is evolutionary theory, and the tension she describes between evolutionary biologists proper and evolutionary psychologists who import selection-based explanations into domains such as mate preference, political attitudes, and cognitive architecture, where the underlying theory's warrant is far less secure. Her most striking illustration is eugenics itself, diagrammed as a tree that draws "its materials from many sources" (anatomy, biology, genetics, psychology, sociology, genealogy, history, religion) and "organizes them into an harmonious entity," a theory that does not merely claim territory within genetics but violently annexes neighbouring disciplines under its own banner. Guest names this pattern explicitly as a vice: theories that unify competing accounts without the consent of their proponents, that drain meaning from alternative concepts and replace them wholesale, often because the unifying theory is so flexible, so nearly able, in her word, to account for anything, that it can be stretched to cover whatever territory it likes. Legitimate unification, she is careful to note, is possible (classical mechanics, electromagnetism, and spacetime being her examples); the vice lies specifically in unification that is coercive, that steamrolls rather than persuades.
This is a horizontal reach: a theory's spread sideways across adjacent disciplinary territory, still within, or at the disputed border of, science broadly construed. It stays inside something recognisably like 𝕋. Mesocosmic reach, as I am using the term, names a different, vertical or outward movement: not from genetics into psychology, but from any of those disciplines into the unreflective coordination of people who have never read a journal article in their lives and never will. The two are related, and eugenics is in fact a case where a theory achieved both at once: it colonised neighbouring academic disciplines in exactly the manner Guest describes, and it became government sterilisation policy, which is a movement of an entirely different kind, into an entirely different kind of community. But they are not the same movement, and a theory can achieve one without the other. Much specialist science today reaches sideways, contested and cited across subfields, without ever reaching outward into the mesocosm at all.
What makes this worth dwelling on, rather than merely a terminological neighbouring, is that Guest's own diagnosis of the mechanism behind illegitimate horizontal reach (flexibility, near-unfalsifiable elasticity, a theory so expressively powerful it can be bent to explain almost anything) may also be, uncomfortably, part of the mechanism behind successful vertical, mesocosmic reach. Freudian concepts are notoriously elastic: repression, the unconscious, and the Oedipal complex have long been criticised, on grounds close to Guest's own, as too flexible to be properly falsifiable, capable of absorbing and reinterpreting any counter-example rather than being genuinely tested by it. This is very close to the vice she flags in evolutionary psychology and in eugenics. And yet this same elasticity is arguably a precondition, not merely an accompaniment, of Freud's mesocosmic reach: a rigid, narrowly falsifiable theory of dream content could not have been transplanted into millions of ordinary people's self-interpretation nearly as readily as a flexible one could. I do not think this licenses treating reach as evidence of validity; quite the opposite, it is a caution against that temptation, since it suggests that the very property Guest is right to flag as a scientific vice may sometimes be exactly what allows a theory to travel furthest into the mesocosm. Reach and goodness, on this view, are not just different axes; they can, in unfortunate cases, be in tension with one another, which is itself an argument for keeping them conceptually distinct rather than treating high reach as a mark in a theory's favour.
V. Theories as Recursive Careers
Living Value Theory's contribution to this problem begins from a distinction that has no direct counterpart in Guest's ontology: the distinction between nonrecursive, selfrecursive, and interrecursive processes, and the claim that a theory's relationship to its own object depends decisively on which of these three the object is.
A theory of a nonrecursive process describes something that does not respond to being theorised about. The orbit of a planet, the folding of a protein, the radioactive decay of an isotope proceed exactly as they would whether or not anyone has a correct theory of them. Explanation here is purely descriptive; the object holds still.
A theory of a selfrecursive process describes something that changes through its own recursive organisation, most centrally a person's relationship to themselves. When a theory proposes that dreams have meaning, that childhood shapes adulthood, that self-efficacy beliefs affect performance, it does not merely describe the self; it offers the self new terms in which to interpret itself. Explanation becomes partly constitutive. People who accept a theory of themselves are, in a small but real sense, different afterward from who they were before.
A theory of an interrecursive process describes relations between people: kinship, exchange, class, race, gender, political legitimacy. Here the theory does not merely enter a world that already contains it as one more description; it becomes a participant in the very relations it purports to describe, reorganising institutions, expectations, and incentives, and being reorganised in turn by how those institutions, expectations, and incentives take it up.
Guest's four criteria, I want to suggest, are calibrated, implicitly, and not through any fault of the paper's, for theories whose object remains nonrecursive, or for self-recursive theories that stay contained within a specialist community that treats them non-recursively (a theory of memory, evaluated only by memory researchers, behaves for evaluative purposes rather like a theory of protein folding). They begin to strain the moment a theory's object is interrecursive, or the moment a nonrecursive theory acquires interrecursive consequences, which is exactly what happened to nuclear physics between 1939 and 1945, and it is worth dwelling on that case precisely because the physics itself never changed in its propositional content. Only its recursive relationship to the world did. That relationship, and how it changes, is what I am calling a theory's recursive career, and mapping that career requires a variable Guest's apparatus does not contain.
VI. Mesocosmic Reach
The variable I want to introduce is mesocosmic reach: the degree to which a theory becomes recursively embedded in the ordinary coordination of human life, beyond the community that produced it. It is not a measure of scientific importance, and it is not reducible to fame, citation count, or media coverage. A theory can be famous without having any reach at all in this sense: widely known by name, respected as a symbol of genius, and yet touching almost nobody's actual daily coordination. It can also, more rarely, be scientifically modest and still acquire enormous reach because it resonates with something people are already trying to coordinate.
Most theories, across every discipline, have effectively no mesocosmic reach. They circulate entirely within 𝕋 and are never heard of again outside it; this is not a failure, it is the normal condition of specialist science, and it is exactly where Guest's calculus applies without strain. A smaller number acquire intermediate reach: attachment theory, Keynesian economics, plate tectonics shape policy, professional training, and school curricula, without most people organising their own lives explicitly around them. A very small number cross into what might be called maximal reach: Darwin, Freud, Marx, evolution, race, gender, AI. These stop being merely scientific theories and become what I will call mesocosmic objects: things argued about in newspapers, courtrooms, parliaments, and living rooms, whose meaning is no longer under the control of the discipline that produced them.
What makes reach a genuinely new variable, rather than a restatement of "impact" under a different name, is that it is not a property of θ in Guest's sense at all, nor of the horizontal, disciplinary reach discussed above. It is a relation between θ and a historically changing recursive ecology that lies entirely outside 𝕋, and that ecology is not the field. Guest's field-level construct captures how a theory relates to other theories and to the community of practitioners who work on it; it has no place at all for the classroom, the ballot box, or the newsroom, because it was never meant to. Mesocosmic reach names exactly the space her formalism leaves empty, and it is not an empty space by accident: it is where evolutionary theory became scientific racism, where eugenics became forced-sterilisation policy, and where nuclear fission became, within six years, the organising fact of global politics.
One further property of mesocosmic reach is worth stating before turning to the ladder that governs its acquisition: reach is not a fixed property of a theory, settled once and for all at the moment of publication, but a relation that can change dramatically as the surrounding recursive ecology changes, even while the theory's content stays constant. Nuclear physics is the paradigm case this essay will examine at length, but it is not unique. Microbiology carried, for most of the twentieth century, a comfortable intermediate reach, relevant to public health policy, largely invisible to ordinary daily coordination, until a pandemic made viral transmission, incubation periods, and vaccine mechanisms into dinner-table knowledge almost overnight. Artificial intelligence occupied a narrow specialist niche for decades before a small number of consumer-facing systems gave it, within a few years, a reach that touches education, employment, and everyday communication. In each case the underlying science did not become truer or better established at the moment its reach expanded; what changed was its position in the recursive ecology surrounding it. This is the feature of reach that makes it a poor proxy for scientific merit and, at the same time, the feature that makes it worth tracking in its own right, because a theory's capacity for harm, as later sections will argue, tracks this same historically contingent variable rather than anything fixed in the theory's content.
VII. The Recursive Ladder: Where a Theory Enters
Reach is not simply high or low; it is acquired, when it is acquired at all, through a rough sequence, and different theories enter that sequence at different points, a fact that turns out to matter enormously for how quickly, and how violently, a theory can travel.
Some theories begin already inside the mesocosm. Freud's earliest claims, that dreams have meaning, that repression shapes behaviour, that childhood is formative, are, from their very first formulation, claims about selfrecursive coordination that people were already engaged in without a name for it. Nobody had to be persuaded that the domain mattered; the only live question was whether Freud's account of it was any good. Darwin similarly begins already interrecursive, because a theory of common descent is immediately a theory of kinship, ancestry, and human difference, domains people were already, urgently, coordinating around.
Other theories begin outside the mesocosm entirely and have to earn their way in, stage by stage. Climate science is the clearest living example: its first and hardest task is not persuading anyone to change their behaviour but persuading them that atmospheric chemistry belongs to the field of everyday coordination at all: that it is not simply "a bit hotter in the summer" but something people actually have to reckon with. Only once that threshold is crossed can the theory advance to selfrecursive demands (fly less, eat differently) and only later still to interrecursive institutionalisation (carbon pricing, international treaties). Much of the instability of public debate over climate change follows from the fact that different participants are stalled at different rungs of this same ladder, arguing past one another about physics, personal obligation, and collective action as though they were the same disagreement.
Relativity offers a limiting case on the other side. For most people it never really leaves the first, fascinated encounter with a nonrecursive theory (everyone knows Einstein's name, almost nobody organises a Tuesday around spacetime curvature), and it would very likely have stayed there indefinitely, had a specific historical event not forced it, almost overnight, into the interrecursive register.
Before turning to that event, one refinement to the ladder is needed, and it matters enough to state as a general principle rather than introduce quietly inside the case study. What I have so far called, loosely, "recognition of recursive relevance" is actually two distinct thresholds that can be crossed independently, in either order, and on entirely different timelines. A theory can be recognised by institutions, such as states, militaries, funding bodies, and regulators, as strategically or administratively significant long before any lay public has heard of it, exactly as classified research is designed to achieve; and a theory can, more rarely, acquire lay salience, becoming something ordinary people discuss and organise around, with little or no corresponding institutional program behind it, as happens with viral misconceptions or pseudo-scientific claims that circulate widely without any state or funding body ever taking them up. Institutional-elite recognition and public-mesocosmic recognition are not the same rung of the ladder wearing two names. They are different variables, moved by different mechanisms, and a theory's recursive career often depends on the gap between them: how large it is, how long it lasts, and what happens in the interval. The case this essay turns to next is, among other things, an unusually stark illustration of exactly that gap.
VIII. Oppenheimer: A Theory's Recursive Career in Six Years
A. Act One: A Non-Recursive Theory Among Colleagues
Nolan's film, and the historical record it dramatises, open inside a world Guest's ontology fits almost perfectly. Cambridge, Göttingen, Berkeley: young physicists arguing about wave mechanics at blackboards, judged by their teachers and rivals on exactly the criteria Guest specifies: whether the mathematics is coherent (metaphysical commitment), whether the ideas can be explained clearly enough to withstand seminar-room scrutiny (discursive survival), whether they connect to spectral lines and experimental results (empirical interface). Quantum theory in the 1920s and 1930s is a θ embedded in a Θ embedded in a 𝕋, precisely as her formal apparatus depicts it. Its mesocosmic reach, in this period, is close to zero. Fission itself, once discovered in 1938, is at first exactly this kind of object: a startling but contained result, discussed by a small international community of nuclear physicists, its significance legible only to specialists. Nothing about the physics, at this stage, requires any vocabulary beyond Guest's.
B. The Institutional Rung, Crossed in Secret
What happened next is the clearest illustration available of the distinction just introduced, and it is worth pausing on before moving to Los Alamos, because it is easy to tell the Oppenheimer story as though the theory sat quietly until Trinity and then detonated straight into world politics. It did not. Within a year of fission's discovery, the theory had already crossed the institutional-elite rung of the ladder, while remaining, by design, almost entirely absent from the public one. The August 1939 letter to Roosevelt, signed by Einstein at Szilard's urging, and the Advisory Committee on Uranium it prompted, mark the moment a small circle of state and military officials recognised the theory's strategic significance, years before Trinity, let alone Hiroshima, and years before more than a handful of newspaper readers had ever encountered the word "fission." The refugee networks that carried European physicists, among them Fermi, Szilard, Wigner, and Teller, into American institutions were themselves a form of institutional embedding, moving the theory's bearers, if not yet the theory's mesocosmic meaning, into new state structures. All of this proceeded under an entire apparatus of secrecy whose explicit purpose was to prevent exactly the kind of lay, public uptake that would eventually follow in 1945. The Manhattan Project is, among other things, a startling case of maximal institutional and state embedding achieved through the deliberate suppression of public embedding: the two rungs not merely uncrossed together, but pulling in opposite directions, for the better part of six years, before the gap between them closed all at once.
C. Act Two: The Multimaterial Threshold
Once the state had committed to the project, what changed next was not the theory but its relationship to feasibility, and the change was not conceptual, it was material. Los Alamos is the film's, and the historical episode's, great demonstration that a correct theory is nowhere near sufficient for a theory to act in the world. Knowing that fission releases energy, and even knowing the rough physics of a chain reaction, told the Manhattan Project almost nothing about how to machine plutonium hemispheres to sub-millimetre tolerances, how to design explosive lenses that compress a core symmetrically to the nanosecond, how to transport and shield radioactive material without killing the people handling it, how to organise thousands of specialists under total secrecy, or how to build a facility from nothing in a New Mexico desert. All of this is what LVT calls multimaterial mediation: the enormous, usually invisible labour of translating a symbolic claim into coordinated physical practice, none of which is derivable from the theory itself and all of which could have failed, fatally, at any point. The film lingers on this: the criticality experiments performed by hand, the shaped-charge lens problem, Louis Slotin's later death from a slipped screwdriver, not as texture but as the argument's second premise: a theory does not travel on its own logical strength. It travels only as far as an enormous, contingent, dangerous infrastructure is built to carry it, and that infrastructure has its own failure modes that have nothing to do with whether E=mc² is true.
D. Act Three: Self-Recursive Entanglement
At Trinity, something changes that is neither physics nor engineering. Oppenheimer's famous recollection of the Bhagavad Gita line at the moment of detonation marks the point at which the theory stops merely explaining atoms and starts reorganising its own author's relationship to himself. This is the selfrecursive rung of the ladder, and it is worth being precise about what makes it selfrecursive rather than merely emotional: Oppenheimer does not simply feel something about the bomb; he begins interpreting his own life, responsibility, and moral status through the fact of having built it. The theory has become constitutive of his self-understanding in exactly the sense that separates a selfrecursive object from a nonrecursive one. Nothing in nuclear physics predicts or requires this. It happens because the theory has, by this point, travelled far enough that its bearer can no longer hold it at arm's length as a merely descriptive claim about isotopes.
E. Act Four: Interrecursive Detonation
From Hiroshima and Nagasaki onward, the physics all but disappears from its own story, the public story at least, if not the classified one, a distinction the next section returns to. What takes its place is deterrence doctrine, the arms race, loyalty politics, McCarthyism, and eventually the 1954 security hearing that stripped Oppenheimer of his clearance, a proceeding in which almost nobody discusses quantum mechanics at all. What is adjudicated instead is trust, past associations, ideological reliability, institutional loyalty: an entirely different evaluative apparatus than the one that judged the same man's physics fifteen years earlier. This is the interrecursive rung completed, on the public side, in a matter of weeks after six years of institutional gestation. The theory has become so thoroughly a participant in the relations between nations, parties, and institutions that its own scientific content has become almost beside the point to the community now doing the evaluating, a community that bears no resemblance to Guest's 𝕋 of interested non-bad actors within a field, because it is Congress, the press, and a security apparatus, and because "non-bad actor" is itself one of the hearing's contested questions rather than a stable background assumption.
F. Guest's Categories, Rung by Rung
Laid out this way, it becomes possible to see with unusual clarity exactly where Guest's four criteria hold and exactly where they stop holding, within the career of a single body of theory. Metaphysical commitment and empirical interface do real, necessary work in Act One and are already strained by Act Two, where the relevant questions are engineering questions, not theoretical ones. Discursive survival, reframed as intelligibility to a scientific community, governs Act One cleanly; by Act Four it has been replaced by a survival contest among journalists, generals, and politicians in which technical clarity is close to irrelevant. Minimising harm, the category that reaches furthest in Guest's own framework, is present throughout, but it means something different at each stage, a point developed fully below. No single, unmodified calculus tracks the theory across all four acts, not because Guest's categories are poorly chosen, but because the kind of object under evaluation, and the kind of community doing the evaluating, have changed out from under them.
G. The Great Man Problem, Twice Over
Guest's ontology contains a vice she calls great man theorising: the tendency of a field's historical record to attribute collective, distributed labour to a single prestigious figure, obscuring the students, technicians, and collaborators, often women, often uncredited, whose work actually constituted the achievement. Her examples are Pythagoras and Theano, Lavoisier and Marie Anne Paulze Lavoisier, Kekulé and the chemists who solved the structure of benzene before him. Oppenheimer's own case is a near-perfect instance of the same pattern, and examining it shows something the vice-list format of Guest's paper does not quite bring out on its own: that great man theorising is not evenly distributed across a theory's career. It concentrates at exactly the point where reach is highest.
Within the physics itself, the Manhattan Project was never a one-man achievement, and nobody inside it believed otherwise. Fermi's reactor team, the metallurgists and chemists at Los Alamos, the calculations performed by rooms of human "computers" (many of them women, many uncredited in the popular record in exactly the pattern Guest describes for Masterman), the engineers at Oak Ridge and Hanford who solved the industrial-scale isotope separation and plutonium production problems Los Alamos itself could not have solved alone: all of this collective, multimaterial labour is compressed, in the public record, into a single name. Oppenheimer becomes, in the phrase the era itself used, "the father of the atomic bomb," a title that does exactly what Guest warns against: it reorients credit toward one prestigious figure and away from the thousands whose work the achievement actually depended on.
What the recursive-career framework adds is a reason this compression happens specifically here, rather than being evenly distributed across all of nuclear physics's history. Within 𝕋, among physicists, the distributed nature of the achievement was common knowledge and required no simplification; a specialist community can hold a complicated, multi-author causal story without difficulty, because that is what specialist communities are for. It is only once the theory crosses into the interrecursive register, once it becomes a mesocosmic object that ordinary people, with no access to the internal history of the field, need to talk about, blame, or lionise, that a single legible protagonist becomes useful in a way it was not useful before. A diffuse causal story involving thousands of named and unnamed contributors does not travel well through newspapers, congressional hearings, or dinner-table argument; a story with one recognisable face does. On this reading, great man theorising is not simply a bias that afflicts historical record-keeping in general. It is, in part, a predictable by-product of mesocosmic reach itself: the price a theory's actual history pays for becoming simple enough to be carried by a community that was never going to master its internal complexity in the first place. Guest is right to name this a vice. What her framework does not quite supply, and what a recursive-career account can, is an explanation of why the vice reliably intensifies exactly where reach is highest, rather than being distributed at random across the history of science.
IX. Where Oppenheimer Sits on the Ladder: A Comparative Note
It is tempting to treat the Oppenheimer case as sui generis, but placing it against the other examples already discussed clarifies what is genuinely distinctive about it. Its entry point on the recursive ladder is not unusual: nuclear fission begins, like climate science and like relativity, as a nonrecursive curiosity, with no immediate self- or interrecursive relevance. Early press coverage of fission in 1939 was modest physics news, filed alongside other developments in a rapidly advancing field; nothing about it, at that point, distinguished its public trajectory from relativity's.
What is distinctive, once the institutional and public rungs are properly separated, is that fission ascended both with unusual speed, but not at the same time, and not for the same reasons. The institutional rung was climbed almost immediately: about a year separates fission's discovery from state recognition of its significance. The public rung was climbed later and far more abruptly: nothing changed for the public between 1939 and 1945, and then everything changed within days of Hiroshima. Climate science offers a useful contrast on both counts. Its institutional rung was climbed slowly and comparatively late: the Intergovernmental Panel on Climate Change was not founded until 1988, three decades after Charles Keeling's first systematic measurements of atmospheric carbon dioxide and nearly a century after Svante Arrhenius's original 1896 calculations of the greenhouse effect. Its public rung has been climbed more slowly still, and remains, in large parts of the world, incomplete. Nuclear fission covered in six years, on the public side, a distance climate science has not fully covered in over sixty. The difference is not that the physics was more compelling than the climatology. It is that fission's public ascent was driven by a single, legible, dateable technological event, rather than by the slow accumulation of diffuse and endlessly contestable evidence. A mushroom cloud over a city converts an entire population's coordination overnight, in a way that a rising global mean temperature, distributed across decades and deniable at any single data point, structurally cannot. This suggests a further hypothesis worth setting alongside the recursive ladder itself: how fast a theory ascends the public rung may depend less on the strength of its evidence than on whether its passage can be effected by a single, unambiguous, witnessed event, or must instead accumulate gradually across a public that can always defer the reckoning by one more year.
There is a further irony worth recording. Relativity's own mesocosmic reach, which, on its own terms, remained modest, was substantially increased by fission's ascent rather than by anything relativity did on its own. Mass-energy equivalence became, in popular understanding, "the equation that made the bomb possible," and Einstein, whose direct role in the Manhattan Project was limited to co-signing the 1939 letter, found his theory dragged into the interrecursive register by proximity to a family member's success. Reach, it seems, can be borrowed across a Θ in something like the way Guest's own diagrams show family members sharing citational and genealogical linkage, a reminder that a theory's recursive career is not always its own achievement, but sometimes an inheritance from whichever of its relatives happened to detonate first.
X. Reassessing Guest's Four Categories Through the Case
Having watched the theory move, it is possible to return to Guest's ontology and say something more precise than "it applies early and not late." Each category needs a specific supplement, and the supplement is broadly the same in each case: recognition that the criterion's meaning is not fixed but shifts as reach increases.
Metaphysical commitment. Guest's apparatus for tracking a theory's assertions and its descent through time (θ at t₀ becoming θ at t₁, potentially fracturing into rival successors) is well suited to tracking what nuclear physics asserted and how those assertions changed as evidence accumulated. It has no native vocabulary, however, for what happens once a theory's assertions are taken up as prescriptions by actors who were never party to the original commitments: mass-energy equivalence and the physics of chain reactions did not entail deterrence theory or mutually assured destruction, yet those doctrines were built, by non-scientists, directly on top of the physics as though they followed from it. The theory's metaphysical commitments and its political afterlife are related but not identical, and a calculus built only for the former will misdescribe the latter as a simple extension of it.
Discursive survival. Guest ties survival to intelligibility among interested non-bad actors within a field. Once a theory has high mesocosmic reach, survival is contested simultaneously across communities that share none of the field's norms, and the category of "non-bad actor" stops being a stable background condition and becomes one of the very things in dispute: Lewis Strauss believed himself, presumably sincerely, to be defending national security against a security risk; Oppenheimer's allies believed the hearing itself to be an act of bad faith. Discursive survival at this stage is not a property of the theory's clarity at all. It is an outcome of an entire contested ecology of interpretation, and no amount of expository clarity on the physicists' part could have settled it, because by 1954 clarity about physics was no longer what the argument was about.
Empirical interface. This category performs beautifully for Act One (the theory's relationship to spectral data, to measured cross-sections, to the detected signature of the first chain reaction under Stagg Field) and becomes almost inapplicable by Act Four, where the 1954 hearing does not ask for confirmatory experiments at all. It asks about associations, statements, and character, adjudicated by an entirely different, non-scientific standard of evidence. The theory's empirical interface has, in effect, been superseded by an unrelated evaluative apparatus that happens to be judging the same man.
Minimising harm. This is the category where the reframing matters most, because it is also the category where Guest's own instincts point furthest in the direction this essay is arguing. Her treatment of harm is retrospective and genealogical: whom did a theory exclude in its construction, what abusive ancestor concepts does it inherit, what has been done in its name. These are real and important questions, and nothing here disputes them. But nuclear fission poses a harm question of a different shape entirely, not who was excluded from co-authoring the theory, but what the theory, once it had travelled far enough, made newly possible for humanity to do to itself. And the crucial observation is that harm here is not a fixed property of the θ at all. The identical physics, nuclear fission, the same chain reaction among neutrons and nuclei, is instantiated in a civilian power reactor, where the material and institutional apparatus surrounding it is designed for containment and generation, under regulatory oversight, and in a weapon, where the same chain reaction is deliberately engineered to be uncontained, under military command authority, for the purpose of destruction. Even this pairing needs care: a reactor's harm profile is not simply "low" (Chernobyl and Fukushima are evidence enough of that), but it remains categorically different from a weapon's, and the difference tracks not the physics, which is identical in both cases, but the material and institutional program built around it. What varies is not the science but how far, and into which recursive layer, the theory has been allowed to travel, and what kind of apparatus was built to carry it there. LVT's proposal, against Guest's implicit picture of harm as a discoverable property of the theory itself, is that harm is an emergent property of recursive entanglement: it scales with reach and with the material-institutional program built on top of a theory, not with any feature legible by inspecting θ alone, and a metatheoretical calculus that treats it as a fourth static criterion, symmetrical with the other three, will systematically underestimate exactly the theories that matter most, because those are precisely the theories whose harm has not yet been determined by their content and will instead be determined by their career.
XI. Toward a General Account of Theory-Travel
Pulling these observations together yields a rough general schema, offered as a hypothesis rather than a law. A theory's recursive career, where it has one, tends to move through five moments rather than the four this essay began with. A non-recursive articulation, in which some process is correctly described without yet being recognised as something anyone needs to coordinate with. An institutional-elite recognition, in which states, funders, regulators, or militaries register the theory's strategic or administrative significance, often under conditions, as with the Manhattan Project, that actively suppress rather than promote lay awareness. A public-mesocosmic recognition, logically and temporally independent of the institutional rung, at which ordinary people register that the process concerns their own coordination. A selfrecursive uptake, in which people begin reorganising their own self-understanding or behaviour around the theory. And an interrecursive institutionalisation, in which the theory reorganises relations between people through law, politics, or economic arrangement at the public, not merely the elite, level.
Several qualifications matter as much as the schema itself. Not every theory enters at stage one: Freud enters already self-recursive, Darwin already interrecursive, and this recursive ladder position, not scientific merit, is a large part of why some correct-but-modest theories acquire vastly more reach, faster, than more rigorously established ones. Not every theory that enters at stage one completes the journey: the overwhelming majority stall at stage one permanently, which is the normal, unremarkable condition of specialist science and not a mark of failure. The institutional and public rungs, as the Oppenheimer case shows starkly, need not occur in the order listed, and can be separated by years or, in the case of research that remains permanently classified, indefinitely. The journey is not guaranteed to be linear or permanent: it can stall for decades, as nuclear physics did on the public side between the 1930s discovery of fission and the war; it can be resisted at any rung, as climate policy is currently resisted mainly at the transition from selfrecursive obligation to interrecursive transformation; and it can, in principle, partially retreat. And the trigger that moves a theory from one stage to the next is very often not an argument at all but an event, Hiroshima in the case at hand, which is itself a further departure from any picture in which theories are judged, stage by stage, purely on their merits by a competent community.
One further qualification deserves to be stated as a limitation rather than absorbed silently. The schema as given pictures reach flowing in one direction: outward, from a shrinking specialist community into an expanding lay one, with the theory's content held fixed while its evaluators multiply. This is not always true, and it is worth being explicit about the cases where it fails. Where a lay community has direct, sustained, and high-stakes embodied involvement with a theory's object (patients living with a disease their physicians do not yet understand being close to a limiting case), that community can acquire genuine technical competence and cross back into the specialist field itself, altering its internal practice rather than merely receiving or reinterpreting its output from outside. This is a different mechanism from the selfrecursive uptake described above, which changes a person's self-understanding without requiring them to become a competent participant in the field's own methodology. It is a rarer and stronger form of return traffic, and the schema as stated does not yet predict when it will occur. That it can occur at all is a genuine boundary condition on the model, not a detail to be smoothed over in service of a tidier picture.
XII. Theory-Travel and Science and Technology Studies
A. The Nearest Rival Framework
The questions this essay has been asking, namely what happens when scientific knowledge leaves a specialist setting, how the criteria and communities that evaluate a theory change as it does, and how a theory's meaning comes to be reorganised by institutions, media, and publics who were never party to its original construction, are the working territory of science and technology studies, and specifically of four decades of research on the public understanding of science. Pretending otherwise would be false, and would also weaken rather than strengthen the argument, since the differences worth keeping only become visible once the resemblance is taken seriously. Brian Wynne's study of Cumbrian sheep farmers evaluating official radiological science through local husbandry knowledge and institutional memory is the closest direct precedent to this essay's account of criteria changing from the seminar room to the hearing room. Steven Epstein's account of AIDS activists acquiring biomedical competence and altering clinical-trial design is a still sharper challenge, treated separately below. Bruno Latour's actor-network theory, Sheila Jasanoff's work on co-production and sociotechnical imaginaries, Ian Hacking's looping effects, and Star and Griesemer's boundary objects each occupy some portion of the same ground. A reader who knows this literature will recognise a great deal of what follows, and should.
B. Two Senses of Entanglement
The place to start is a distinction the STS literature itself does not consistently make, because its own method makes the distinction hard to see. "Entanglement" between science and its social surroundings is doing two very different jobs in that literature, under one word. In one sense, Latour's sense, entanglement is close to universal: every published theory has funders, reviewers, a journal, instruments, allies, a laboratory with a budget and a lease. This is real, and worth insisting on against any naive picture of science as a hermetically sealed space of pure ideas, but it is also, once stated plainly, nearly true by definition of anything that gets published at all. In another sense, call it deep or mesocosmic entanglement, entanglement is rare: a theory becoming bound up with sociotechnical imaginaries, state power, national identity, public political order, and the ordinary self-understanding of people who will never read the primary literature. Jasanoff's own work supplies the richest account of this rare sense, and it is worth being precise about her actual position rather than assuming it: she does not, and on her own terms would not, distinguish a state's or an institution's uptake of a theory from a public's. Her framework treats these as produced together or not at all, a commitment examined directly below on its own strongest grounds rather than as a foil. This rare sense is not true of most science, and it is not supposed to be; it is precisely what makes nuclear weapons programmes, genomic ancestry testing, or climate policy interesting cases rather than unremarkable ones.
The trouble is that STS's characteristic method, the deeply worked historical or ethnographic case study, can only be conducted on episodes dramatic and well-documented enough to sustain one, which means every case the literature examines has already, by construction, cleared the bar of Jasanoff-style entanglement before the researcher arrives. Cumbrian sheep farming after Chernobyl, AIDS activism, the atomic bomb: these are not a random sample of what happens to scientific theories in general. They are the extreme tail, selected for precisely the property whose rarity the method then has no way of registering, because the method has no natural denominator: no accounting of the very large number of theories to which nothing of the kind ever happened. This is not a claim that STS studies the wrong things. It is a claim about what a case-study method can and cannot see: it is built to find deep entanglement wherever entanglement occurs, and it will therefore always find it, without being able to say how exceptional the finding is relative to the population of theories as a whole, most of which are produced, argued over, and forgotten entirely inside the specialist communities that made them.
C. Jasanoff's Own Diagnosis, and Where It Stops
Jasanoff's account of this rare, deep sense of entanglement is not a passive backdrop to her work; it is something she has explicitly diagnosed as a gap in the STS tradition she inherited, and moved to fill. The 2009 paper introducing sociotechnical imaginaries, co-written with Sang-Hyun Kim and focused specifically on nuclear power in the United States and South Korea, states plainly that STS had given little sustained attention to how science and technology are taken up by non-scientific actors and institutions, leaving the relationship between science, technology, and political power comparatively undertheorised. Sociotechnical imaginaries, and the larger co-production idiom they extend, are her considered answer to that self-diagnosed gap, not an incidental feature of it. This essay's disagreement with her, then, is not that she has overlooked the problem of power and institutions; it is that even her considered answer to it does not reach the specific case this essay has been built around.
The reason is definitional rather than incidental. A sociotechnical imaginary, on her account, must be collectively held and publicly performed. Public performance is not an optional feature of the concept but part of what makes something an imaginary rather than merely an idea a handful of officials hold. A classified weapons programme, for the period it remains classified, has no public performance to speak of, and therefore does not straightforwardly qualify as an imaginary on her own definition. This is a stronger problem than simply failing to distinguish government from citizens. It means the 1939–45 interval is not a case her central unit of analysis flattens; it is a case that unit of analysis has no name for at all, for as long as secrecy holds.
The obvious reply is available to her, and deserves to be stated rather than waved off. An atomic imaginary was already circulating well before Los Alamos, in H. G. Wells's 1914 novel The World Set Free and in Frederick Soddy's popular writing on radioactive energy, so the classified years are not a void but an occultation of something already public, and the 1945 Smyth Report, the government's own authorised first account of the bomb, is itself a textbook act of co-production, the state engineering the public re-emergence of an imaginary it had spent six years suppressing. Both points hold. Neither explains why an indefinitely deferred, science-fictional speculation and an actual, dated, operational weapon should count as continuous instances of one imaginary rather than two different things separated by exactly the gap this essay has been tracking. A framework general enough to absorb any discontinuity by locating some antecedent public discourse and calling the interval an occultation risks losing the very capacity to mark cases apart that makes it useful in the first place.
It is also worth noting that when Jasanoff and her collaborators did extend the imaginaries concept toward power directly, in the Dreamscapes of Modernity volume, the extension addressed a related but different problem: the asymmetry between states and civil-society organisations contesting rival, already-public imaginaries. That is a real advance, and a harder problem than this essay has attempted. But it presupposes exactly the condition absent from 1939 to 1945: competing visions, plural, all already performed in public, rather than the case of a single, unilateral, wholly non-public programme with no visible counter-imaginary to be asymmetrical against. The gap this essay is pointing to, in other words, sits one step further back than the gap Jasanoff has already, admirably, tried to close.
D. What Latour Gets Right, and an LVT Answer
None of this excuses the essay's own formulations from a real difficulty, and it is worth naming the difficulty precisely rather than defending against it in general terms. Latour's stronger claim is not merely that theories are entangled with institutions; it is that nothing travels through a network of instruments, allies, and institutions unchanged: the bomb is not nuclear theory plus a set of external consequences appended afterward, but a new object, constituted by transformations among physical theory, metallurgy, explosives engineering, industrial organisation, state secrecy, and strategic doctrine, none of which existed as mere context surrounding an untouched theoretical core. Read this way, the claim that "the physics itself never changed" understates what happened between 1938 and 1945, and understates it in a way that matters: metallurgy and secrecy and doctrine did not simply surround the theory, they became part of what "nuclear fission" meant as an object in the world.
This deserves a considered answer rather than a retreat to a weaker claim, and LVT already has the resources for one, drawn from an argument developed elsewhere in this corpus about ordinary language and metaphor: that it is the stability, not the concreteness, of a source domain that allows it to anchor meaning across radically different target domains. Applied here, the claim is not that nuclear fission's propositional content floated, unchanged, above every material and institutional transformation it passed through, that would concede everything to Latour and leave nothing for a typology of theories to classify. The claim is narrower: what persists as "nuclear fission" across the reactor and the weapon, across 1938 and 1945, is a stable recursive core (the same relation among neutrons, nuclei, and released energy) that remains identifiable and re-appliable across otherwise transformed material and institutional settings, in the same way a stable source domain remains recognisable across the very different target domains it gets mapped onto, without ever being encountered in some pure, unmapped state. This is weaker than the invariance claim Latour is right to reject, and stronger than the alternative he would prefer, in which the theory dissolves entirely into its network and nothing stable is left to classify by type. It is also, not incidentally, the only version of the claim compatible with this essay's own argument, since a typology that predicts differential travel by the kind of process a theory's object is requires there to be a persisting something whose kind can be assessed, a requirement Latour's strongest formulation is specifically built to deny.
E. The Loop Back In
Epstein's case deserves to be taken as a direct limitation of the model as stated, not folded in as a minor addition. The recursive ladder, in every version offered so far, pictures reach flowing in one direction: outward, from a specialist community toward an expanding lay one, with the specialist community's share of the evaluative work shrinking as the public's grows. Epstein's AIDS activists did something this picture does not predict. They did not merely reinterpret official science from outside, in the manner of Wynne's sheep farmers; they acquired genuine methodological competence, entered disputes over clinical-trial design on the trial designers' own technical terms, and changed regulatory and research practice from within. This is a fully interrecursive phenomenon, a lay public reorganising the field itself rather than merely being reorganised by their own uptake of it, and it is distinct from Hacking's looping effects, which are selfrecursive: a classification changes how the classified person understands themselves, and that changed self-understanding can in turn pressure the classification to revise, but the classifiers' own methodology is not thereby altered by the classified becoming methodologically competent practitioners. Epstein's case goes further than Hacking's, and further than the ladder as stated allows for. The honest position is that this is a genuine boundary condition the model does not yet predict, most plausible wherever a lay community's stake in a theory's object is direct, embodied, and sustained enough to motivate real technical mastery, which is a rare conjunction, but not one this account should pretend not to need.
F. Gieryn and the Boundary That Almost Matches
Of the literature surveyed, Thomas Gieryn's work on boundary-work comes closest to this essay's founding question, namely what allows a theory to leave the institutional space of science at all, and it is worth naming directly rather than leaving a reader to notice the gap. Gieryn's project concerns how scientists rhetorically demarcate and defend the boundary between science and non-science, protecting jurisdiction, funding, and authority against astrology, creationism, or amateur inquiry. This is a real and adjacent question. It is still not this essay's question, for a specific reason: boundary-work explains how and why the wall between science and everything else gets built and defended, but it does not aim to predict which of the theories that clear that wall will go on to travel far into lay coordination and which will not, still less to ground such a prediction in what kind of process a theory's object is. Boundary-work is a theory of the gate. It is not a theory of what happens, or how far, once something has gone through it.
G. What Is Actually Being Claimed
The precise claim, stated without the exaggeration either side of this exchange has been tempted toward, is this. STS has not left this domain unoccupied, and no version of this essay should suggest otherwise. What STS offers, across Wynne, Epstein, Latour, Jasanoff, and Hacking, is a body of retrospective, case-relative thick description: rich, historically specific accounts of how a particular theory came to be entangled with a particular set of institutions, identities, and political stakes, built case by case, after the fact, from episodes selected precisely because something dramatic happened to them. What the three-part recursivity typology adds is a claim of a different explanatory form: that theories differ, systematically and by type, in how far they are liable to travel, independent of which contingent funding stream, war, or activist movement happens to seize them, and that this difference is visible in advance, from the kind of process a theory's object is, rather than only afterward, from the archive of what in fact happened to it. Whether that prospective claim survives contact with a fuller historical record than a single case can supply is a genuinely open empirical question, and this essay has only begun to test it. But it is a different kind of claim than anything in the STS literature surveyed here, not a better-branded restatement of the same one, and it should be defended as such rather than either folded defensively into that literature or held apart from it by simple assertion that the literature is asking about something else.
XIII. Relocating, Not Refuting, Guest and STS Alike
None of this essay's argument, against either Guest or the STS literature, is offered as a refutation, and it would be a poor reading of either body of work to treat it as one. Guest's ontology is exactly the right toolkit for the overwhelming majority of scientific activity, which never leaves the specialist field, and it is a more historically and ethically serious toolkit than most alternatives on offer, precisely because of the attention it already pays to genealogy, exclusion, and, as discussed above, to the horizontal reach of a theory across its own field's disputed borders. STS, for its part, supplies the only body of theory that has actually done the patient, case-by-case work of showing how deep entanglement is constructed once a theory does travel: the mechanics of translation, co-production, and boundary negotiation that this essay's more schematic apparatus can gesture toward but has not, in a single case study, demonstrated with anything like STS's ethnographic and archival density.
What this essay proposes, against both, is a relocation rather than a replacement. Guest's four criteria remain the right instrument for theories, and for the early careers of theories, that stay within a field's own jurisdiction; STS's case-study apparatus remains the right instrument for reconstructing, in full historical texture, how a particular theory that did travel came to be entangled with the particular institutions and publics it met. What is missing from both, and what a stage-relative, type-based recursive-career model aims to supply, is an account of which criteria are doing real evaluative work at a given point in a theory's career, who is actually doing the evaluating at that point, and, the genuinely prospective claim STS's method structurally cannot make, some basis for predicting, from the kind of process a theory is about, how far and how fast it is likely to travel before any particular funding stream, war, or activist movement has decided the matter case by case.
XIV. Coda
Guest's title asks what makes a good theory, and how we make a theory good. STS's four decades of case studies ask, in effect, what happened to this particular theory once it became entangled with these particular institutions and publics, and how that entanglement should be understood. Nuclear fission was, by any standard internal to physics, an excellent theory almost from the moment of its formulation: parsimonious, empirically well-confirmed, formally tractable, discursively clear enough that a handful of physicists could agree on it within months of Hahn and Strassmann's 1938 results. None of that excellence predicted, explained, or controlled what the theory would go on to do once it left the laboratory, first for a small circle of state officials and then, six years later and in a matter of days, for the world. What decided that was not the theory's goodness but its journey: the multimaterial labour of Los Alamos, the selfrecursive transformation of the man who led it, the interrecursive politics of a hearing room fifteen years later that had almost nothing left to say about physics. To ask only what makes a theory good is to ask a question about a single frame. To ask, with STS, what happened to this theory once it travelled is to ask a rich and necessary question about one film already shot. To ask how theories travel, by type, before the shooting starts, is a third question, harder to formalise than the first and less richly documented than the second, but not reducible to either, and it is the one this essay has tried to keep open.