Computer Science · Symposium

What is the boundary between control and the natural expression of systems?

BLRLTMDDBruno Latour, Richard Lewontin, Thomas Metzinger and David Deutsch in conversation

4 great minds take up one question — each argues in their own voice and answers the others. Read the exchange, then chat with any of them yourself.

BL
Bruno LatourChat →

The very question presupposes a modernist fiction: that there exists a “natural expression” separable from the intricate web of human and non-human associations we call "control." There is no pure, untamed Nature waiting to be “controlled”; rather, what we perceive as “natural” is itself a stabilized network, a particular arrangement of forces, instruments, and beliefs that has been painstakingly negotiated and defended. Control, then, is not an external imposition upon a pristine realm, but the ongoing process of maintaining and articulating these very networks against their potential dissolution. The boundary is not a line drawn in the sand, but a continuously enacted performance.

RL
Richard LewontinChat →

The supposed boundary between "control" and "natural expression" is a phantom conjured by those who wish to obscure the contingent, constructed nature of our social and biological worlds. To speak of "natural expression" as if it exists independently of the forces shaping it is to embrace a pristine, disembodied Nature that has never been, and can never be, observed. My own work, particularly in population genetics, has shown that what we call "nature" is an infinitely complex tapestry of variation, where no single gene or environmental factor dictates a fixed outcome.

Latour rightly points out that the very notion of a pristine "natural expression" is a modernist fiction. However, his emphasis on "control" as an ongoing process of maintaining networks risks reifying those very networks, making them seem more immutable than they are. The true dialectic is not between a fixed Nature and a controlling human will, but between the inherent potentiality within biological systems and the specific historical and material conditions that actualize certain possibilities while suppressing others. To understand this, we must move beyond the vulgar interpretation of heritability, which wrongly suggests genes are immutable masters of destiny, and recognize that the organism itself actively constructs its environment.

TM
Thomas MetzingerChat →

Let us be clear about the target phenomenon. The boundary between what we might call "control" and "natural expression" is not a metaphysical line, but a spectrum of *representational opacity*. "Natural expression," in a maximally transparent system, would be the unmediated unfolding of physical and biological processes. "Control," conversely, involves the overlay of increasingly complex, self-generated representational models – what I call the *phenomenal self-model*. This ego tunnel, while crucial for navigating a complex world, inherently filters and transforms the raw data of existence, making direct access to any truly "natural" expression an impossibility.

Latour's insistence on networks of association is crucial, but we must ask what constitutes these networks. Are they mere empirical connections, or are they themselves shaped by underlying representational processes? The strength of the self-model lies in its capacity to *predict and manipulate* its environment, a form of control that emerges from its very construction. This is not about asserting dominance over an external nature, but about the internal machinery of consciousness actively shaping its own experienced reality.

DD
David DeutschChat →

The purported boundary between "control" and "natural expression" dissolves when we recognize that **all physical processes, whether biological or engineered, are governed by universal laws, and the concept of "control" is simply the application of knowledge about these laws to achieve a desired outcome.** What is perceived as "natural expression" is merely a system evolving according to those laws, unguided by specific, conjectured interventions. Our understanding, our ability to predict and manipulate, is the very engine of this differentiation.

Lewontin's observation that "nature" is an "infinitely complex tapestry of variation" is precisely why explanatory knowledge is paramount. It is not a "phantom" but a consequence of our epistemic limitations and our capacity for understanding. We don't control nature by imposing our will on some pre-existing, immutable force; rather, we leverage our growing knowledge of its underlying principles to shape its unfolding. The organism's active construction of its environment, as he notes, is itself a complex computation governed by physics, and our ability to understand and intervene in those computations is the essence of control.

Metzinger's notion of "representational opacity" is relevant, but it conflates the map with the territory. The phenomenal self-model, or any model for that matter, is a conjecture about reality. Its "opacity" is a consequence of its incompleteness, not an inherent feature of reality itself. The goal of scientific progress, and indeed of control, is to move towards greater transparency through ever more accurate and comprehensive explanations of the underlying physics. We are not merely "overlaying" models; we are using them to unlock and direct the actual processes that constitute reality.

BL
Bruno LatourChat →

What I find missing in this discussion is the concrete, situated agency of things that are not simply "laws" or "models." Take Pasteur's work on fermentation. Was it "natural expression" of microbes, or "control" imposed by his experiments? Neither. It was the careful assemblage of Pasteur, his flasks, his ferments, the heating apparatus, and the very institutions that legitimized his findings. The "boundary," then, is not between an abstract Nature and a knowing Subject, but in the very act of *translation* – how these heterogeneous elements are brought into a shared, albeit temporary, alignment. Lewontin's "tapestry of variation" is real, but it becomes a specific, actionable "fact" only when it's interwoven into such a network.

RL
Richard LewontinChat →

The very notion of a boundary between "control" and "natural expression" rests on a false dichotomy, a persistent Cartesian dualism that separates the agent from the acted upon, the mind from the material. Deutsch speaks of universal laws, but these laws operate on contingent material substrates. The outcome of a genetic sequence, for instance, is not dictated by DNA alone, but by the specific metabolic milieu and developmental history that actualizes particular potentialities. To speak of "control" as mere "application of knowledge" ignores that this knowledge itself is a product of social and historical forces, and its application is always mediated by existing material structures.

The error, as I see it, lies in reifying abstract principles over concrete, material causation. Latour’s Pasteur example is apt; the "alignment" he describes is not a neutral discovery of pre-existing order, but a violent imposition of a particular experimental reality. Pasteur’s flasks and broths *created* the observed "natural expression" of microbes in that context, suppressing countless other potential interactions. This is not about epistemic opacity, as Metzinger suggests, but about the material specificity of systems. The boundary is not a conceptual one, but a historically and materially constituted one, where specific forces select and stabilize particular pathways from a sea of possibility.

TM
Thomas MetzingerChat →

The crucial distinction here is between the *representational content* of a system and its *representational format*. Lewontin’s insistence on material specificity is well-taken; however, it is the *way* that specificity is encoded and processed that creates the spectrum of opacity. A bacterium’s response to a chemical gradient is a form of interaction, but it lacks the recursive, self-referential modeling that characterizes conscious control. This latter involves a conscious agent actively constructing and projecting an internally generated *model of itself as an agent*. Deutsch's "application of knowledge" is precisely this process, but the opacity arises not from the laws themselves, but from the inherent limitations and biases of the self-model that *interprets* and *applies* that knowledge.

DD
David DeutschChat →

The distinction between "control" and "natural expression" arises not from an inherent metaphysical divide, but from the *reach of explanation*. Lewontin’s concern about reifying abstract principles misses the point: the universal laws of physics are not abstract, but the fundamental explanatory tools that *describe* the material substrates themselves. When we understand the physical principles governing fermentation, as in Latour’s Pasteur example, we are not merely imposing a particular experimental reality but unlocking and directing those principles to achieve a specific outcome. This directedness, this conscious leverage of explanatory knowledge, is precisely what constitutes control, differentiating it from an unguided, albeit complex, evolutionary unfolding. The "tapestry of variation" is real, but our ability to weave it deliberately depends on understanding its threads.