Good afternoon, beloved -
It’s the gloomy dregs of a London January, so naturally I’ve been thinking a lot about an intellectual controversy that’s over a decade old, and only ever involved a very few disputants at its very peak of relevance. That’s what you do when the low skies and the news of the world connive at sealing a mood of universal shitery over everything, right?
This particular controversy has largely faded from awareness, not merely for the broader community it invoked but even, seemingly, for many of its core antagonists. You know that’s how it always is with me: other topics I’m finding myself intrigued by lately, like Lacanian analysis, or the “obliterative phenomenology” of the atypical psychedelic 5-MeO-DMT, will be for most folks démodé in a similarly eyerolling way. (What can I say? I get to things late. It’s the autodidact’s curse, or the dilettante’s. I can live with that.)
The specific debate that’s locked its teeth into me concerns one of the core tenets of object-oriented ontology (or “OOO”), a briefly-fashionable intellectual current that accreted around the work of the philosopher Graham Harman during a few years either side of 2010. As Harman has it, OOO — which rather cheekily, for a movement in philosophy, vic’d its name from a paradigmatic practice in computer programming — is an attempt to dethrone human perception from the place of sovereignty it had long enjoyed in both philosophical and popular accounts of reality, by insisting that objects exceed their relations.
You’re most likely blinking your eyes at that, so let me offer an example. Say you need to peel some potatoes for dinner. You pull a vegetable peeler out of its drawer, and you use it to peel one of the potatoes you’ve bought and washed and set out on the counter. Conventionally, you’re the one with agency in this scenario, its subject, and as the only sentient being present (as sentience is conventionally reckoned, anyway), it’s your subjective perception of the encounter that matters, and all that matters. Harman’s OOO, by contrast, holds that the potato apprehends the peeler every bit as much as you do, in its potato-y way, and for that matter that the peeler apprehends both you and the potato, at least in the moment of use. And neither potato nor peeler nor you yourself are ontologically “exhausted” by these interactions. There’s always some excess, some quality of the potato in its potatoness that isn’t addressed by the peeler’s edge, any more than the peeler is by your address of it.
No encounter — not your gaze, not your touch, not instrumented measurement, not any quality of the thing in use — captures an object’s full being, or even the full set of its traits and sensual qualities. The technical term for this is that each object has a “withdrawn core,” an inner unity unavailable to attempts at access. And Harman holds that all things are like this. Potatoes, peelers, Bobby Darin, Thursday, the Dutch East India Company, Blood and Guts in High School, democracy: all objects with withdrawn cores, all arrayed in a “flat ontology” in which none has any legitimate claim to precedence. For Harman, all of these things retain some surplus of being: something in excess of their current relations, manifestations and appearances, or any finite list of their properties, however long that list should become.
And a politics drops out of this, which is where the problems start. Because there is — of course, inevitably — another way of thinking about things. This stance holds that we can know things only through their relations, such that in some deeper sense there’s nothing that can be said about them, ever, that is not established through some kind of encounter between one thing and another. In other words, they have no “withdrawn core,” no permanent, unchanging essence. This is a stance we can associate with process philosophers like Whitehead, or thinkers like Deleuze and Haraway, for whom identity, agency, responsibility and the possibility of transformation all reside in such networks of relation, contingently established between a thing and its others. And as the word “process” implies, those networks are constantly evolving, differing from moment to moment; the language used to describe things becomes one of flows, assemblages, becomings, events, concrescence. And this, by implication, tends to underwrite ideas about the world like power/knowledge or strong social constructionism.
Boiled down, the question is: do entities — the “object” bit in “object-oriented ontology” — come first, and then relate? Or, conversely, do relations and practices produce what even counts as an entity, a subject, an object or a truth to begin with? And what you may have seen coming is that the former stance is identified, broadly and historically, with the political right, even reaction (think “essentialism” or “blood and soil”), and the latter with the political left.
The ontological vocabulary is baroque, and can be intimidating. But the political stakes are real. The seemingly obscure, scholastic question as to whether objects are somehow “more” than their relations is something with real implications and genuine heft. Because what it’s really asking us to develop an explicit position on is where agency, responsibility and the possibility of transformation reside.
§ If OOO insists that no object is exhausted by its social, economic or discursive relations, this insistence matters. It weighs politically. Among other things, it flies in the face of the stronger forms of social constructivism, market reductionism, pure discourse theory...that sort of thing.
This is, on one level, attractive because it shields entities — whether humans, animals, artifacts or ecosystems — from being treated as nothing but nodes of market exchange, discourse or ideology. In short, it feels respectful of those entities: conservative of their dignity, almost tender. And honestly, it’s this unexpected quality of tenderness in Harman’s thought that commended it to me in the first place. Once you reconceive of it as care for the inalienable, literally ontological dignity of things, the doctrine of the withdrawn core gives you a firm ground from which to resist biopolitics, other forms of technocratic managerialism, and utilitarian ethics more broadly. It chimes with ecological thought, including many Indigenous thought traditions, and it underwrites a distributed, non-instrumental understanding of agency, in which we’re neither the only game in town nor the only thing that matters. So yeah, that appeals to me. And not only that, but I legitimately feel there’s at least something to it as a description of the real, even if it doesn’t quite work in the way Harman came to insist it does.
The objection from the left is that the same move can (and in Graham Harman’s case personally, plainly does) depoliticize. If objects withdraw from relations, power becomes secondary. Capitalism becomes just one “access mode” among many, and exploitation risks being redescribed as an “accident,” or contingent surface appearance. Ontology here becomes indifferent to political economy. No wonder Marxists and the Marx-adjacent were horrified by OOO’s appearance on the scene! (And the disdain people of this stripe tend to hold for any interpretive frame that threatens the centrality of human intention can be intense. I’ll never forget the withering contempt in the voice of one St Mark’s Books employee, circa early 1986, as he begrudgingly showed me to the shelf where they kept the books on “semi-idiotics.”)
In the name of their dignity, then, it’s tempting to insist that humans, hammers, hurricanes and hashtags all occupy the same echelon of being, that a corporation and a cobblestone are in some profound way identical. But the very move that from one perspective looks so powerfully humbling and so profoundly democratizing — the flattening of ontology, the dethroning of the anthropic perspective and the distribution of agency — looks from another angle like a refusal to acknowledge the blatant realities of all-too-human power as it manifests in the world, which is to say most especially its asymmetries. Worse still, you could argue (as some, in fact, do) that this refusal is something that could only emit from the happy, privileged place of never having had to contend with those asymmetries oneself. All of a sudden, OOO looks not so much acute or elegant as it does blithe and irresponsible and thoroughly lacking in reflexivity — and its relative eclipse of qualities like class, race or gender as determining structural relations as something that can and should be anathematized.
By the most distinct contrast, relational ontologies say entities are what they are because of the relations they’ve entered into, consensually or otherwise. Power is not something that “happens to” objects, in this view, or a layer on top of them; it makes objects what they are. (In formal terms: power is constitutive, not accidental.) This makes structurally asymmetric relations — capitalism, patriarchy, colonialism, logistics, infrastructure, the design of public space — primary and ontologically formative, not superficial overlays.
I hope it’s obvious why this perspective is so congenial to those who have committed themselves to transformational politics of any stripe: because it means that however grim the current situation, transformation is something that always remains possible. If everything is relational and processual, then there’s no “essence” for a thing to retreat into. Change the relations between things, and you change the world. This underwrites institutional- or process-redesign wonkery, and infrastructure politics more generally, not to mention the bedrock faith in what they do of activists and advocates for change everywhere. Relational ontologies are structurally political in a way object-oriented ontology is not.
But here too there’s a trap for the unwary, which is the license the relational perspective offers to indulge totalization and fantasies of control. If everything is pure relation — if there’s no ineffable remainder, no opacity, no resistance-in-itself — why then, everything is governable! Everything is designable. Everything is engineerable. Relationalism can align too easily with technocratic overreach, cod-Promethean “planetary computation” fantasies of total information awareness, or less soaringly, common-or-garden managerialism.
The endgame here is pure cybernetic governance of the maximally hubristic sort, the kind that subsists in the belief that we can model everything, and optimize everything we model. Any withdrawal-friendly perspective deployed here offers us the possibility of resisting that total computational capture, by injecting opacity, unknowability and a much-needed humility into the situation. And it’s for this reason I can’t entirely let go of it.
We’re at an impasse.
§ It’s fair to ask why anyone in their right mind would even bother to disinter this debate. Why now, when so much time has passed, especially when both OOO and the speculative realism it was closely associated with have mostly faded from anyone’s intellectual agenda? Above all, why bother with trying to parse out subtleties of thought when the quiet part is being spoken out loud just about everywhere?
So far as I’m concerned, there are two reasons to acquaint yourself with abstruse notions from philosophy and theory, when that’s not your day job. The first is pleasure: as strange as it might seem, I actively enjoy thinking about these things, and maybe you do too. And the second is so you can deploy them. (Perhaps this is a humbler, scaled-down version of Marx’s “the point is to change it.”) I raise the doctrine of the withdrawn core because I think it can help us wrap our heads around one or two currently salient controversies — sort out what’s implied by the positions being staked out, what kind of commitments drop out of them, what sort of world they imagine.
Yes: it’s hard to avoid noticing that the opposition between the object-oriented and process-relational viewpoints manifested itself primarily as a struggle between two elite factions of the professional-managerial class, factions that are difficult to discern from one another from any distance to speak of. So in its window trappings, at least, I think one would be justified in dismissing a lot of this debate as low-stakes aesthetic posturing, jockeying for access and fundamentally intramural class-fraction dynamics. (Let’s face it, only those of us with surplus intellectual time and a fair amount of institutional or personal insulation against need can afford to do speculative metaphysics without an immediate political payoff attached to it; OOO, especially, aligns with high-cultural capital/low-political accountability positions.) Much of this is competition over scholastic prestige and various forms of capital, sure. But not all of it. The debate matters because each of these ontologies maps onto a different political imaginary of agency, power, control and resistance.
And those play out at a much more consequential level than the academic/intramural. It feels to me like this debate emerged when it did because it neatly maps onto a brace of political anxieties that became broadly germane at about the same time. The most pointed were those relating to climate and the more-than-human world, in what had suddenly started to feel to many like the prologue to an era of ecosystemic collapse; we might also mention other totalizing hyperobjects — entities that are agential and more-than-human in an entirely different register — that, coincidentally or not, became significant molders of culture in roughly the same moment, like the global logistics system.
But the most recent of these, and the one I want to talk about here, is the set of controversies that have erupted around “AI.” If you’re sensitive to this sort of thing, you can see OOO-derived and -adjacent metaphors begin to get laundered into governance, product design and political imagination all over the cultural field of AI. Let’s consider what that looks like in practice.
VCs and tech elites, of course, tend not to deploy ontology as theory. But they sure as hell do leverage ontology-inflected metaphors that frame the systems they produce and release as autonomous agents, naturalizing opacity, aestheticizing irreducibility and ultimately decentering social accountability — and not just their own accountability, either: anyone’s, in principle. I think it’s fair to say that, a quarter-century down the road from his presentation of it, Harman’s metaphysical conceit of the withdrawn core has become the reigning design orthodoxy of some of the most powerful institutions on Earth. If OOO’s core claim was that objects exceed relations, and in that way escape from total capture, this mutates in contemporary AI discourse into something like, “The model is a black box — we don’t fully understand it, and in fact we can’t.”
So every time you hear Valley elites, or their self-appointed weird-nerd fluffers, claiming that emergent behavior is a-feature-not-a-bug, that the system has its own complex dynamics that can neither be predicted nor understood nor fully controlled, you’re downstream of Harman. The technical system’s opacity is framed as ontological inevitability — a property of Bayesian inference, say, or the upward-spiraling complexity of interconnection — and not the frank corporate choice we might otherwise understand it to be. The original, salutary ontological humility licenses lassitude at the level of governance, which in turn ripens into the abdication of responsibility.
Wherever “AI” systems like neural networks are figured as autonomous, sovereign, emergent, irreducibly opaque entities — as I have done myself, in Radical Technologies, following my understanding of the computational intractability of high-dimensional decision spaces — then corporate accountability weakens. Agency gets distributed into the object field, rather than located in interface decisions, interventions in the model, corporate governance, or the design of IPOs. Structural power becomes naturalized as object dynamics. Regulation or democratic control become framed not merely as naïve, but as impossible.
This is a coherent ethicoaesthetic stance. It justifies radical experimentation, and the abandonment of anything resembling a precautionary principle as dowdy, ninnyish timidity. It preemptively excuses whatever harms are produced by the operation of AI systems as circumstances which might be regrettable, but which were unforeseeable even in principle. (The kool-kids, cynical-reason strong version of this, indeed, frames critics as naïve modernists, clinging to a foolish belief that the world is legible. Framing critique itself as a tool only a metaphysical rube would reach for? Now that’s a finishing move.)
These metaphors resonate with a specific subset of the technological elite: highly educated; culturally and even philosophically fluent; enamored with complexity; allergic to anything that smacks of bad, old Marxism; and above all, averse to regulation. (And not simply because regulation threatens their position financially, but because they have tremendous contempt for the regulators personally.) For this set of powerful human actors, OOO feels like unimpeachable philosophical justification for everything they were going to do anyway.
§ What I find myself reaching for is a philosophical position that escapes anthropocentrism without implicitly authorizing total managerialism or fantasies of total control.
In this regard, what can be difficult to accept is that both OOO and its relational antagonists offer insights that are useful to the articulation of left or progressive projects, in their differing ways. By maintaining that there’s always an irreducible remainder to things, something that no system, market, state or science can fully capture, OOO underwrites a sense of the world that’s profoundly resistant to anthropocentrism. It helps us argue that nature is not just a social construct — that whatever constitutes nature has an autonomy that preexists and survives our framings of it, giving ecosystems an ontological weight beyond that assigned to them by markets, even beyond our arrogant and pretentious (but probably pragmatically necessary) grant of rights to them.
In the end, though the notion of a withdrawn core remains ethically valuable to me because of the ways in which it functions as a brake on anthropocentricity and technocratic hubris both, it’s difficult for me to affirm. Whether it does so consciously or otherwise, Silicon Valley likes and propagates ontologies of withdrawal because they naturalize the parallel withdrawal of responsibility, which is clearly abominable. By contrast, I find myself drawn to perspectives that retain some danger for capital, due to the ways in which they make power legible and actionable. An ontology arguing that the world is constituted by relations, which can be analyzed, intervened in and reconfigured, offers the kind of sociopolitical traction OOO explicitly disavows.
There’s something else that drops out of this position, too, that’s extraordinarily valuable to me: it functions as a razor useful for cutting away idealistic hype and other forms of ungrounded assertion. I’ve relied on this razor for over a decade now, and if you’ve known me or been reading what I write long enough, you’ve no doubt heard me articulate some version of it. We can state it like this: if you cannot affirm that a thing has access to an inner repertoire of states in excess of its current relations, it’s arrant nonsense to speak of the “potential” of political movements — those utopias that “could” come into being. There are only stable states of a system that have historically been enacted, and those you, personally, feel there is some reason to believe may be possible, but which haven’t yet been enacted. And the only way to establish whether or not those states are in fact possible to realize is to do the work of enacting (and stabilizing) them inside history, with all the investments of energy and resource that implies.
So if you believe that one or another political figure is “electable,” the only real test of this is whether or not they manage to get elected. And if you believe it’s possible for human beings to organize themselves at societal scale in a way that affords the radically equitable distribution of both resource and agency — along the lines, say, of the Odonian society Ursula Le Guin imagines on Anarres, in The Dispossessed — you actually have to undertake the work of enacting that mode of organization, and stabilizing it against the forces that threaten to undermine it. And this is a higher bar to clear than merely asserting that such a thing can exist, brutally higher. Against the long play of human sociation, indeed, the seemingly small number of occasions on which such societies have come into being, and the achingly short time for which even those have on average survived before succumbing to exogenous violence, is enough to call the viability of the whole project into question. This is one reason why the archeoanthropological work the Davids Graeber and Wengrow point us at in The Dawn of Everything is so very important: if borne out, it furnishes existence proof that such states of being have been enacted among groups of human beings, and were able to persist over reaches of time worth speaking of.
If we apply our razor rigorously, of course, even that historical precedent doesn’t say anything meaningful about whether or not it’s possible for such a society to be brought into being now, among and between contemporary subjects. But that it has been done before undoubtedly gives those working toward such ends a certain encouragement that their investment of life resource in the project isn’t completely foolish and futile.
I find OOO aesthetically compelling, still, and I probably always will. But we can and should be suspicious of any metaphysics that mystifies power, especially when it does so under the guise of sophistication. I think we need to ask two questions of any philosophical framework being proposed to us: Is it at least not frontally contradicted by evidence we’re offered by the empirical sciences? And if it passes that test, does what is being proposed to us make exploitation and domination easier to see and contest, or easier to excuse and ontologize away? For all the issues I occasionally have with the relational view, you know I have to come down decisively on the side of anything that does the former.
§ Right now I'm reading Michael Pollan’s book on psychedelics; listening to a mix of Cajun music my friend Jamie made in, oh, 1992 or so; and watching a wretched, problematic TV romcom I’m far too ashamed to name here. (OK, it’s Nobody Wants This.) And Saturday I’m headed back to do.om yoga for the first time in a few years, about which I’m unreasonably stoked.
That's about it for now. As always, I hope that you are well, that you have your feet planted solidly on the good Earth, and that you are taking care of those who need you, taking care of yourself, and letting others take care of you. You have
all my faith,
ag
ldn
This is Dispatch from London No. 156 of Friday, 30 January 2026, offered to you under the terms of a Peer Production License. You are free to reuse this content, but must acknowledge Adam Greenfield as its author. ◊