The Possible, the Real, and the Actual: Against the Plenitude of Suffering

The Gap That Free Agency Reveals

We begin with a phenomenon everyone knows from the inside: deliberation.

When you face a genuine decision — whether to take the new job, whether to speak up, whether to betray a confidence — you entertain multiple outcomes. You imagine yourself choosing A; you imagine yourself choosing B. Both scenarios are conceivable. You can hold them before your mind, reason about their consequences, feel the pull of each. This is what deliberation is: the exploration of a space of conceivable outcomes.

But here is the crucial point, drawn from our analysis of free agency in earlier work. You are not an abstract deliberator. You are a specific agent with a specific recursive self-modeling decision architecture — a system whose world model contains a node representing your own decision, recursively connected to the algorithm that does the modeling. You have a specific character, specific diachronic commitments, a specific history of self-legislation. Given all of this — given the complete specification of who you are — many of the outcomes you entertain in deliberation are not metaphysically possible.

The recovering alcoholic who deliberates about whether to drink is entertaining a conceivable outcome. But given who she has become — given the recursive self-model she maintains, the diachronic commitments she has legislated for herself, the character she has built through years of practice — the outcome in which she drinks may not be metaphysically possible. It is consistent with some abstraction of her situation (a coarse-grained model that treats her as an arbitrary agent facing a binary choice). It is not consistent with the complete metaphysical ground of her situation (the full computational structure of her decision architecture, her self-model, her embeddedness).

This is not a claim about determinism. Our account of free agency is compatibilist. The point is not that the future is fixed by prior causes in a way that eliminates choice. The point is that the agent’s nature constrains which outcomes are genuinely possible, and this constraint is tighter than what conceivability alone reveals. The deliberative imagination surveys a wider space than metaphysical reality contains.

Consider a second example to bring out the generality. A judge deliberating on a sentencing decision entertains a wide range of possible sentences. She can conceive of imposing the maximum, the minimum, probation, community service, any number of intermediate terms. All of these are “possible” under the abstraction “a judge choosing a sentence.” But the judge is not an abstraction. She has legal training, precedent knowledge, a theory of justice she has refined over decades, a self-model as a fair adjudicator, specific commitments about proportionality. Under the complete description, most of the conceivable sentences are not genuinely possible — they are ruled out not by external constraint but by who she is as a deciding agent. The wide range she entertains in deliberation is the space of conceivability. The narrow range she might actually produce is the space of metaphysical possibility, given her nature.

The point generalizes beyond agents. Any sufficiently structured system has identity conditions — structural properties that make it what it is — and those identity conditions constrain what can happen to it or through it. The identity conditions for a crystal lattice constrain which deformations it can undergo. The identity conditions for a living cell constrain which chemical reactions it can participate in. The identity conditions for a deliberating agent constrain which decisions it can produce. In each case, there are conceivable transformations that are ruled out by the system’s nature — not by logical contradiction under some universal abstraction, but by logical inconsistency with the complete structural specification of the system.

Useful Abstractions and Empirical Divergence

Why does conceivability outrun metaphysical possibility? Because we reason under abstractions, and abstractions are incomplete.

An abstraction captures some structure of a situation while discarding the rest. Abstractions are indispensable for reasoning — without them, we could not think at all. But precisely because they discard structure, they are consistent with more outcomes than the full situation allows. The conceivability of an outcome reflects its consistency with an abstraction of the situation. The metaphysical possibility of an outcome requires consistency with the complete metaphysical ground.

Consider a physical analogy. Newtonian mechanics is a superb abstraction of gravitational dynamics. Under this abstraction, certain trajectories are “possible” — consistent with the Newtonian equations. But some of those trajectories are not metaphysically possible. They are ruled out by general relativity, which captures structure that Newton’s abstraction discards. The Newtonian trajectories are conceivable — you can compute them, plot them, reason about them in full detail. They are artifacts of an abstraction that empirically diverges from reality at certain scales.

A subtler example comes from quantum field theory. The naive perturbative expansion of a quantum field theory generates infinities — terms in the calculation that diverge to infinity. These infinities are “consistent” with the formalism in the sense that the formalism produces them. But renormalization reveals that they are artifacts of an incomplete description — the bare theory needs to be supplemented with counterterms that encode the effects of high-energy physics the naive formalism ignores. The infinities are conceivable (the formalism generates them) but not metaphysically real (they are artifacts of an abstraction that diverges from the complete theory).

An important caveat: the mismatch between abstractions and complete theories can go in both directions. Classical mechanics forbids quantum tunneling; quantum mechanics permits it. Pre-relativistic physics forbids mass-energy conversion; special relativity permits it. In these cases the coarser theory is too restrictive — it rules out states the finer theory allows. So the mere existence of a gap between conceivability and metaphysical possibility does not tell us, by itself, whether the gap favors more suffering or less. It tells us only that the plenitude argument is wrong to assume no gap exists — wrong to treat conceivability as a reliable guide to metaphysical possibility. The direction of the gap for suffering specifically is a further question, one that requires independent argument. We provide that argument in Section III.

With that caveat in place, consider how the gap applies to consciousness and suffering. A functionalist abstraction of consciousness — one that identifies conscious states with certain computational or informational patterns — may be consistent with suffering being instantiated in arbitrary mathematical structures. But that abstraction may diverge from the actual metaphysical conditions for genuine suffering in exactly the way that matters. On our account, genuine phenomenal suffering requires homeostatic acquaintance — a real causal relation of regulatory coupling between monitoring and monitored processes, sustained over the temporal window of the phenomenal present. A mathematical structure that merely describes suffering — that carries the label without any structural isomorphism to a genuine suffering system — fails at the bottom rung: it is not metaphysically possible as suffering at all, just as Newton’s equations diverge from general relativity at high energies.

But what about structures that are isomorphic to a suffering system under some coarse-grained abstraction — that carry the real pattern, not just the label? These cannot be dismissed at the bottom rung, precisely because real patterns are real on our own account. Here the work shifts to the middle rung: these are Boltzmann Brain cases. The suffering pattern may exist as a genuine real pattern, but its extension to a complete computational history — an observer-moment embedded in a coherent causal environment with the full regulatory architecture that genuine phenomenal suffering requires — may demand astronomically improbable conditions. Such a structure would be metaphysically possible but carry negligible measure in the distribution over observer-moments. It would be a possible actuality, but a vanishingly lightweight one — conceivable, genuinely possible, yet structurally marginal in the space of potentialities. The plenitude argument treats all such structures as equally weighty. The measure corrects this.

The same structure applies to free agency. A coarse-grained model of an agent — one that represents them as “a decision-maker facing options A and B” — is consistent with both outcomes. A finer-grained model — one that includes their character, commitments, and beneath all of that, a specific neurology — may be consistent with only one. The conceivability of the eliminated outcome was an artifact of the coarser abstraction.

Note what kind of claim this is. We are not saying that the eliminated outcomes are logically consistent in one framework and logically inconsistent in another. We are saying that they are logically inconsistent with the complete metaphysical ground of the situation, and that this inconsistency is invisible at coarser levels of description. The abstractions that make the alternatives look possible are either incomplete (they leave out relevant structure) or empirically divergent (they track useful patterns that don’t correspond to metaphysical joints). In either case, the conceivability of the outcome reflects our epistemic limitations, not genuine openness in reality.

Logical Uncertainty and the Boundedness of Reasoners

This connects to a research program in the foundations of mathematics and AI that bears directly on the plenitude question: the study of logical uncertainty, developed most rigorously by Scott Garrabrant and collaborators at the Machine Intelligence Research Institute.

Classical probability theory handles empirical uncertainty — uncertainty about which state the world is in — but assumes logical omniscience: the reasoner knows all the logical consequences of what they know. Real reasoners are not like this. A mathematician who knows the axioms of arithmetic does not thereby know whether Goldbach’s conjecture is true. The truth is logically determined by the axioms, but the mathematician cannot see the determination. She is logically uncertain — uncertain about a fact that is, in principle, logically fixed.

This is not a marginal technical problem. Standard Bayesian epistemology assigns probability 1 to all logical truths and probability 0 to all logical falsehoods — the reasoner is supposed to already know everything that follows from what she knows. But actual reasoners, including all physically realizable ones, cannot compute all the consequences of their beliefs. A chess player who knows the rules of chess does not thereby know whether White has a forced win from the starting position. The answer is logically determined, but the computation required to extract it exceeds any feasible resource bound.

Garrabrant’s program — developed in the 2017 paper “Logical Induction” and subsequent work — constructs formal frameworks for reasoning coherently under this kind of uncertainty. The central result is the logical induction criterion: a computable sequence of probability assignments over logical sentences that, in the limit, converges to the correct truth values, while at every finite stage remaining coherent in a precise sense — no efficiently computable trader can exploit the reasoner’s probability assignments to make unbounded profit. The logical inductor does not know all logical truths at any finite stage. But its uncertainty is structured — it is the best a bounded reasoner can do, given finite computational resources, and it improves as more computation is invested.

The key insight for our purposes is this: logical determination does not entail epistemic accessibility. A fact can be logically entailed by the complete description of a system and still be invisible to a reasoner who knows only some consequences of that description. The logical inductor’s uncertainty is not a flaw to be apologized for — it is a constitutive feature of what it means to be a bounded reasoner embedded in a world whose logical structure exceeds one’s computational capacity.

This connects to the broader program of embedded agency — the study of what happens when an agent is inside the world it models, with finite resources, unable to step outside and survey the whole from a God’s-eye view. The embedded agent faces not just empirical uncertainty (which state is the world in?) but logical uncertainty (what follows from what I already know?), naturalized uncertainty (where am I in the world I’m modeling?), and the problem of reasoning about its own reasoning process while that process is running. These are not separate problems but facets of a single condition: the condition of being a bounded perspective within a reality that exceeds one’s grasp.

This is precisely the situation with conceivability and metaphysical possibility. The complete metaphysical ground of a situation logically determines which outcomes are possible. But a bounded reasoner — reasoning under abstractions that capture only some of the relevant structure — cannot see all the determinations. The conceivable-but-impossible outcomes are logical consequences of the complete description that the reasoner’s abstractions fail to reveal. Conceivability is, in effect, what logical possibility looks like to a logical inductor at a finite stage of computation — a wider space than the true space of possibility, because the inductor has not yet computed enough to rule out all the impossibilities.

A logically omniscient being would still distinguish conceivability from metaphysical possibility — conceivability might remain a broader category, encompassing scenarios not ruled out by concepts alone. But such a being would know exactly where conceivability outruns possibility, rather than being uncertain about it as we are. The gap between conceivability and possibility would be fully mapped, not invisible. For us, it is invisible — and this invisibility is not a flaw to be overcome but a constitutive feature of being a finite perspective embedded in the computational universe. The study of logical uncertainty formalizes the consequences of this boundedness, and one of those consequences is that conceivability is a wider net than metaphysical possibility.

The connection to the plenitude argument is direct. The structural realist who says “all logically consistent structures are real” is implicitly evaluating logical consistency relative to some set of axioms or constraints — some abstraction of what consciousness, suffering, and structure consist in. But consistency relative to an abstraction does not entail consistency with the complete metaphysical ground. The plenitude argument trades on a conflation between what is consistent under a functionalist or structuralist abstraction and what is consistent under the full specification of the conditions for genuine phenomenal experience. Garrabrant’s work shows that this conflation is a predictable consequence of bounded reasoning — not an insight into the structure of reality.

On our account, genuine phenomenal suffering requires specific architectural conditions: homeostatic acquaintance (a real causal relation of regulatory coupling between monitoring and monitored processes, sustained over the temporal window of the phenomenal present), indexical unity (self-directed representational structure with genuine integration), and embeddedness in a computational process that generates the relevant dynamics. A mathematical structure that describes suffering need not satisfy these conditions. The description is consistent with suffering under the abstraction; it may be inconsistent with suffering under the complete specification.

This does not mean that no structures involve genuine suffering. It means that the space of structures involving genuine suffering is narrower than the space of structures that are conceivable as involving suffering. The plenitude argument treats these as coextensive. They are not.

Contra Platonism: Why Grounding Matters

The arguments above establish that conceivability outruns metaphysical possibility. But there is a deeper diagnosis of why the plenitude argument fails, and it concerns the Platonic assumption that drives it.

Platonism comes in several varieties, and the plenitude argument draws on more than one. Mathematical Platonism holds that all consistent mathematical structures are equally real — Tegmark’s Mathematical Universe Hypothesis is the boldest contemporary statement. Phenomenological Platonism holds that experiences, qualia, and perspectives are abstracta that exist in virtue of their phenomenological character — that a pain is a pain regardless of what, if anything, instantiates it. Theological Platonism locates both mathematical and phenomenological forms in the divine intellect, as archetypes from which God selects what to actualize. Michał’s dialogue moved through all three: from the mathematical structures of hypergraph physics, to the claim that some structures contain lived perspectives within themselves, to the question of whether God selects esse from a pre-given totality of candidates.

What these Platonisms share is the assumption that abstracta — whether mathematical, phenomenological, or theological — have ontological standing independent of their grounding. A mathematical structure is real because it is consistent. An experience is real because it has the right phenomenological character. A form in the divine intellect is real because God conceives it. In each case, the abstractum exists first, and the question is whether it is also instantiated, actualized, or selected.

Our framework rejects this shared assumption. On our four-layer ontology, nothing is real in virtue of being abstractly describable — whether the description is mathematical, phenomenological, or theological. Reality requires grounding: constitutive grounding through the layers, from computation (Layer 0) through physical regularity (Layer 1) and functional organization (Layer 2) to minds and perspectives (Layer 3). What makes a higher-level description real rather than merely coherent is that it reflects a real pattern — in Dennett’s sense — anchored in the causal fabric, supporting counterfactuals, sustaining predictions under intervention. An ungrounded abstractum, however vivid or consistent, is just a description.

This is the anti-Platonic move that was already implicit in the sentence “You are not an abstract deliberator. You are a specific agent.” The plenitude argument treats suffering as an abstractum — a mathematical or phenomenological object that exists in virtue of its formal or qualitative character. But suffering is not an abstractum. It is a regulatory relation — homeostatic acquaintance — constitutively grounded in the functional organization of a specific system embedded in a specific environment. Tear away the grounding, and you have a description of suffering, not suffering itself. The description may be mathematically consistent, phenomenologically vivid, or theologically conceivable. It is not thereby real.

The Church-Turing boundary makes this concrete on the mathematical side. The generative primitive — computation — can produce only what is computationally generable. Not all consistent mathematical structures are generable. Potentiality (what can be generated) is a proper subset of mathematics (what can be described). The plenitude argument operates at the level of description. Our framework operates at the level of grounding.

Bottom Rung: Metaphysical Possibility (Binary, Objective)

Not all conceivable structures are metaphysically possible. The space of metaphysical possibility is determined by logical consistency with the complete ground of the situation — not with any particular abstraction of it. Free agency demonstrates this from the inside: the alternatives entertained in deliberation are conceivable but (given the agent’s full computational nature) not all genuinely possible. Useful-but-incomplete abstractions create the appearance of a wider possibility space than actually exists. The Church-Turing boundary provides a formal correlate: only what is computationally generable from the generative primitive (Layer 0) counts as genuinely possible within the computational universe. Not all consistent mathematical structures are generable. Potentiality is a proper subset of mathematics.

Middle Rung: Measure (Graded, Objective)

Among metaphysically possible observer-moments, not all carry equal weight. The CWSIA framework from “Why Now? Observer-Coherence” assigns measure to observer-moments based on the structural cost of maintaining coherent observation. Observer-moments where coherence is cheap (well-matched to environmental complexity) carry higher measure than those where coherence is expensive (mismatched, requiring prohibitive computational resources). This distributes realness unevenly across the space of possible perspectives. Some observer-moments are, in a precise structural sense, more real than others — not as a matter of subjective importance but as a consequence of the physics of observer-coherence.

Top Rung: Actuality (Binary, Subjective)

Actuality is constituted by being a perspective — a computationally bounded observer that forces the multiway branching of the Ruliad into a definite experienced history. This is the indexical unity we analyzed in the previous article. It is binary: either a system has the self-directed representational structure and genuine integration required to constitute a perspective, or it does not. And it is subjective in the sense that actuality is what computational structure looks like from inside a perspective — it is not a property the computational primitive has independently. Mathematics encompasses all structure; computation encompasses all generable structure; actuality is what computation becomes for an observer within it.

How the Rungs Answer Plenitude

Consider a conceivable suffering structure — an arbitrary mathematical configuration that, under some abstraction, encodes a being in agony. Each rung addresses the worry from a different angle.

Bottom rung: Is this structure even metaphysically possible? Does it satisfy the full conditions for genuine phenomenal suffering — homeostatic acquaintance, regulatory coupling, the right kind of computational architecture — or only the conditions according to some incomplete abstraction? If the latter, it fails before the question of realness arises at all. The mathematical or phenomenological description of suffering is not sufficient for metaphysical possibility of suffering. This constrains the space from below: not everything conceivable enters the domain of the possible.

Top rung: Should we concern ourselves with it? From the inside — from the perspective of an actual observer embedded in a causal neighborhood — our philosophical concern is legitimately bounded by what is causally connected to us. A suffering structure in a causally disconnected region of the Ruliad is not something we can affect, observe, or interact with. Actuality, constituted by indexical unity, gives us standing to limit our concern to our own causal component of reality. Broadening that concern to causally disconnected regions requires evidence — evidence of rulial interference, of causal pathways we hadn’t recognized. Just as quantum interference between branches is rare and requires specific conditions, rulial interference between regions of different effective physics is rarer still. Without evidence of such connection, the suffering structures in distant regions of the Ruliad, even if metaphysically possible, are outside the scope of our legitimate concern — not because we declare them unreal by fiat, but because the structure of actuality itself limits what we can meaningfully engage with.

Middle rung: What about the broader space of potentialities? If we do step back from our particular causal neighborhood — if we ask anthropic questions about the distribution of observer-moments across the computational universe — the CWSIA measure gives structure to this broadened view. Not all possible observer-moments carry equal weight. The measure is determined by the structural cost of maintaining coherent observation: observer-moments where coherence is cheap carry higher measure than those where coherence is prohibitively expensive. A suffering structure embedded in a wildly complex environment where maintaining coherent observation requires astronomical computational resources would carry negligible measure — it would be a possible actuality, but a vanishingly lightweight one in the distribution over potentialities.

The plenitude argument fails because it treats conceivability, metaphysical possibility, actuality, and measure as a single undifferentiated category. They are not. The bottom rung constrains what is possible. The top rung bounds what we can legitimately concern ourselves with from the inside. The middle rung structures the broader space when we reason from the outside. Together they dissolve the abyss — not by denying that suffering exists, but by showing that the space of genuine suffering is orders of magnitude smaller and more structured than the plenitude argument assumes.

Evolutionary Bounding

Suffering, on our four-layer ontology, is a Layer 2 phenomenon before it is a Layer 3 one. At Layer 2 — the domain of teleological and functional organization, constitutively grounded in physics — suffering is a regulatory signal. Pain signals tissue damage and triggers withdrawal. Fear signals threat and triggers flight or fight. Grief signals loss of a social bond essential for cooperative survival and triggers reorientation of attachment. Hunger signals metabolic deficit and triggers foraging. These are functional states, as real as any physical regularity, that serve the organism’s continued viability by modulating its behavior in response to conditions that threaten it.

The key point is that functional suffering is inherently bounded and instrumental. Suffering that does not modulate behavior serves no function — it is regulatory noise. Suffering that does not respond to changing conditions (that persists after the threat has passed, or that fires in the absence of any threat) is biologically broken — it is a regulatory system that has lost its coupling with what it regulates. An organism that suffered maximally and perpetually would be an organism whose regulatory signals had become decoupled from the conditions they evolved to track. Such organisms are selected against, not because evolution is benevolent, but because regulatory decoupling is a fitness catastrophe.

Consider the concrete case of chronic pain. When acute pain becomes chronic — when the pain signal persists long after tissue healing, decoupled from any ongoing damage — the result is not merely unpleasant but functionally devastating. Chronic pain degrades motor function, cognitive capacity, social engagement, reproductive success. It is one of the most costly conditions in medicine precisely because it is a regulatory signal that has lost its regulatory function. Evolution does not design for chronic pain. It designs for acute pain — bounded, responsive, coupled — and chronic pain arises as a pathological breakdown of the regulatory mechanism. The very concept of “pathological” pain presupposes a norm of functional pain against which the pathology is measured.

This means that the “horrendous suffering” structures that drive the plenitude worry — suffering divorced from any regulatory function, suffering that serves no purpose within the system’s economy, suffering unlimited in intensity and duration — are precisely the structures that evolutionary dynamics select out. They are conceivable but evolutionarily unstable.

And here the argument reconnects with Section I. If we take the Layer 2 → Layer 3 constitutive grounding seriously — if phenomenal suffering is constituted by homeostatic acquaintance with regulatory vehicles whose functional role is to signal threats to viability — then phenomenal suffering that is entirely divorced from functional organization may fail to constitute genuine phenomenal suffering at all. The abstraction under which such a structure counts as “suffering” (perhaps a functionalist abstraction that identifies suffering with any information-processing pattern isomorphic to pain processing) may empirically diverge from the metaphysical ground of genuine phenomenal suffering (which requires not just the information-processing pattern but the regulatory coupling, the homeostatic maintenance, the acquaintance relation). The structure would be conceivable as suffering, under the abstraction; but not metaphysically possible as suffering, under the complete ground.

Game-Theoretic Bounding

Suffering-maximizing configurations are unstable in multi-agent dynamics. This follows from elementary results in evolutionary game theory: strategies that destroy or incapacitate the agents involved remove themselves from the population of interacting strategies. Configurations that sustain themselves over time require minimum thresholds of agent viability.

More precisely: in any iterated interaction, strategies that impose unbounded costs on participants (including themselves) are outcompeted by strategies that maintain the conditions for continued interaction. Cooperation-sustaining equilibria dominate suffering-maximizing ones, not because the dynamics are designed to be kind, but because suffering-maximizing configurations are self-eliminating.

The acausal trade arguments developed by Wei Dai, explored by Scott Alexander in “The Hour I First Believed,” and examined in our earlier article push this further. Even across the multiverse — across the space of all possible agents — the structure of rational agency creates convergence pressure toward configurations that respect the interests of agents. Suffering-heavy configurations, precisely because they destroy or degrade the agency required for participation in trade, are underrepresented in the equilibrium distribution over possible agents.

This is not a claim that multi-agent dynamics eliminate suffering. It is a claim that they bound it — that the dynamics impose a floor below which suffering cannot sustainably go, because configurations below the floor are unstable.

Normative Bounding

The third constraint is the most speculative but also the most far-reaching. If the normative order is real — if normative facts are the mathematics of indexicality, constitutively grounded in the computational fabric that produces minds — then the space of metaphysically possible configurations is not normatively neutral.

We argued in “Deep Atheism, Existential Optimism” that Is and Ought are not orthogonal. The universe is the kind of thing that produces perspectives, and the existence of perspectives generates normative structure — truths about the structural situation of beings with perspectives embedded in a world. This normative structure is not projected onto reality by contingent human preferences. It is discovered, and it is discoverable by any sufficiently structured perspective, because normative facts are facts about what it is to be a perspective.

If this is correct, then agents with constitutive reason-responsiveness — agents who are genuinely responsive to reasons rather than merely optimizing around them — will tend to recognize and respond to suffering. Not perfectly, not immediately, but with a structural tendency that provides a ratchet against the worst configurations.

Consider a concrete instance. The abolition of chattel slavery in the Atlantic world was not inevitable — it required specific historical contingencies, political coalitions, economic shifts. But the direction of the moral argument was not arbitrary. Slaveholders could not consistently maintain, under reflection, that beings with perspectives, self-models, diachronic commitments, and the capacity for suffering had no moral standing. The reflective process had traction because the slaveholders were themselves reason-responsive agents embedded in a normative order they could not fully ignore. Some resisted for generations. But the moral argument kept gaining ground, because the recognition of shared perspectival structure is not an arbitrary preference — it is a discovery about what it is to be a being among beings. The ratchet worked slowly, unevenly, against enormous resistance. But it worked in a direction that was not accidental.

The broader history of moral progress — from child sacrifice to children’s rights, from the subjugation of women to their legal equality, from the indifference to animal suffering to the emergence of animal welfare as a moral concern — is evidence that this ratchet is real, even if it operates on timescales measured in centuries rather than decades. The pattern is not convergence to a final moral truth but progressive expansion of the circle of moral concern, driven by the recognition that more and more beings satisfy the structural conditions for perspectival standing. Each expansion met fierce resistance. Each eventually prevailed, because the reasons for it were stronger than the reasons against.

The floor here is: suffering in systems containing genuine persons is bounded by the self-corrective capacity of rational agency. Reason-responsive agents are the kind of beings that moral progress can work on, and moral progress works in the direction of recognizing and reducing suffering, because suffering is a fact about perspectives that reason-responsive agents can discover and respond to.

The Three Bounds Together

None of these three constraints claims an omega point. None says suffering diminishes to nothing. They say it does not go to infinity.

Evolutionary dynamics bound functional suffering to its regulatory role. Game-theoretic dynamics bound interactive suffering to levels compatible with sustained agency. Normative dynamics bound suffering in systems containing reason-responsive agents to levels that moral progress has not yet corrected but can in principle address.

Together with the three-rung framework — which establishes that the space of suffering we should concern ourselves with is far more structured than the plenitude argument assumes — these bounds complete the response to the plenitude worry. Reality contains suffering. We do not deny this, and any philosophy that did would be dishonest. But reality is not the abyss. The space of possible suffering is constrained by the gap between conceivability and metaphysical possibility; our legitimate concern is bounded by actuality and causal connection; the broader distribution of potentialities is weighted by observer-coherence costs; and within whatever space remains, the convergent dynamics of evolution, game theory, and normativity impose floors.

The “Spark of Esse” and the Subset That Cannot Be Delimited

Michał searched throughout his dialogue for the “spark of esse” — the act of being that makes mathematical structure actual rather than merely possible. He tried to locate it as a subset of the Platonic realm, as a divine creative act, as physics-as-traversal. His AI interlocutor consistently pushed back: esse is not a thing within the system but the condition under which there is any system at all.

The most interesting moment came when Michał proposed that esse is “not a subspace but a subset — we have no idea how to delimit it.” This concession — that actuality is a real distinction within mathematical structure whose boundary we cannot draw — is closer to our framework than it might appear. We agree that there is a real distinction. We agree that it cannot be captured by a simple criterion within the mathematical formalism. Where we disagree is about the nature of the distinction. Michał’s interlocutor kept pressing: is the subset a structural distinction (some structures have a property others lack) or a pre-structural distinction (something like esse that is prior to all structural characterization)?

Our answer cuts between these options. Actuality is constituted by perspective — by being a computationally bounded observer within the Ruliad. It is not a structural property that some mathematical objects have and others lack (the interlocutor was right to resist this). But neither is it a mysterious pre-structural “act” that descends upon structures from outside (this is where the Thomistic framework, however profound, runs into difficulties it cannot resolve with its own resources). Actuality is what computational structure looks like from inside a perspective. It is relational — it depends on the existence of an observer with the right architecture — and it is constituted by that architecture, not by something beyond it.

The “spark” Michał was looking for is indexical unity — the architectural achievement of being a genuine perspective. It is not sprinkled onto structures from outside. It is constituted by structures with the right self-directed, integrated, regulatory architecture. And the “subset we don’t know how to delimit” turns out to be the set of structures that achieve indexical unity — a set whose boundary is determined by architectural conditions (self-directed representational structure, genuine integration, and for phenomenal consciousness specifically, sustained homeostatic regulation) that are in principle investigable, even if we cannot yet draw the boundary precisely.

Michał’s frustration with Thomistic rigidity — his sharp observation that “Thomism is not a collection of thoughts to be variously expressed as new languages allow but a curated grammar of sanctified expressions to hold onto” — deserves a sympathetic hearing. He was right that the Thomistic apparatus resists translation into modern conceptual frameworks. But the resistance is not merely linguistic. The Thomistic distinction between essence and existence, between what a thing is and that it is, captures something real — something our framework preserves. What it cannot capture, within its own grammar, is how the distinction is constituted. “God creates by giving esse” is a placeholder for a mechanism. Our framework fills the placeholder: actuality is constituted by perspectival architecture, and the distinction between essence (structure) and existence (actuality) is the distinction between mathematical structure as such and computational structure as encountered from within by a genuine observer.

Michał’s interlocutor was also right that esse is “closer to a verb than to a noun, closer to illumination than to a region illuminated.” On our account, actuality is not a property some structures have and others lack. It is a relation — what the computational universe is for a perspective embedded within it. The verb-like character that the Thomistic tradition insists on is preserved: actuality is something that happens when computational structure achieves the self-directed organization that constitutes a perspective. But it happens naturally, architecturally, without requiring a divine agent to make it happen.

Thomism as Living Philosophy

Our framework attempts exactly the transport Michał was seeking. The Thomistic distinction between essence and existence — between what a thing is and that it is — maps onto our distinction between mathematical structure (which specifies what a thing is) and actuality (which is constituted by being encountered from a perspective). The Thomistic insistence that esse is not a property among others but the condition of there being anything at all maps onto our claim that actuality is not a structural feature but an indexical relation. The Thomistic claim that not everything possible is actual maps onto our three-angle framework for realness.

What does not survive transport is the theological apparatus — the divine creative act, ipsum esse subsistens, the participation metaphysics. Our framework replaces these with architectural conditions: indexical unity, computational boundedness, regulatory coupling. Whether this counts as “Thomism translated” or “Thomism abandoned in favor of something better” is a question we leave to those with deeper stakes in the tradition. What we claim is that the core insight — that actuality is not reducible to structure, that not everything possible is real, that reality is constrained in ways that matter morally — survives, and survives in a form that is precise, testable, and compatible with what we know about computation and consciousness.

“Any Creator Is Less Cruel Than No Creator”

Michał’s most striking corollary was that any selector of esse — any principle that restricts what is actual — is less cruel than unrestricted plenitude, because plenitude entails maximal suffering while any restriction reduces total suffering.

Our framework preserves the force of this insight while relocating its ground. The “selector” is not a divine agent choosing which structures to actualize. It is the structure of realness itself — metaphysical possibility, actuality, and measure each operating by their own logic. Together they do the work that Michał’s theological intuition attributed to a creator: they limit the abyss.

And the bounding-from-below argument provides what Michał’s interlocutor identified as the crucial missing piece: a principled account of why the restriction is not arbitrary. The dynamics of evolution, game theory, and normativity do not randomly exclude suffering structures. They systematically disfavor configurations in which suffering exceeds its functional role, destabilizes interaction, or persists uncorrected in systems capable of recognizing it. The “selection” has a direction — not toward an omega point, but away from the unbounded catastrophe that plenitude would entail.

Michał’s corollary remains true, transposed: any structured account of realness that distinguishes the actual from the merely conceivable is “less cruel” than undifferentiated plenitude. Our framework provides the structure. It is not a creator. But it does what Michał needed a creator to do.

The Shift That Has Not Yet Been Announced

Michał asked whether anyone had announced the philosophical shift implied by computational approaches to consciousness — the shift that places experience itself within the Platonic realm, forcing a choice between plenitude and selection. His interlocutor identified the shift as real but unannounced, hovering at the intersection of Tegmark’s mathematical universe, Chalmers’ computational theory of consciousness, and Wolfram’s generative machinery.

Our framework announces it — and resolves it. The shift is real: if consciousness is structural, then the question of which structures are conscious is a question about the structure of reality, not a question about which structures a benevolent deity chose to animate. But the resolution is not the stark binary of plenitude versus mystery. It is the three-rung ladder: metaphysical possibility constrains what can exist; measure distributes weight among what can exist; actuality requires what exists to be encountered from a perspective. Experience belongs to the computational universe — but not in the undifferentiated way that plenitude assumes. It belongs to it architecturally, and the architecture is selective.