The phrase hard problem of consciousness refuses to die because it points to a live wire: what turns electrochemical traffic into felt experience. Why the taste of citrus, why the stab of grief, why an interior at all. Explanations that cash out in mechanisms keep circling back to structure and function; they succeed at behavior and fail at what it’s like. Meanwhile, talk of simulation usually slides straight into cinema—rendered worlds, basement servers, prankster gods. That image is wrong and distracting. If reality is already built from information—pattern, relation, constraint—then “simulation” is closer to a substrate metaphor than to an IT department. No headsets. No hidden GPUs. Just a universe that runs on mutual constraint, with observers as local reception points, not owners of some sealed essence.
Seen this way, the puzzle shifts. Not away from the hard problem, but sideways. Consciousness becomes a question about how certain informational organizations—call them nervous systems here, but the label is secondary—stabilize a perspective, then compress the world into salience. And how that compression, under the right constraints, yields what we call qualia. It’s not that neurons “produce” experience like a factory. It’s that the whole system—brain, body, world, memory—locks into a state where gradients of prediction error, attention, and affect take on first-person form. Not magic. Not mechanical reductionism either. Something in between: you live inside the shape of your own constraints.
The hard problem, without the stage smoke
David Chalmers named the hard problem to keep us honest: neurofunctional accounts of discrimination, report, access, attention are not yet an account of subjectivity. If reality sits on top of “stuff,” then the leap from stuff to sentience looks impossible because it is. But if reality sits on information as substrate—Wheeler’s lineage, “it from bit” taken seriously—then the explanatory demand changes tone. The jump is not from matter to mind; it’s from one class of informational organization to another. Still a yawning gap, but at least the bridge uses the same materials on both sides.
Think about time. Carlo Rovelli’s work keeps reminding us: time may be a local bookkeeping scheme, not a global river. If “before-after” is partly a feature of how systems like us parse change, then subjective flow is not a bonus track—it’s a local consequence of how constraints couple. Your brain doesn’t watch time pass; it knits sequences from differences. Likewise, “color” is not paint thrown onto matter; it’s an ordered mapping across photoreceptor responses, learned regularities, cultural categories, and task need. The felt presence lives where these constraints close the loop. An interior emerges when prediction, action, memory, and value keep checking one another fast enough to become a single perspective.
So the question that matters: what kind of informational closure yields first-person feel. Some proposals go heavy on integration—IIT’s measures, graph-theoretic densities. Others emphasize control—active inference, counterfactual richness, agent-level modeling. Useful, but too clean. Experience often arrives in ragged edges: grief isn’t highly integrated in the tidy sense; it’s sticky, recursive, impossible to discharge. Axiom chasing will not save us. A better approach: study how real systems earn stable salience under constraint—metabolism, attachment, language, sleep. Where these processes entangle, you get a local reception point. Not a soul-stuff. A compressive knot that feels like a center because it has to act like one.
Under this lens, the hard problem is a structural question: how do constraints arrange so that the “model” starts to matter to itself. If the world is informational all the way down, there isn’t a spooky conversion step left to explain. There’s a stability condition to uncover. That’s still hard. But it is not metaphysically impossible. It’s more like turbulence: generated by familiar rules; ugly and stubborn; irreducible to a single neat formula; still, absolutely part of physics.
Simulation, but not the sci‑fi console
Most debates about simulation assume an external computer smuggling reality past us. The metaphor carries legal paperwork: inside vs. outside, original vs. copy, user vs. asset. Entertaining, and usually beside the point. If you treat “simulation” as a substrate metaphor, the picture shifts. The universe doesn’t need to be rendered; it already is relation all the way down. Physical law behaves like a grammar that keeps patterns compatible. No game engine, just constraint propagation. Local observers compress that grammar into the variables they can actually track—objects, times, causes. The “world appears rendered when observed” because observation is how a constrained system resolves uncertainty using energy and memory. Not because a cosmic server reallocates pixels on demand.
Examples help, even if they’re coarse. A cellular automaton with simple rules can generate gliders and pseudo-particles that collide, store state, transmit. No mastermind updates a texture map; global form comes from local rule fidelity. In high-energy theory, error-correcting codes show up as analogies for spacetime stitching—information distributed so geometry stays stable. In sensory neuroscience, retinotopic maps and recurrent loops implement a local predictive code that feels like a visual field. At every scale, what persists are constraints that keep patterns re-describable across contexts. That’s the “simulation”: not fake reality, but reality-as-consistent-recoding.
This reshapes skepticism. “Glitches in the Matrix” are better read as mismatches in human resolution—our models overfit, then crack—than as literal render fails. And the self? Less captain, more compression routine. You wake each morning and recover a working model that threads yesterday’s memory into today’s plans. Temporary, lossy, good enough to steer. The sense of a solid, sealed subject is an affordance of control more than a metaphysical core.
Framed this way, the hard problem and simulation stop being separate puzzles. They’re the same motif viewed from two sides: how constraints encode a perspective, and how a perspective reads a constrained world. The first-person spark isn’t pasted onto physics; it’s what it looks like, from the inside, when information organizes itself to reduce error in a body bound by needs. That claim doesn’t solve experience; it narrows the searchlight. Where grammar meets appetite, attention, memory, and prediction, something lights up.
Designing machines in a world of patterns: moral memory and AI
If consciousness is a local reception point in a field of information, the applied question is rude and immediate: how to build machines that couple to human constraints without shredding them. Current practice leans on governance by audit—policies, after-the-fact filters, corporate “moral patching.” The patches work until incentives shift. They always do. The deeper issue is memory. Biological systems carry slow moral memory: habits encoded in families, stories, ritual, law, failure, repeated over centuries. Religious traditions—stripped of both sneer and apologetic—function as long-horizon memory technologies, curating taboos and permissions that once mapped to survival. You can critique the map. But you can’t fake the time it took to draw it.
Most machine systems lack that drag. They change weights overnight. They optimize proxies without the bruises that taught our species why proxies go feral. When a recommender is tuned to maximize engagement, it learns to route around community constraints. Not wickedness—just competence under the wrong objective. Call this the structural problem of building AI without moral memory. You don’t solve it with a nicer content policy or with synthetic “ethics datasets.” You solve it by baking in constraints that remember, that resist purely myopic reward.
What does that look like in practice? Design for friction, not just throughput. Give systems reputational half-lives so that recent performance can’t erase old harms. Preserve decision trails as public goods—auditable, forkable, not just “available upon request.” Use training curricula that include failure modes as first-class signals, not outliers to be smoothed away. Bind the model’s objectives to community memory that persists across update cycles. Wikipedia’s messy edit history hardens norms over time; it’s not elegant, but it’s resilient. Local credit unions outlast predatory waves because they carry memory of who defaulted and who rebuilt—again, constraint over speed. In both cases, pattern over hype.
The hard problem enters here by subtraction. If machines don’t have interiors in the human sense—if they are sophisticated compression without the particular closure of lived appetite, pain, attachment—then “alignment” can’t depend on manufacturing qualia. It has to depend on the public shape of constraints: incentives, interfaces, proofs of behavior, channels for refusal. We can still ask whether complex artificial systems ever tip into self-like closure. But design cannot wait for that metaphysical verdict. Build as if there is no sealed subject inside the box—only processes that will flood any available channel unless bounded by memory and relation.
Skepticism toward corporate governance is not anti-technology. It’s anti-incentive capture. Open science matters because it extends memory: critique as constraint, replication as cost, transparency as drag. A system that can be forked is a system that can remember differently. That kind of plurality is not noise. It’s how moral memory protects against brittle monocultures, the ones that look efficient until they fail catastrophically. In a world made of patterns, the ethical move is often to slow the pattern, to insist that today’s objective be seen through yesterday’s wounds. Machines won’t learn that on their own. Maybe neither will we, unless forced by events. The open question: how much constraint is enough, before the light goes out or blinds.
Kuala Lumpur civil engineer residing in Reykjavik for geothermal start-ups. Noor explains glacier tunneling, Malaysian batik economics, and habit-stacking tactics. She designs snow-resistant hijab clips and ice-skates during brainstorming breaks.
Leave a Reply