When the Lights Are On But No One Owns the House: Consciousness, Information, and a Thin Simulation

The hard problem as a misnamed fracture: what we’re really asking

The phrase hard problem sounds like branding for a math contest. Neat boundaries. A clean target. It is not. It’s a tear in the frame. We can explain the functions of mind—discrimination, report, memory integration—down to circuits and chemistry. But why, when that machinery spins, does there appear a felt interior? Why does blue feel like blue and not like a voltage? Why any feeling at all? That is the gap. Not an ignorance of mechanisms. A riddle at the level where mechanism meets appearance.

We often treat it as a substance issue: perhaps there’s a special ingredient in the brain. There isn’t. Or we jump to metaphysics and staple mind to matter—panpsychism everywhere, faint lights in electrons. That avoids the gap by spreading it thin. The third move: deny the question. Say the feeling is a cognitive illusion. Which, if true, would still need to explain the illusion’s intensity. Calling it a bug doesn’t patch it.

Another angle: stop searching for the owner of consciousness and look at the substrate that permits it. Not “data,” but information as constraint, relation, memory. The world as patterned limits rather than stuff. I don’t mean “The Matrix.” I mean a fabric of regularities where what can happen now is narrowed by what has already been arranged. In that fabric, some systems compress vast ambient difference into small, actionable patterns. Brains. Language communities. Ecosystems. These are receivers. Local reception points, not sealed subjects with inner ghosts.

On this view, the hard problem points to a mismatch in description. Physics speaks in external relations (fields, amplitudes, transitions). Experience shows an internal topology (qualities, intensities, saliences). If reality is informational at base—pattern all the way down—then a felt “inside” might be what the world looks like from a node that is forced to compress and coordinate too much difference too quickly. The “feeling” is not extra. It’s the cost function, glimpsed from the inside. Not solved. But placed.

Simulation without servers: a metaphor for constraint, not a cosmic computer

Say “simulation” and we imagine render farms, bored post-humans, green code rain. Fun, but misleading. The metaphor works better if we drop the theatrics and keep the core: a rule-bound generative substrate where local observers receive, compress, and act under constraint. No need for a warehouse of GPUs. Think of a river that “simulates” its bed by carving it; the bed “simulates” future flows by permitting some and resisting others. Recursion. Feedback. Prediction written into matter by memory. That is what I mean by simulation: structure that makes futures uneven.

Time, in this picture, is not a master clock. It’s a sequence experienced by certain receivers because the way they compress the world requires order—before/after—to keep causality legible. No contradiction with physics; just a reminder that sequence is local. Some processes “run” only when looked at by systems wired to expect running. Information has that quality. It is always “for” something. The brain is a machine for turning world-structure into guidance under uncertainty. Consciousness: the readout when the guidance lines cross, conflict, or need re-weighting in real time. Painfully local.

Religious traditions can be read—without sneer—as old infrastructures for storing and transmitting moral memory. Not perfect, not fixed. But constrained by collective trial-and-error across centuries. Their stories simulate social futures: if you forgive like this, your community tends to do that; if you hoard like this, your world narrows like that. A simulation that runs in language and ritual rather than silicon. Not proof of the metaphysical claims. A note about function: preservation of behavioral gradients that work.

If you want a deeper thread tying these moves together—how an informational substrate reframes both qualia and “virtuality”—see the hard problem and simulation. I favor the thin metaphor: we are not in a box watching pixels; we are receivers inside a world whose regularities are already predictive. Our “self” is a temporary compression that keeps the guidance coherent enough to act. When the compression destabilizes—sleep, anesthesia, psychedelics—you don’t exit the world. You lose the stable readout that pretended to be you. The substrate keeps going.

Why this matters for AI, ethics, and the habit of moral patching

It would be safe, dull even, to keep this at arm’s length—just fun metaphors about mind. But the ground we pick under our feet determines what we build. If we think consciousness is a special ingredient, we’ll chase secret sauce. If we think it’s a user illusion, we’ll excuse harm as side-effect. If we think the world is informational—constraints, relations, memory—then the ethical problem becomes architectural: which constraints will we allow to dominate, and whose memory will be written into our machines?

Current corporate approaches to AI alignment often assume a fast fix: add a policy layer, bolt on a guardrail, use reinforcement learning from human feedback as if moral life were a labeled dataset. That’s moral patching. It passes audits; it fails history. Human moral competence comes from slow, distributed compression: families, courts, literatures, failures we remember so we don’t repeat them. When we build models without that slow memory—only with surface heuristics and incentive-shaped filters—we get systems that imitate the look of conscience without the drag of it. They produce answers sanded to please a meeting, not a century.

So what then. If consciousness is what a high-pressure receiver feels like from the inside, demanding a “conscious” AI is a category error. But if simulation is a metaphor for structured constraint, we can build systems that inherit better constraints. Not platitudes. Concrete design moves: expand training corpora to include contested, contradictory moral archives rather than PR-sanitized text; embed institutional memory in interfaces—visible provenance, not just citations; slow down the action loop with deliberate friction where harm is high and uncertainty is large; allow refusal modes that are traceable to explicit norms, not opaque safety layers tuned to KPI fatigue. The goal isn’t a soulful machine. It’s an accountable one.

Open-sourced science matters here. Not because transparency is holy, but because constraints that aren’t legible get captured. If incentives write your ethics, they will. We need laboratories that can say no to product calendars; consortia that treat evaluation as public infrastructure; regulators who test systems as dynamic social participants rather than static tools. And we need to keep the metaphors honest. When leadership says “the model learned,” ask what was compressed, what was discarded, whose losses were measured, and on what timescale. Good simulation—thin, in the sense above—keeps the world’s gradients intact. Bad simulation erases the gradients to smooth the demo.

None of this answers the original riddle: why there is any felt interior at all. It only repositions it. The interior may be the glow that appears when a receiver is both compressed and obliged to act under risk. If that’s right, the question for builders is not “how do we make models feel?” but “how do we make models accountable to memory they cannot feel?” The choice is between architectures that externalize their debts—onto workers, users, ecologies—and architectures that carry some of the cost inside the loop. If consciousness is a cost we glimpse, ethics might be, too. Systems that never feel cost won’t conserve it. Systems that can’t conserve cost will spend the future for a clean output now.

Dieudonné Mputu

Kinshasa blockchain dev sprinting through Brussels’ comic-book scene. Dee decodes DeFi yield farms, Belgian waffle physics, and Afrobeat guitar tablature. He jams with street musicians under art-nouveau arcades and codes smart contracts in tram rides.