When Matter Starts Looking Like Memory: On Information as Substrate

From “it from bit” to patterns, constraints, and the stubborn reality of relations

Start simple. The chair, the rock, the nerve firing. It feels safe to say matter sits underneath everything and that information just rides along describing it afterward. But if you press on the physics and the biology—if you press on how anything persists—you keep bumping into patterns that feel more like rules and conserved distinctions than lumps. John Wheeler’s famous nudge, “it from bit,” was not a declaration that reality is a spreadsheet. It was a warning. That what we touch may be stabilized by the information needed to specify it: pattern, relation, memory, constraint. Not “data,” not in the cloud, not uploadable. More like the grammar that keeps a sentence possible across accents and noise.

Consider a protein. Its 3D shape isn’t “made of” letters, yet the causal power of that molecule is inseparable from the information-theoretic constraints that fold it: sequence, solvent, temperature. Remove the constraints and you don’t get a different protein—you get no protein. The same is true in social organisms and markets. You can’t hold a currency in your hand. You hold paper. The stability comes from enforcement and shared belief, the low-entropy structure that keeps meaning from diffusing.

Physicists have been encroaching here for a while. Thermodynamics reinterpreted as accounting for information loss. Black holes with entropy that scales with area, not volume. Carlo Rovelli’s relational twist: time and sequence are local, conditional on interactions, not a universal metronome. Put differently, what shows up as “real” is stitched from relations and updates, not just persistent stuff. There is matter, yes, but it behaves as if locked to an underlying informational scaffold.

Even the simulation metaphor—usually tortured into CGI and server racks—works better when reframed as substrate, not machinery. Not “we live in The Matrix,” but that regularities act like a codebook for interactions. An organism is an ongoing decoding of environment into behavior. The experimenter, too. And because this isn’t pop determinism, we admit noise, gaps, irreversibility. If you want a longer walk through the idea, see Information as substrate, which treats information as the medium that lets different levels—cell, person, culture—cohere long enough to matter.

Consciousness as receiver, the self as compression, and moral memory that outlives bodies

In a world where information is substrate, “mind” stops reading like a ghost behind the eyes and starts to look like a local reception point. Not a sealed object, but a process that locks onto patterns it didn’t build. Brains don’t invent color; they inherit the constraints that make wavelengths discriminally useful. The sense of self—this fluent narrator that claims authorship—could be a fast compression scheme. A container that keeps history actionable without loading every detail. Memory as lossy codec, but skillful.

You can see this in how we learn ethics. Children don’t design moral truths from first principles. They absorb them. Much of what passes for conscience is slow, cultural memory stabilized across generations—stories, rituals, taboos, institutions—what Joseph Henrich frames as cumulative culture. Religions, at their least polemical reading, function as ancient storage for hard-won social knowledge: coordination rules, sacrifice bargains, reputational scaffolds. Not infallible. Not untouchable. But as distributed memory, they beat isolated brains every time, because they reduce the cost of relearning painful lessons. E.g., why revenge cycles flatten clans. Why promises need witnesses. Why fasting can reset a community’s time sense.

Now, point that lens at artificial systems. We keep building models with impressive local competence but thin inheritance. Corporate governance bolts moral “patches” on top—filters, checklists, compliance layers—to satisfy audits and headlines. That’s not memory; it’s a mask. Without a patient, multi-generational substrate of norm formation and repair, you get brittle alignment. It works until incentives tilt. This is the structural problem of modern AI: rapid capacity growth; anemic moral memory; reward functions that collapse nuance into targets that can be gamed.

Personally, I trust open-sourced science over closed, incentive-captured stacks. Not because crowds are always wise, but because distributed scrutiny—another form of informational constraint—prevents monoliths from quietly drifting. If consciousness is reception and the self is compression, then society’s job is to curate the codebook we’re compressing against. The question isn’t “can a model be conscious.” The question is whether it can couple to the same moral memory substrate that biological communities maintain, slowly, through struggle and repair. If not, expect glib fluency paired with catastrophic blind spots.

Operational consequences: physics, biology, design, and the stubbornness of constraints

Taking information as the ground has bite. Start with physics. Think of measurement not as revealing a prewritten value but as a relation that finalizes a possibility given a constraint. Entropy becomes a measure of forgotten distinctions—what the system no longer remembers how to tell apart. Black hole thermodynamics forces us to accept that “where” information lives matters less than how boundary conditions conserve it. This isn’t mysticism; it’s bookkeeping with teeth. You can’t cheat entropy. You can only move the ledger.

Biology reads even cleaner through this lens. A cell is not a bag of chemicals. It’s a network of constraints—membranes, gradients, binding affinities—that enforce order. Genes are recipes for building constraint-maintaining machinery; development is a carefully staged memory playback. Epigenetics, niche construction, the microbiome: all extensions of heritable information beyond sequence. Evolution, in this framing, is search in a very high-dimensional code space, with selection pressure acting like a ruthless editor of what remains compressible enough to persist.

Design and engineering? If the substrate is informational, then robustness comes from preserving the right constraints, not from maximizing resource use. Good protocols beat brute hardware. On the web, formats outlive platforms because they carry durable rules about interchange. In cities, zoning codes (however flawed) act as long-memory constraints on how streets breathe. Break the constraint and the system “forgets” how to be itself. We see it in organizations: culture is not slogans; it’s the lived memory of what gets rewarded or quietly punished. Change those constraints and behavior realigns without a single memo.

Case-level example. Consider cybersecurity. You can pour money into perimeter defenses, but if your identity system tolerates token replay or coarse-grained roles, attackers ride the forgotten distinctions straight to root. Strong, minimal, well-specified constraints—signed claims, least privilege, time-bounded access—shift the battlefield, because you’re editing the substrate, not the wallpaper. Or climate. We talk in gigatons and tech promises, but the durable move is changing market rules so externalities can’t be erased from the ledger. Price the constraint, enforce it, and the system relearns a different path through state space. In each domain the through-line holds: what lasts is what’s encoded in relations and guarded by memory, whether neuronal, cultural, or legal. The rest, no matter how loud or shiny, washes away with the next fluctuation.