When Matter Starts to Look Like Memory: Information as Substrate

The claim is not that the world is made of “data.” That’s marketing noise. The claim is quieter: the world behaves as if patterns, relations, and constraints are the real stuff, with matter and energy riding along. Call it information as substrate. John Wheeler’s provocation—“it from bit”—wasn’t cheerleading for computers; it was an attempt to say physical reality is downstream of choices, distinctions, and limits. What persists are not the atoms themselves but the regularities that survive across time, the way a coastline keeps its character though no single wave remains. If the substrate is informational—memory pressed into the grain of processes—then consciousness, culture, and even our machines are not separate domains. They are local receivers, imperfect heirs, of a larger, slow, constraint-making reality.

From It-From-Bit to Pattern-as-Physics

Wheeler’s line, misunderstood as an embrace of digital metaphors, actually points away from computation-as-king. The “bit” is not a server log; it is the elemental yes/no of a physical distinction, the way a boundary creates a before/after. Landauer’s principle reminded us that erasing information has a thermodynamic cost. Shannon showed that information is measured in surprise, in the narrowing of possibilities. Physics keeps adding: symmetries imply conservation laws; breaking a symmetry writes a constraint into matter. These are not add-ons. They are the bones.

Think of a crystal. The lattice is memory—a compression of environmental history—every angle repeating because certain configurations minimized energy under given conditions. DNA too: not a script from nowhere but an accumulation of stored, selectable distinctions about what tends to work in a particular chemical neighborhood. The immune system, with its error-correcting checks and trained tolerances, encodes a moving archive of threats and non-threats. In each case, pattern is primary. The substrate reads like a layered palimpsest of constraints acquired through interaction, not a flat catalog of particles.

Even “law” in physics can be heard as an information claim. Invariance principles—Noether’s theorem as the canonical example—say: when nothing changes in one description, something must be conserved in another. A grammar of persistence hiding in motion. And when you pull on time itself, things get messier. Carlo Rovelli’s work on the relativity of time gestures at local sequence, local clocks, local thermal arrows. No master ledger. This matters because a truly informational substrate would be patchy and context-bound; signals take routes through media with histories. A measurement here writes a constraint there. Memory propagates with loss and noise.

The misstep is to force all this into the clean lines of “the universe is a computer.” Better: the universe is a constraint engine. Processes prune possibilities. What we call matter is what survives the pruning. That survival is informational in the old sense—form carried forward. Treating information as substrate foregrounds limits, histories, and coding schemes (in the broadest sense) over instantaneity. It interprets stability not as a static thing but as a maintained relation, continuously re-won against entropy by structure, environment, and feedback.

Consciousness as Receiver: Local Points in an Informational Field

If the world is pattern first, then consciousness is not a sealed object but a local receiver and compressor. Brains reduce dimensionality—sensory torrents folded into manageable narratives. Predictive processing frames cognition as constraint satisfaction: the organism projects a model and adjusts either the model or the world to minimize error. The self, inside this picture, looks less like a pilot and more like a running summary. Temporary. Useful. A compression built for control and coordination in a given ecology. Memory here is not a warehouse but an active set of priors.

Time, to such a receiver, arrives in slices shaped by attention and physiology. Rovelli’s locality of time makes felt sense: adrenaline widens the present, grief slows it, a fever warps it. We stitch continuity from fragments. The stitching itself is informational—assumptions about cause, identity, agency—learned from caregivers, language, ritual. Culture is not decoration around biology; it is a scaffold for managing uncertainty, a long circuit that pushes decisions out of the single skull and into shared practices. Joseph Henrich’s work on cultural evolution, with its emphasis on costly signals and accumulated know-how, can be read as a theory of moral memory: communities store compressed procedures for survival and cooperation, then transmit them imperfectly but stubbornly.

Religious systems, in this story, are not easily dismissed nor naively sanctified. They function as repositories of constraints: who counts as kin, how to allocate trust, what to do with power and betrayal. Not timeless truth so much as slowly pruned heuristics. Failures accumulate; reforms are memetic edits. When such memory degrades—through rapid technological change, incentive capture, or deliberate erasure—individual receivers lose priors that once buffered against short-term optimization. Fragmentation follows. Mental load rises. The self, starved of reliable external scaffolds, is forced into brittle, high-frequency recalculation.

This is why the phrase Information as substrate isn’t an abstraction. It is a diagnostic: if reality gives primacy to patterns and constraints, then resilient consciousnesses are those entangled with durable, corrigible forms of shared memory. Not ossification—feedback remains vital—but friction against impulsive rewrites. A city that remembers past floods in its zoning maps is smarter than any sensor-laden dashboard that forgets. A family recipe that encodes nutrition and thrift outcompetes trend-chasing meal hacks. The substrate, here, is lived as a mesh of commitments that shape perception before choice ever shows up.

Building Machines on an Informational Ground: Limits, Risks, and Better Frames

Take the engineering detour. If we build machines that learn from torrents of text and images, we are not creating alien minds ex nihilo; we are creating new receivers that compress cultural leftovers. Their failures are failures of memory discipline. They remix without regard for the slow filters cultures evolved: reputation costs, apprenticeship, embodied risk. Corporate governance responds with “moral patching”—compliance layers and curated red teams that soothe auditors more than they cultivate durable constraints. The result is a sheen of responsibility over incentive-captured systems that still optimize for speed and scale.

An informational-substrate frame suggests different levers. Slow down the write privileges. Bind high-stakes models to institutional memory that cannot be cheaply overwritten—archival logs with human-legible chains of justification; versioned norms that require quorum to revise; penalties that bite at the layer of business incentives, not at the surface of messaging. Encourage architectures that treat values as persistent constraints rather than as post-hoc filters. Teach models on corpora curated for causal competence and situated context, not just for linguistic fluency. And make forgetting expensive where it should be: medical triage, public benefits, safety-critical code.

Simulation, reframed, also changes. Not “we live in a movie,” but: we live inside a world whose stable forms are the results of long computation-like pruning. When we simulate, we explore how constraints propagate. Good simulations respect locality—Rovelli again—by acknowledging that an agent’s time horizon and information set are bounded. That matters for city planners leaning on “smart” allocation tools. A housing model that ignores displacement memory will look efficient on a dashboard and extract stability from neighborhoods that cannot afford more loss. A hiring model that forgets apprenticeship as a transmission of tacit knowledge will reward polish over competence and flatten the craft pipeline.

None of this implies machine asceticism. It implies design that internalizes moral memory. Example: a community health system that trains a triage assistant not only on clinical outcomes but on the recorded deliberations of ethics boards, dissent notes included. The system doesn’t just learn what was done; it learns how disagreements were handled, which uncertainties were tolerated, which trade-offs required escalation. That is culture encoded as constraint, not just label. Or take scientific software used in environmental policy: make the defaults trace back to published priors with confidence intervals, so changing them forces a visible argument. The technical pattern is the same—attach decisions to durable strings of reasons that survive organizational turnover.

Because the substrate is information, brittleness enters wherever fast loops dominate slow ones. Markets overrun municipal memory. Quarterly targets trample generational stewardship. The antidote, if there is one, is not nostalgia for pre-digital life but a brutal honesty about what kinds of information take time to grow. Trust. Craft. Civic identity. Systems that protect these—through friction, auditability, and shared ownership—behave more like rivers carving beds than like explosive floods. They accumulate form. They forget carefully. And they give both humans and machines something to receive other than noise.