Information Engine

Before meaning, there is difference. A system carries information when its state could have been otherwise — when there is uncertainty for an observer to resolve. Claude Shannon made this precise in 1948: information is measured not by what a message means but by how much it reduces uncertainty, quantified in bits. From the spin of a particle to the firing of a neuron to a letter on a page, information begins wherever one configuration is selected out of many possible ones.

A symbol is a compression: a small, shareable mark that stands for something larger. Speech encoded thought into sound; writing froze sound into durable form; the alphabet reduced thousands of pictures to a few dozen reusable letters; mathematics and logic compressed reasoning into manipulable notation; code turned instructions into executable text. Each leap traded raw detail for composability — fewer primitives, recombined without limit — letting a finite set of symbols express an unbounded space of meaning.

Life is the oldest information technology. Four nucleotide letters spell out, in three-letter words, the recipes for every protein in every cell — a code billions of years old, copied with extraordinary fidelity and proofread against errors. The genome is not a blueprint but a program: read, transcribed, translated, and executed by molecular machines. Before brains, before language, evolution had already discovered storage, copying, mutation, and selection — the core operations of any system that learns.

In 1936 Alan Turing stripped computation to its skeleton: a tape of symbols, a head that reads and writes, and a finite table of rules. Astonishingly, this minimal machine can compute anything any computer can — it defines the very boundary of the computable. Around this core grew the rest: algorithms that transform information into action, compression that squeezes out redundancy, and error-correcting codes that let signals survive a noisy world. Computation is how information stops being merely stored and starts doing work.

Communication is information defeating distance. The telegraph severed the link between message and messenger; the telephone, radio, and television widened the channel; the internet made every node addressable and every message routable. Each layer collapsed latency and multiplied bandwidth, knitting billions of people and devices into one continuously updating fabric. Humanity now behaves, in part, like a single distributed organism — sensing, signaling, and coordinating at the speed of light.

Money is memory: a ledger of who owes what, abstracted into a token that anyone will accept. For most of history that ledger needed a trusted center — a temple, a bank, a state. Cryptography changed the terms. Hash functions fingerprint data; digital signatures prove authorship; consensus protocols let mutually distrustful strangers agree on one shared history without a central authority. Zero-knowledge proofs go further still — letting one party prove a statement is true while revealing nothing else. Economic coordination, at its root, is information coordination.

A large model is a compression of its training data into a few hundred billion numbers — and to compress well, it must discover the regularities that generate the data. Meaning becomes geometry: words and concepts become points in a high-dimensional space where nearness is relatedness, and analogy becomes arithmetic. From this compressed world-model the system predicts what comes next, token by token. It is an old hypothesis made concrete: that to predict well is to understand, and that intelligence may be, at bottom, very good compression.

At the smallest scale, information stops behaving classically. A qubit is not a 0 or a 1 but a superposition of both, and entangled qubits share a single state across any distance. This is not a metaphor for computation — it is a different physics of information, one that may factor numbers and simulate molecules beyond any classical reach. Stranger still, physics suggests the information a region can hold scales with its surface, not its volume, and that black holes obey a thermodynamics of bits. Some physicists take the next step: perhaps reality is not described by information but made of it.

Genes are not the only replicators. Ideas — stories, songs, beliefs, techniques, jokes — copy themselves from mind to mind, mutating and competing for the scarce resource of human attention. Richard Dawkins called them memes, and the analogy runs deep: what survives in a culture is not necessarily what is true or good, but what is most transmissible. Religions, myths, ideologies, and viral posts are all information that has learned to spread. Networks accelerate the process; a culture is the slowly evolving memory of a species.

Run the trend forward. Memory grows cheaper and vaster; bandwidth approaches the physical limit; models compress more of the world each year; brains and machines edge toward direct interface. The boundary between individual and collective cognition is thinning. It is at least plausible that civilization is condensing into a single planetary-scale information system — one that senses, remembers, models, and decides as a whole. Whether that becomes a tool we wield or a mind in its own right is the open question of the century. The experiment is already running; we are inside it.

What we still do not know

The deeper one follows information, the more the easy answers dissolve. These are not rhetorical questions — they are live problems at the edge of physics, biology, computer science and philosophy.

The anatomy of information power

Every information system — a genome, a language, the internet, a model — can be read as a different weighting of seven capacities. Plot their profiles and the systems that reshaped history reveal why: each one pushed some of these terms far past what came before.