Information Engine
信息引擎
|
Psyverse · An atlas of information
EN · 中文 · bits → DNA → language → computation → networks → AI → qubits

Information Engine

信息引擎

Civilization is the history of information becoming denser, faster, more abstract, more networked — and increasingly intelligent. From the bit to the genome to the embedding to the qubit, one structure keeps recurring at ever-larger scale. This is an atlas of that structure.

Central thesis · 核心论点

Information may not merely describe reality. It may be one of the deepest substrates from which matter, life, intelligence and reality itself emerge.

10 systems · 十大系统live simulations · 实时模拟entropy · code · networks · qubits
BIT · ENTROPY · SHANNON · SYMBOL · DNA · TURING MACHINE · ALGORITHM · ERROR CORRECTION · NETWORK · CONSENSUS · ZERO-KNOWLEDGE · EMBEDDING · COMPRESSION · QUBIT · ENTANGLEMENT · MEME · COLLECTIVE INTELLIGENCE · PLANETARY COGNITION · BIT · ENTROPY · SHANNON · SYMBOL · DNA · TURING MACHINE · ALGORITHM · ERROR CORRECTION · NETWORK · CONSENSUS · ZERO-KNOWLEDGE · EMBEDDING · COMPRESSION · QUBIT · ENTANGLEMENT · MEME · COLLECTIVE INTELLIGENCE · PLANETARY COGNITION ·
01

The Origin of Information

Signal, entropy, and the birth of the bit

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.

Symbol Stream · Shannon Source

// biased distribution — partly predictable

Predictability0.32
noiseorder
Alphabet size N

max H = log₂(4) = 2.00 bits/sym

Entropy H
1.85
bits/sym
Max H
2.00
bits/sym
Redundancy
7
%
Surprise/sym
93
%
Compression · 4096-symbol message
Raw (fixed-width)8,192 bits
Entropy limit7,580 bits

≈ 93% of size · saves 7% (Shannon's source-coding theorem)

Symbol probabilities p(x)
46
0
18
1
18
2
18
3
Signal

The structured part — what a receiver can predict.

Noise

The unpredictable part — pure randomness, no pattern.

Entropy

Average surprise per symbol, in bits. Measures uncertainty.

Pattern

Repetition lowers entropy — and lets us compress.

Meaning

Not measured by Shannon. Entropy is surprise, not sense.

Entropy measures surprise, not meaning. — Shannon, 1948

02

Language & Symbol

Compressing the world into shared marks

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.

Evolution of Symbol Systems
Gesture & SpeechPictographsCuneiformLogographs 汉字SyllabariesPhoenicianGreek AlphabetLatin AlphabetHindu–Arabic NumeralsMathematical NotationFormal LogicBinary / CodeEmojiAI Tokens / Embeddings

← earlier click any node later →

Logographs 汉字

~1200 BCE
What it encodes

Chinese characters: each glyph carries a whole morpheme of meaning.

Key property

Thousands of logographs, dense per-symbol meaning, sound-independent.

Example glyphs
信息 文字
50,000
symbols
96%
info/sym
03

DNA & Biological Information

Life as self-replicating, error-correcting code

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.

Life is self-replicating, error-correcting information. Four chemical symbols — A, T, G, C — encode every organism, copied with a fidelity no human storage medium has ever matched.

Double Helix
A Adenine
T Thymine
G Guanine
C Cytosine

Pairing rule: A bonds T, G bonds C. Each rung is one complementary base pair — the redundancy that lets a single strand rebuild its partner.

Genetic Code · Codon → Amino Acid
Pos 1
Pos 2
Pos 3
AUG
Methionine · Start
single-letter · M
Redundancy1× codons

This codon is the lone spelling for its outcome. 64 codons map to 20 amino acids + Stop — so the code is degenerate, and many single-base typos still translate correctly. That is built-in error tolerance.

2 bit
per base
6 bit
per codon
1 / 20
addresses an AA
DNA template

The gene is stored as the double-stranded sequence 5′-TAC-3′. Each base is one of 4 symbols — exactly 2 bits.

3.2 B
base pairs
the human genome
~750 MB
raw 2-bit code
fits on one disc
~20,000
protein-coding genes
in only ~1.5% of the DNA
1 / 10⁹
error rate
after proofreading + repair

A 2-bit alphabet, 6-bit words, a 64→20 redundant code, and a copy machine accurate to one part in a billion — biology solved information storage and error correction billions of years before Shannon named them.

04

Computation & Algorithms

The minimal definition of what can be known

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.

A · TURING MACHINE

Symbols, rules, memory

Walk to the least-significant bit, then carry leftward: 1→0 and continue, 0→1 and halt.

HEAD
_
_
_
_
_
_
_
1
0
1
1
_
_
_
_
-6
-5
-4
-3
-2
-1
0
1
2
3
4
5
6
7
8
TRANSITION TABLE δ
STATEREADWRITEMOVENEXT
seek00Rseek
seek11Rseek
seek__Lcarry
carry10Lcarry
carry01Sdone
carry_1Sdone
STATE
seek end
STEPS
0
SPEED5×

A Turing machine is the minimal definition of “what is computable” — symbols on a tape, a finite table of rules, and a movable head. Memory + rules + symbols = universal computation. Every algorithm you have ever run is some elaboration of this 1936 idea.

B · ERROR CORRECTION

Hamming(7,4) · surviving noise

CORRECTED0
4 DATA BITS — CLICK TO SET
d1 d2 d3 d4
RECEIVED 7-BIT CODEWORD — CLICK ANY BIT TO INJECT NOISE
SYNDROME000(= 0)
✓ NO ERROR — codeword valid
p1 = d1 ⊕ d2 ⊕ d4
p2 = d1 ⊕ d3 ⊕ d4
p3 = d2 ⊕ d3 ⊕ d4

Error-correcting codes are why information survives transmission and storage. Three parity bits over four data bits build a structure where any single flipped bit announces its own position in the syndrome — so the receiver repairs it without asking for a resend. The same redundancy guards deep-space probes, QR codes, RAM, and DNA.

05

Networks & the Planetary Nervous System

From the telegraph to a single low-latency planet

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.

PLANETARY PACKET FLOWLIVE · 1991
COMMUNICATION EVOLUTION
REACH
Documents linked globally
LATENCY
~100 ms
BANDWIDTH
~10 Mbit/s broadband
BANDWIDTH · LOG SCALE107 bit/s
10¹10⁶10¹²

Each layer collapsed distance and raised bandwidth — telegraph to fiber spans eleven orders of magnitude — knitting humanity into a single, low-latency, planet-spanning nervous system.

06

Money, Cryptography & Consensus

Coordinating strangers without a center

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.

LONGEST-CHAIN CONSENSUSHEIGHT 0
BLOCK #0
521980c3
prev
00000000
canonicalcompeting forkorphaned
AVALANCHE EFFECTILLUSTRATIVE · NOT CRYPTOGRAPHIC
DIGEST
a4496db5

A one-character edit changes nearly every output digit — yet the same input always yields the same digest. That determinism + sensitivity is what lets a ledger detect any tampering.

Money = Memory

Money is a society's shared memory of who owes whom — a portable, fungible record of past contribution and trust.

Ledger = Shared State

Accounting is information bookkeeping; a ledger is shared memory. The hard problem is agreeing on one copy.

Cryptography = Sealed Truth

Hashes and signatures let anyone verify a record without trusting its author — math replaces the middleman.

Bitcoin · Proof-of-Work

Bitcoin spends real energy so the longest chain is costly to forge — turning electricity into agreement.

Ethereum · Programmable

Ethereum makes the ledger programmable: contracts are agreements that execute themselves on shared state.

ZK · Identity

Zero-knowledge proofs verify a fact while hiding the data — digital identity that proves a claim without exposing you.

ZERO-KNOWLEDGE · PROVE WITHOUT REVEALING
CLAIM
“I know a secret x — e.g. I am over 18.”
SECRET x
•••••••• (never sent)
WHAT THE VERIFIER LEARNS
Exactly one bit: “the claim is true.”
Verifier is convinced x exists, learns nothing about x itself.
ψ

Psy Protocol is a privacy-focused, independent public blockchain — its own chain — that uses zero-knowledge proofs (built on Plonky2 over the Goldilocks field with Poseidon hashing) to settle fully private transactions.

Trust used to require a central authority. Cryptography lets strangers agree on shared information — one history, one ledger — without one.

07

AI & the Compression of Reality

Intelligence as compression plus prediction

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.

Embedding Space · 2D Projection51 tokens
dogcatwolflionhorsefishbirdkingqueenmanwomanprinceprincesschildonetwothreeseventenhundredjoyfearangerlovesorrowhopecomputernetworkcodedatamodelsignalredbluegreengoldvioletbreadriceappleteawineParisFranceItalyRomeJapanTokyopastnowfuture
dim 1 →
↑ dim 2
animalspeople / royaltynumbersemotionstechnologycolorsfoodplacestime
Nearest neighbours · similarity
king(24, 20)

// similarity = 1 − normalised distance. click any token to explore.

Vector arithmetic · analogy

// the same geometric offset maps gender, capital → another word.

An embedding compresses meaning into a position. Trained only to predict the next token, models discover that nearness should encode relatedness — so directions in the space come to mean things (gender, plurality, capital-of). Prediction over such a geometry is itself a powerful form of compression: lossless of structure, lossy of detail.

Bengio's neural language model (2003) → word2vec (2013) → transformers: meaning as geometry, learned from prediction alone.

08

Quantum Information

Does the universe compute?

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.

Qubit · Bloch spheresuperposition
|0⟩|1⟩
θ polar1.05 rad · 60°
φ azimuth0.85 rad · 49°
|ψ⟩ = 0.865 |0⟩ + 0.331+0.377i |1⟩
P(0) = cos²(θ/2)
74.9%
P(1) = sin²(θ/2)
25.1%
measurement histogram (Born rule)n = 0
0
|0
0
|1

// each outcome is random; many measurements reproduce P(0), P(1).

Entanglement · Bell pair (Φ⁺)

|Φ⁺⟩ = (|00⟩ + |11⟩) / √2

?
qubit A
?
qubit B

// measuring A instantly fixes B's correlated outcome. real, yet cannot send information faster than light — the outcome itself is random.

Information is physical
Bit vs Qubit

A bit is 0 or 1. A qubit is a direction on a sphere — a superposition that, on measurement, becomes one classical bit.

Bekenstein bound · holography

The information a region can hold scales with its bounding surface area, not its volume — as if reality's storage lives on a boundary.

Black-hole information paradox

If a black hole evaporates, where does the information that fell in go? Quantum mechanics says it cannot vanish. Still debated.

Wheeler's 'it from bit'

Does reality fundamentally compute? Digital physics asks whether every 'it' derives from yes/no answers. An open question — stated as open.

Classical information is the bit; quantum information is the qubit, where superposition and entanglement let a system hold correlations no list of bits can express. Measurement converts the quantum to the classical — probabilistically, by the Born rule. Whether the universe is, at bottom, a computation remains genuinely open.

09

Memes & the Collective Mind

Culture as information under selection

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.

Memetic Contagion · Social Grapht = 0
Origin idea· 0
Variant β· 0
Variant γ· 0
Variant δ· 0

// dim = susceptible · glowing = carrier · faded = immune / forgotten

Transmissibility R₀0.42
weakviral
Mutation

copies occasionally drift into new strains

Controls

"Reset" rebuilds the network

Susceptible
58
Carriers
0
Immune / fading
0
Variant share over timecarrier count
running to record…
Ideas are replicators

Dawkins called them memes: units of culture that copy from mind to mind. Myths, melodies, scientific theories, slogans and internet memes spread along social ties, mutate as they are retold, and compete for the scarce resource of human attention. What survives is not the truest idea but the most transmissible — culture is information under selection.

10

The Planetary Information Organism

Where the engine is heading

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.

information-stack.sim
recursion · 0/12
$ ready. press RUN to simulate the evolution of information.
The Information Stack
  1. 00Physics
  2. 01Chemistry
  3. 02Biology · DNA
  4. 03Neural systems
  5. 04Language
  6. 05Writing
  7. 06Computation
  8. 07Networks · Internet
  9. 08Economics · Cryptography
  10. 09AI systems
  11. 10Collective intelligence
  12. 11Planetary cognition
The Recurring Law

Each layer compresses, transmits, error-corrects and computes the layer below — then becomes substrate for the next.

// compress → transmit → correct → compute → substrate

The same process, recursing

From quantum events to planetary networks, one motif repeats: information is captured, copied, corrected, and computed — and each completed layer becomes the raw material of the next. Run the console and watch the metric climb through orders of magnitude. The open question is whether this ascent converges on a single planetary-scale information organism that thinks at the speed of light.

Artificial General Intelligencenear

Systems that compress and predict across every domain at once — a general-purpose engine for turning information into understanding and action.

Brain–Computer Interfacesemerging

Direct channels between neural signals and machines, narrowing the gap between thought and digital information to near zero latency.

Planetary Cognitionthis century

Billions of minds, sensors, and models coupled tightly enough to sense, model, and decide as one distributed organism.

Synthetic Memoryunderway

External, searchable, perfectly persistent memory — civilization's collective recall offloaded into infrastructure that never forgets.

Digital Consciousnessspeculative

The open frontier: whether information processed in the right pattern can give rise to genuine inner experience — and how we would ever know.

Programmable Societiescontested

Rules, money, and coordination encoded as transparent, executable information — and the deep question of who gets to write the code.

Open questions · 未解之问

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.

01
Is information physical, or is the physical informational?

Landauer · Wheeler · digital physics

02
Does meaning reduce to information, or escape it?

semantics vs. Shannon's syntax

03
Is intelligence simply very good compression and prediction?

Solomonoff · Hutter · large models

04
Can information ever truly be destroyed?

the black-hole information paradox

05
Where does the individual mind end and the collective begin?

extended mind · networked cognition

06
If reality computes, what is it computing — and for whom?

the hardest question of all

Meta-model · 元模型

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.

CCompression

Squeezing redundancy out — saying more with less.

TTransmission Speed

How fast information crosses distance.

EError Correction

Surviving noise; copying without drift.

ASymbolic Abstraction

Standing for the world with reusable marks.

NNetwork Connectivity

How many nodes can reach how many others.

MMemory Persistence

How long information endures intact.

PComputational Capacity

Turning stored information into new information.

CCompressTTransmitECorrectAAbstractNNetworkMMemoryPCompute
Overlay systems
Seven capacities

Hover an axis to read its definition. Every information system is a different weighting of the same seven terms.

Information Power
= C + T + E + A + N + M + P

It from bit — or bit from it?

We began with difference and ended with a planet wired into one updating mind. The same operations — store, copy, compress, transmit, correct, compute — recur from molecules to markets to models. Whether information merely describes the world or partly constitutes it remains open. But the trajectory is unmistakable: matter learned to remember, then to speak, then to think. The engine is still accelerating, and we are its latest medium.

An educational synthesis of information theory, computer science, biology, cryptography and philosophy. The simulations are illustrative simplifications, not exact replicas of production systems. Open questions are stated as open.

Information Engine · 信息引擎 · Psyverse · 2026