← all systems · see it in The System
Computer
A computer is a machine that can be programmed to automatically carry out sequences of arithmetic or logical operations (computation). Modern digital electronic computers can perform generic sets of operations known as programs (src:wp:computer). In computer science's usage, 'merely some physical machinery that can in theory compute any computable function'; outside it, 'human-made devices that engage in sequential processing of inputs to produce outputs' (Richards & Lillicrap 2022; contention C-596).
Properties published scienceour proposal
Sourced fields, and our own readings.
| born at | 1830s design (Babbage's Analytical Engine) to 1941-1948 working machines (Z3, Colossus, ENIAC, Manchester Baby); no single first (contention C-568) |
| born precision | decade range; contested |
| boundary | Inside: the machine at work, from its input devices to its output devices: input organs, memory, processor (control and arithmetic), the operating system and the program it is running (von Neumann's five parts plus what runs them). Outside: the outside recording medium as stored (R), the people who write programs and read results, the power supply grid, the networks it joins, the heat sink. Where the boundary sits depends on which definition of 'computer' is used (contention C-596). |
| nested level | Built from switching elements (relays, vacuum tubes, then transistors in integrated circuits) arranged as logic gates, memory and processor; carries the stored-program Seed; embedded in networks (the Internet), data centres and the products it controls; used by people. |
| scale | technological |
Measures (17) published science
| characteristic clock: Manchester Baby instruction rate | about 700 instructions per second |
| Manchester Baby memory word access | 360 microseconds |
| Manchester Baby store refresh | 20 milliseconds for 32 words |
| Manchester Baby first program | 21 June 1948 date |
| ENIAC addition rate | 5000 additions or subtractions per second |
| Moore 1965: growth of components per chip at minimum cost | roughly a factor of two per year |
| Moore 1965 projection for 1975 | 65,000 components per integrated circuit |
| Alpha 21264 rated maximum power | 95 W (72 W typical applications) |
| DRAM: machines with a correctable memory error | about one third of machines per year (Google fleet, 2006-2008) |
| DRAM: DIMMs with correctable errors | 8.2% of DIMMs per year |
| DRAM error rate | 25,000 to 70,000 errors per billion device hours per Mbit |
| machines with uncorrectable memory errors | 1.3% of machines per year (2-4% on some platforms) |
| memory scrubbing rate | 1 GB every 45 minutes typical |
| high-energy neutron flux, New York, sea level | about 13 neutrons cm^-2 h^-1 (E > 10 MeV) |
| neutron flux above 50 MeV across 23 exascale sites | 3.7 +/- 0.2 (Guangzhou, sea level) to 26.4 +/- 1.1 (Los Alamos, 2,125 m) cm^-2 h^-1 |
| ICT share of global greenhouse gas emissions | 1.8-2.8% (peer-reviewed estimates); 2.1-3.9% adjusted percent of global GHG |
| Landauer bound | k_B T ln 2 energy per erased bit (reached within 1% in a 100 ms protocol) |
Open contentions (3) open contention
Where sources or readings disagree. Both stay on the map.
- C-563 What causes most computer memory errors: cosmic rays or hardware faults?
booked: The map keeps both: cosmic-ray neutrons as a cause of random errors, and field data showing that hardware faults dominated in one large fleet.
the other reading: hard errors dominate: 'We provide strong evidence that memory errors are dominated by hard errors, rather than soft errors, which previous work suspects to be the dominant error mode' (Schroeder 2009; their instruments could not reliably tell the two apart); cosmic-induced radiation is the source of silent errors, neutrons the key (Asorey 2022) - C-568 What was the first computer?
booked: The map names no single first. It keeps each claim (Z3, Colossus, ENIAC, Manchester Baby) with the qualifiers that make it true.
the other reading: Z3 (1941): 'the world's first working electromechanical programmable, fully automatic digital computer'; Colossus: 'the world's first electronic digital programmable computer'; ENIAC: 'the first electronic, Turing-complete device'; Manchester Baby (21 June 1948): 'the world's first stored-program computer'; historians: 'the question of the “first computer” is no longer seen as legitimate' (Haigh 2014); Williams: 'If you add enough adjectives to a description you can always claim your own favorite' - C-596 What is a computer, and is the brain one?
booked: The map uses the everyday sense, a human-made machine programmed to carry out arithmetic or logical steps, and keeps the broader sense in which brains count as computers alongside.
the other reading: computer science's formal sense: 'merely some physical machinery that can in theory compute any computable function'; on it 'the brain is literally a computer; there is no metaphor' (Richards & Lillicrap 2022); outside computer science: 'human-made devices that engage in sequential processing of inputs to produce outputs'; on it brains are not computers and the metaphor is weak (same source; Cisek, Epstein, Cobb as cited there); the human computer: a person who calculates (1640s sense)
Partners and parts published science
Built from / builds (17)
- Internet
- Computer memory
- Transistor
- Input devices (input organs, I)
- Central processing unit
- Output devices (output organs, O)
- Operating system
- Buses
- Competition for a bottleneck link's capacity
- Botnet
- Attackers against defenders for control of computing
- Chaos and nonlinear dynamics (the practice, and the chaotic behaviour it studies)
- Computer network
- Botnets against botnets for the same devices
- Attack traffic against legitimate users for the bottleneck
- Computers and workers for routine tasks
- Competition among matrix multiplication algorithms for real computing time
Sits inside / holds (13)
- Computer memory
- Software
- Internet
- Human society (from local community up to societies)
- Data centre
- Algorithm (the practice of step-by-step procedures and the structure they compute)
- Central processing unit
- Operating system
- Linear algebra (the practice of solving, transforming and taking apart linear systems with vectors and matrices, and the structures it works with)
- Electric circuit
- Computer hardware
- Database
- Data structure
Trades with (valence) (74)
- Engineering (design) via sys:sensory-environment · Computer gain
- Software via sys:electromagnetic-field · Computer mixed
- Internet via sys:electromagnetic-field · Computer gain
- Internet protocol suite (TCP/IP): the layered datagram protocols every host and router runs, the RFC record that specifies them, and the community that keeps them via sys:electromagnetic-field · Computer gain
- Vector space: sets whose elements can be added and scaled, as mathematics describes it, with bases and dimension, linear maps, inner products, and their use as the state spaces of physics and data via sys:computer-memory · Computer gain
- Real number (the practice of measuring the continuum, and the complete ordered field) via sys:sensory-environment · Computer gain
- Entropy via sys:thermal-energy · Computer loss
- Hamiltonian and Lagrangian mechanics (analytical mechanics: classical mechanics built from energy functions and the stationary action, and the practice of using them) via sys:computer · Computer gain
- Statistical mechanics via sys:computer · Computer gain
- Electrical grid via sys:electromagnetic-field · Computer gain
- Number and counting (the number sense, the practice of counting, and the structure it works on) via sys:sensory-environment · Computer gain
- Photography via sys:sensory-environment · Computer gain
- Mathematics (the field: the practice and the structures it studies) via sys:sensory-environment · Computer gain
- Arithmetic (the practice of calculating, and the operations and laws it works with) via sys:electromagnetic-field · Computer gain
- Map (maps and map-making) via sys:written-record · Computer gain
- Cosmic rays via sys:cosmic-rays · Computer loss
- Thermal energy (heat, temperature and heat flow) via sys:thermal-energy · Computer loss
- Human (Homo sapiens) via sys:sensory-environment · Computer gain
- Climate system (climate) via sys:atmosphere-of-earth · Computer even
- Electrical power grid via sys:electromagnetic-field · Computer gain
- Algorithm (the practice of step-by-step procedures and the structure they compute) via sys:written-record · Computer gain
- Operating system via sys:computer-hardware · Computer gain
- Labour market via sys:economy · Computer even
- Turing machine: the abstract machine that reads and writes one symbol at a time on an unbounded tape by a finite table of rules, the model of computation of logic and computer science via sys:written-record · Computer gain
- Algebra (the practice of solving for unknowns and transforming expressions, and the structures it works with) via sys:sensory-environment · Computer gain
- Electricity via sys:electromagnetic-field · Computer gain
- Electronics via sys:electromagnetic-field · Computer gain
- Statistics: the practice of collecting and reasoning from data, the intuitive statistics it grows from, and the structure (samples, distributions, estimators) it works on via sys:written-record · Computer gain
- Linear algebra (the practice of solving, transforming and taking apart linear systems with vectors and matrices, and the structures it works with) via sys:sensory-environment · Computer gain
- Cryptography via sys:electromagnetic-field · Computer gain
- Accounting (the field and practice) via sys:computer · Computer gain
- Computer-aided design via sys:electromagnetic-field · Computer even
- Function (mathematics): the assignment of exactly one output to each input, and the practice of defining, drawing, composing and using it via sys:sensory-environment · Computer gain
- Trigonometry (the practice of tying angles to lengths, and the structure it works on) via sys:sensory-environment · Computer gain
- Calculus (the practice of handling change and accumulation by limits, and the structures it works on) via sys:sensory-environment · Computer gain
- Engineering design process via sys:sensory-environment · Computer even
- Mechanical engineering (the field) via sys:computer · Computer even
- Software engineering via sys:computer-hardware · Computer gain
- Limit (mathematics): the value a sequence or function approaches, and the practice of approaching, bounding and proving it via sys:sensory-environment · Computer gain
- Differential equations (the practice of writing and solving laws of change, and the equations it works on) via sys:sensory-environment · Computer mixed
- Complex analysis (the practice of working with functions of a complex variable, and the holomorphic structure it works on) via sys:sensory-environment · Computer mixed
- Signals intelligence via sys:computer · Computer gain
- Tokamak via sys:electromagnetic-field · Computer even
- Cryptographic hash function via sys:electromagnetic-field · Computer mixed
- Public-key cryptography via sys:sensory-environment · Computer gain
- Computer security via sys:electromagnetic-field · Computer mixed
- Artificial intelligence via sys:sensory-environment · Computer gain
- Botnet via sys:computer · Computer loss
- Information security via sys:electromagnetic-field · Computer gain
- Logic: the study and practice of correct reasoning (formal and informal, across its traditions) and the valid inference it works on, together with how people actually reason via sys:electromagnetic-field · Computer gain
- Physics (the field) via sys:electromagnetic-field · Computer gain
- Classical mechanics (Newtonian mechanics: bodies, forces and motion, and the practice of describing them) via sys:computer · Computer gain
- Electric circuit via sys:electromagnetic-field · Computer gain
- Thermodynamics via sys:thermal-energy · Computer loss
- Computational physics (computer simulation of physical systems: the practice and the simulated systems it stands for) via sys:electromagnetic-field · Computer even
- Matrix (mathematics): the rectangular array of numbers and its algebra, determinants and eigenvalues, matrices as linear maps, their history, and their use in physics, graphics and machine learning; the practice and the structure both booked via a matrix in a machine: entries held in memory and streamed through a matrix multiply unit · Computer mixed
- Fourier analysis (the practice of splitting signals into waves and rebuilding them, and the structure it works on) via sys:sensory-environment · Computer gain
- Wireless network via sys:electromagnetic-field · Computer gain
- Chaos and nonlinear dynamics (the practice, and the chaotic behaviour it studies) via sys:computer · Computer even
- Database via sys:computer-hardware · Computer even
- Computer network via sys:electromagnetic-field · Computer gain
- Management via sys:organization · Computer gain
- Digital identity via sys:human-society · Computer mixed
- The archive and the historical record via sys:written-record · Computer gain
- Computer law via sys:computer · Computer even
- Biology (the field) via sys:sequential-information · Computer gain
- Computer science (the field: the practice and the computation it studies) via sys:sensory-environment · Computer gain
- Mathematical proof: the practice of showing that a statement follows from axioms and earlier theorems, how it is written, checked (by people and machines) and accepted, and the derivations it works on via sys:written-record · Computer gain
- Mathematical logic: the field that studies logic within mathematics (model theory, proof theory, set theory, computability theory) and the formal systems, structures and limit results it works on via sys:thermal-energy · Computer loss
- Machine via sys:written-record · Computer gain
- Infinity: the boundless or endless as mathematics and philosophy describe it (potential and actual infinity, Cantor's sizes of infinity, infinity in calculus, geometry, physics and computing) and the practice of the people who argue over it via sys:sensory-environment · Computer gain
- Series (mathematics): sums of infinitely many terms, and the practice of summing, testing and using them via sys:sensory-environment · Computer gain
- Graph theory: the practice of studying vertices joined by edges (problems, proofs, algorithms, its people and books) and the graphs it works on via sys:computer-hardware · Computer even
- Prime numbers (the practice of number theory that finds, proves and uses them, and the structure of the primes it works on) via sys:sensory-environment · Computer gain
expressed by (19)
- Generation/Synthesis
- Scaffolding/Enablement (Meta-Tooling)
- Templating / Standardization
- Interoperability/Transduction (Bridging)
- Encapsulation & Modularity (Abstraction)
- Evolvability/Plasticity Engineering (Design for Future Variation)
- Distributed Autonomy/Decentralized Execution (Subsidiarity)
- Internal Coordination/Regulation
- Sensing/Scanning
- Trail-Marking / Externalized Memory (Stigmergy)
- Surprise / Change-Point Detection
- Reactive Response
- Physical/Virtual Barrier
- Self-Monitoring
- Resilience & Repair
- Redundancy & Diversity
- Triage & Prioritized Allocation
- Quality Control & Error Correction
- Bridging Supply Gaps (Store & Dormancy)
floor filled by (77)
- Engineering (design)
- Software
- Internet
- Periodic table (the periodic system of the elements and its law, as chemistry describes it)
- Vector space: sets whose elements can be added and scaled, as mathematics describes it, with bases and dimension, linear maps, inner products, and their use as the state spaces of physics and data
- Real number (the practice of measuring the continuum, and the complete ordered field)
- Hamiltonian and Lagrangian mechanics (analytical mechanics: classical mechanics built from energy functions and the stationary action, and the practice of using them)
- Statistical mechanics
- Mathematics (the field: the practice and the structures it studies)
- Arithmetic (the practice of calculating, and the operations and laws it works with)
- Cosmic rays
- Silicon atom
- Human (Homo sapiens)
- Computer memory
- Stored-program principle (instructions held in the same memory as data)
- Internet protocol suite (TCP/IP): the layered datagram protocols every host and router runs, the RFC record that specifies them, and the community that keeps them
- Transistor
- Electrical grid
- Input devices (input organs, I)
- Central processing unit
- Operating system
- Device driver
- Direct memory access
- Network interface controller
- Algorithm (the practice of step-by-step procedures and the structure they compute)
- Algebra (the practice of solving for unknowns and transforming expressions, and the structures it works with)
- Electricity
- Electronics
- Statistics: the practice of collecting and reasoning from data, the intuitive statistics it grows from, and the structure (samples, distributions, estimators) it works on
- Linear algebra (the practice of solving, transforming and taking apart linear systems with vectors and matrices, and the structures it works with)
- Cryptography
- Computer-aided design
- Function (mathematics): the assignment of exactly one output to each input, and the practice of defining, drawing, composing and using it
- Trigonometry (the practice of tying angles to lengths, and the structure it works on)
- Calculus (the practice of handling change and accumulation by limits, and the structures it works on)
- Engineering design process
- Mechanical engineering (the field)
- Software engineering
- Limit (mathematics): the value a sequence or function approaches, and the practice of approaching, bounding and proving it
- Differential equations (the practice of writing and solving laws of change, and the equations it works on)
- Complex analysis (the practice of working with functions of a complex variable, and the holomorphic structure it works on)
- Signals intelligence
- Tokamak
- Cryptographic hash function
- Public-key cryptography
- Digital signatures and certificates
- Computer security
- Artificial intelligence
- Authentication
- Botnet
- Denial-of-service attack
- Information security
- Logic: the study and practice of correct reasoning (formal and informal, across its traditions) and the valid inference it works on, together with how people actually reason
- Physics (the field)
- Classical mechanics (Newtonian mechanics: bodies, forces and motion, and the practice of describing them)
- Electric circuit
- Computational physics (computer simulation of physical systems: the practice and the simulated systems it stands for)
- Matrix (mathematics): the rectangular array of numbers and its algebra, determinants and eigenvalues, matrices as linear maps, their history, and their use in physics, graphics and machine learning; the practice and the structure both booked
- Fourier analysis (the practice of splitting signals into waves and rebuilding them, and the structure it works on)
- Organization
- Computer hardware
- Chaos and nonlinear dynamics (the practice, and the chaotic behaviour it studies)
- Crystallography
- Computer network
- Management
- Vulnerability management
- Digital identity
- Computer ethics
- The archive and the historical record
- Computer law
- Computer science (the field: the practice and the computation it studies)
- Mathematical proof: the practice of showing that a statement follows from axioms and earlier theorems, how it is written, checked (by people and machines) and accepted, and the derivations it works on
- Financial statement
- Infinity: the boundless or endless as mathematics and philosophy describe it (potential and actual infinity, Cantor's sizes of infinity, infinity in calculus, geometry, physics and computing) and the practice of the people who argue over it
- Series (mathematics): sums of infinitely many terms, and the practice of summing, testing and using them
- Graph theory: the practice of studying vertices joined by edges (problems, proofs, algorithms, its people and books) and the graphs it works on
- Prime numbers (the practice of number theory that finds, proves and uses them, and the structure of the primes it works on)
Kinds (10)
- Stored-program computer (the carrier machine)
- Analog computer
- Electromechanical computer (relays; Z3)
- Mechanical computer (Babbage's engines)
- Embedded computer (single-purpose device)
- Personal computer and mobile device
- Supercomputer
- Human computer (a person who calculates)
- Synchronous (clocked) computer
- Asynchronous (clockless) computer
wired (10)
- Human (Homo sapiens)
- router
- Domain Name System
- Bridges and switches
- Command and control (C&C)
- Protein Data Bank
- Graph theorists and their collaborations (the practitioners)
- Great Internet Mersenne Prime Search
- Repeaters and hubs
- X-ray area detector (film, CCD, pixel detector)
Notes published science
- A computer vision network trained only to tell cats from cars grew its own 'number neurons': about one unit in ten in its top layer became tuned to how many dots it saw, with the same log-scaled blur real neurons have. Another study found such units in networks that had never been trained at all (src:nasr-2019, src:kim-2021).
- The founding document of the stored-program computer draws it as a nervous system: von Neumann wrote in 1945 that memory and the arithmetic and control units 'correspond to the associative neurons in the human nervous system', and that input and output are 'the sensory or afferent and the motor or efferent neurons' (src:vonneumann-1945-edvac). The channels-board-outlets shape was there from the first page.
- Memory errors are not rare: in Google's fleet about a third of machines had at least one memory error a year, and the errors were mostly hard faults rather than the cosmic-ray bit flips earlier work suspected (src:schroeder-2009-dram). The neutrons are real too: about 13 high-energy neutrons cross each square centimetre every hour at sea level in New York (src:asorey-2022-neutron).
- In 1965 Gordon Moore wrote that if you shrank a computer down to just its components, 'we would expect it to glow brightly' (src:moore-1965). Today's chips carry a thermostat: a thermal sensor trips, the operating system steps in, and the chip slows itself until it cools (src:brooks-martonosi-2001-dtm).
- Fly brains solved a famous computer-network problem first: the cells that become the fly's bristle sensors pick themselves using one-bit messages and no knowledge of how many neighbours they have, and copying that gave a better algorithm for wireless networks (src:afek-2011, src:navlakha-2011).
- In October 1986 the Internet's first 'congestion collapse' cut the data rate between two Berkeley sites 400 yards apart from 32 Kbps to 40 bps, a thousandfold drop; the fix was to make every computer slow down when it noticed lost packets, a rule that still runs in TCP today (src:jacobson-1988).
- Monkey motor-cortex cells fire as a cosine of the direction the arm is about to move: about three in four of the tuned cells fit D = b0 + c1 cos(theta - theta0) (src:georgopoulos-1982).
- A bacterium keeps a running total: the chemotaxis pathway works as integral feedback control, the same strategy engineers use, which lets the cell return exactly to its old behaviour whatever its protein levels, so it reacts to changes and not to levels (src:yi-2000).
- A billiard table can compute anything a computer can: 2D billiard trajectories can simulate any Turing machine, so 'There is no general algorithm that can decide basic questions such as whether a trajectory is periodic' (src:miranda-ramos-2026-billiards-turing). Next to chaos, that is a second wall on predicting plain Newtonian motion.
- In 1878 a single human computer at the US Coast Survey, Myrick Doolittle, solved 41 simultaneous equations in a week with paper and pencil; in 1946 four mathematicians, Alan Turing among them, needed about two weeks for 18 with desk calculators. And the method everyone calls 'Gaussian' was named for Gauss only in the 1950s, by confusion over its history (src:grcar-2011, src:wp:gaussian-elimination).
- The world's fastest computers are ranked by one school exercise: how fast they solve a big system of linear equations. The benchmark began in 1979 as an appendix of timings for 23 computers, 'in some sense, an accident' (src:dongarra-2003).
- The textbook way to solve some field problems on a computer quietly gets the wrong answer: on a domain with an inward corner, a plain finite element method stays perfectly stable and still converges to a function that is not the solution. The fix was to carry the structure of gradient, curl and divergence down onto the mesh itself (src:arnold-falk-winther-2009).
- A computer pays a tiny toll on every arithmetic step, and in the best planetary integrators those tolls add up like a drunkard's walk: the energy error grows only as the square root of the number of steps (Brouwer's law). Let one step be slightly biased and the error grows in a straight line instead (src:rein-tamayo-2015-whfast).
- The first 'numerical experiment' (1955) found that energy in a vibrating chain came back instead of spreading out, and it was found only because someone forgot to stop the computer. The programmer, Mary Tsingou, was left off the authors for 50 years (src:dauxois-2008-fput).
- In 1978 the RSA paper reckoned that factoring a 200-digit key would take 3.8 billion years with the best method then known; in 2019 a team factored a 240-digit RSA number in about 1,000 computer-core-years, and a 250-digit one in about 2,700. The mathematics didn't break; the attacks got better, so the keys got longer. (src:rivest-shamir-adleman-1978, src:boudot-2020-rsa240)
- Shrinking the step does not make a computed derivative better forever. The textbook limit definition, run on a computer, has two errors: one falls as the step shrinks, the other rises and takes over, so the estimate has a best step in between, and the survey says it commits 'both cardinal sins of numerical analysis' (src:baydin-2018).
- A sine table built the obvious way, each entry from the one before, does not draw a circle: 'this algorithm would draw a logarithmic spiral rather than a circle', and in a computer even the better recurrence lets rounding errors grow with the table's length (src:wp:trigonometric-tables).
- Bodies made the same bargain long before computers: when a miss costs far more than a false alarm, 'optimal systems tend to produce false alarms', which is why fever, pain and panic can look like overreactions (the smoke detector principle, src:nesse-2025-failures). Network defenders choose it too, cutting misses 'at the cost of increasing false positives' (src:nist-sp800-94).
- A network monitor and the computer it guards can disagree about what arrived. Send a packet the IDS accepts but the host throws away, or the reverse, and the IDS sees a different conversation from the real one; in 1998 every commercial network IDS tested fell to both tricks (src:ptacek-newsham-1998). Snort's author described it as filling an 'ecological niche' (src:roesch-1999-snort).
- The break-in that forced Britain to write its computer crime law used the password 1234. In 1984-85 two men got into British Telecom's Prestel service after one of them saw an engineer's login (username 22222222) at a trade show; they were convicted under a forgery law, acquitted on appeal, and the gap their acquittal exposed led to the Computer Misuse Act 1990, which Canada and Ireland then copied (src:wp:computer-misuse-act-1990).
- A law professor's 2003 article found its way into the Supreme Court 18 years later: Orin Kerr argued that computer crime laws should only cover getting past a real gate in the code, the Court cited his brief when it ruled 6-3 in 2021 that 'exceeds authorized access' is a gates-up-or-down question, and the Justice Department then said it would not charge good-faith security research (src:kerr-2003-cybercrimes-scope; src:scotus-2021-van-buren; src:xiao-2025-anti-fraud-research).
- Codebreaking gave us the computer: the ten Colossi built to break German codes were the first programmable electronic computers, made by a Bletchley team that was 75% women (src:wp:signals-intelligence, src:wp:bletchley-park). And after the June 2013 revelations, views of privacy-sensitive Wikipedia articles dropped by over 30% at once (src:penney-2016-chilling-effects).
- Forgetting costs energy: erasing one bit must dump at least kT ln 2 of entropy as heat, but any computation can in principle be done without that toll if it keeps a copy of its input. The physicist who wrote the notes defending this, Charles Bennett, won computer science's top prize, the Turing Award, for 2025 (src:bennett-2003, src:wp:turing-award).
- Linked lists are older than computers by more than two thousand years: scribes copying papyrus scrolls wrote the first word of the next scroll at the end of each one (the 'reclamans', 'shouting back'), and printers later put the next page's first word at the foot of every page (src:wp:linked-list).
- Researchers who counted the ways to combine data-structure building blocks found at least 10^32 possible two-node designs, more than the stars in the sky, while computer scientists have published only a couple of thousand in sixty years; like Mendeleev's table, the empty cells point to designs nobody has built yet (src:idreos-2018-periodic-table).
- In 2024 an online group of mostly amateur mathematicians settled a 60-year-old question: the longest any 5-state machine can run before stopping is exactly 47,176,870 steps. They had to check 181,385,789 machines, and the whole proof was checked by computer. With one more state, a machine called Antihydra stops only if a Collatz-like puzzle about odd and even numbers is false, so nobody knows. Sources: src:bbchallenge-2025, src:wp:busy-beaver.
- The same limit turns up far from computers: a single ball bouncing around a flat billiard table can act as any Turing machine, so some questions about where the ball will go can never be answered by any method, not even in principle. Source: src:miranda-ramos-2026-billiards-turing.
- Your computer cannot hold almost any real number, and the trouble is a carry that may never arrive: to round log(x) correctly to the 53 bits a double stores, some inputs need about 117 bits of working precision, and adding a list of numbers in a different order gives a different answer (src:boldo-2023).
- A computer cannot take the limit that defines the derivative: shrinking h past about the square root of machine epsilon makes the answer worse, not better, because the subtraction eats the digits; the cure is to skip the limit (automatic differentiation) (src:baydin-2018, src:wp:numerical-differentiation).
- When a computer adds a long list of numbers one after another, each addition passes on its rounding error to the next, a chain n - 1 steps deep; adding in pairs cuts the chain to about log2 n steps, and carrying the lost bits forward (compensated summation) brings the answer back to machine precision (src:hallman-ipsen-2021).
- The electron's magnetic moment, the most precisely computed and measured number in physics, is a series that physicists believe diverges: the four-loop term alone took 891 Feynman diagrams, and a thousand times more computer time bought thirty times more precision (src:todorov-2018).
- When computers compared 20th-century novels by white and black American writers, the biggest single source of shared wording was the King James Bible (550 of over 1,200 shared passages); but white writers tended to quote it in solitary scenes and black writers in scenes where characters argue over it. (src:so-2019-race-writing-computation)
- In a computer model of the Peter principle, promoting people at random kept an organisation more efficient than promoting the best performer, and real data from 214 sales firms fit the same story: each doubling of a new manager's past sales went with a 7.5% drop in their team's sales (src:pluchino-2010-peter, src:benson-2018-peter).
- The same constitutional trick turns up inside genomes: in the "parliament of genes", the majority of genes suppress selfish elements that cheat at meiosis, and Scott and West's models find "the majority interest within the genome generally prevails over 'cabals' of a few genes" (src:scott-west-2019-parliament-of-genes; Leigh 1977). Computer security found it independently as "two keys" (src:saltzer-schroeder-1975).
Layers
The ways it takes in its world, where they meet, and how it acts back, each from L1 (where it meets the world) up.
signals in, programs run, results out (reader's view)
- L1: The world the input devices touch: key presses, touch, light, sound, link signals, stored records, mains power; people make and sense the outside medium
- L2: Input devices (the input organs): keyboard, mouse, touchscreen, camera, microphone, network controller
- L3: The I/O path: device drivers, interrupts and DMA move data into memory
- L4: Memory: instructions and data as bits in one store
- L5: The processor: control unit, ALU, registers; fetch, decode, execute, paced by the clock
- L6: The operating system: resources, scheduling, drivers, error handling
- L7: The running program and its results, recorded or shown back to the world; goals set by people
the boardroom (shared floors)
- L4: Memory (M): one store for instructions and data, bits held in circuits; every input lands here first
- L5: The processor (CA + CC): control unit, ALU and registers in the fetch-decode-execute cycle, paced by the clock and interrupted by devices
- L6: The operating system: manages hardware and software resources, schedules tasks, runs interrupt handlers and drivers, logs and acts on errors
- L7: The running program: the computation the user asked for, carried through to results recorded or shown; in Licklider's partnership the goals stay with the people
the store: stored program and data in fast electronic memory (from the Seed's side)
- L1: Electric charge holding each bit; circuits switched on and off
- L2: Logic gates: circuits controlling circuits
- L3: Words of the store, instructions and data alike (the Seed's quanta)
way in: Camera (light)
- L1: Light from the scene
- L2: Image sensor (CCD or CMOS active-pixel): the pattern of currents carries the image
- L3: A driver and DMA move the image data into memory
way in: Clock
- L1: An electrical oscillator, possibly crystal controlled
- L2: The oscillator circuit gives a periodic square wave, a set number of pulses a second
- L3: Clock distribution; clock gating turns the signal off to unneeded parts
- L4: Memory paced in turn: the Baby's store was refreshed continuously, 20 ms for its 32 words
way in: Electric power
- L1: Mains AC
- L2: The power supply unit converts it to regulated low-voltage DC
- L3: A power-good signal tells the motherboard the voltages are in spec, so it may boot
way in: Keyboard
- L1: A finger presses a key: keys act as mechanical levers or electronic switches
- L2: Key switch closes; each press corresponds to a symbol code
- L3: The code is delivered to the host by an interrupt and handed to software, which interprets it
way in: Microphone (sound)
- L1: Sound pressure at the microphone
- L2: Capacitive MEMS microphone: a membrane against perforated backplates; sensitivity set by tension, gap and perforation
- L3: Sound card and DMA move samples into memory
way in: Network
- L1: Signals on the transmission medium (TCP/IP's link layer runs on them)
- L2: Network interface controller: frames received, queued
- L3: Interrupt and DMA interfaces hand the data to the host
way in: Outside recording medium (storage: cards, tape, disk)
- L1: R, the outside recording medium: punched cards, teletype tape, magnetic tape, later disks; the natural medium for long-time storage
- L2: Input organs I transfer information from R into the machine
- L3: Disk drive controllers move the data into memory by DMA
way in: Stray particles (cosmic-ray neutrons)
- L1: High-energy neutrons made when cosmic rays hit the atmosphere; flux set by place, altitude, season
- L2: A neutron strikes silicon and makes a cascade of secondary charged particles
- L3: Small, low-voltage circuits carrying small charges are upset
way in: Temperature (on-die)
- L1: Heat dissipated by the chip's own switching
- L2: A thermal sensor or power estimator triggers
- L3: An operating system interrupt is raised
- L4: The processor saves its state before the handler runs
way in: Touch and pointing (touchscreen, mouse)
- L1: A finger or stylus on the screen, or a hand moving a mouse
- L2: Capacitive touch sensors (interdigitated electrodes) read touch and its pressure
- L3: Filtering, transfer function and control order turn device motion into pointer motion
way out: Control of machines
- L1: Industrial robots, appliances and vehicles moved by it
- L7: The program acting as a control system
way out: Display
- L1: Light from the screen, 'sensed more or less directly by human organs'
- L3: Graphics card fed by DMA
- L7: Results put into a human-accessible form
way out: Heat
- L1: Heat leaves through the chip's surface ('a surface available for cooling close to each center of heat generation')
- L5: Switching dissipates power; at least the Landauer bound per erased bit
way out: Network out
- L1: Signals on the link to other computers ('the Internet, which links billions of computers and users')
- L2: Network interface controller transmits
- L6: Data queued for transmission
way out: Recording (storage out)
- L1: Results written to R, the medium for long-time storage
- L4: Output organs O transfer from memory, never directly from the processor
way out: Sound
- L1: Sound in the room
- L3: Sound card fed by DMA
- L7: Results as sound (speaker, sound card)
Who fills each layer
- L4: Computer hardware, Database, Computer memory, Sorting algorithm
- L5: Electric circuit, Central processing unit
- L6: Operating system