The Original Computers

The word "computer" meant a person who performs calculations for three hundred and thirty-three years before it meant a machine. The shift happened on a specific mesa in New Mexico, in April 1944, when an IBM punched-card system arrived to compete against a group of women who had been doing the work at desks. The women won for two days. After that, the language changed.

By Stephen "Pseudo Publius"  |  7 May 2026

BLUF — 

The Oxford English Dictionary's first attested use of the word computer is from Richard Brathwaite's 1613 book The Yong Mans Gleanings, where it describes a person who performs mathematical calculations. The job title persisted, in roughly that meaning, until the late 1940s. Through three centuries the world's astronomical tables, navigational almanacs, ballistic firing tables, actuarial calculations, and — by 1944 — the implosion-lens hydrodynamics for the Fat Man bomb were produced by human computers, the great majority of whom were women. The T-5 group at Los Alamos under Donald "Moll" Flanders, organizing five or six women at Marchant electromechanical desk calculators in an explicit pipeline of "one adds, one multiplies, one divides, one passes the index card to the next," was a parallel computing architecture in every sense the word now means. Stephen's instinct — that the human computers have a better claim to "first computer" than ENIAC or Colossus or the ABC — is not only defensible. It is the only claim that uses the word in its original sense. The other machines are computer-shaped objects that were named after the women.^1,2,3

What the word actually meant

"I haue read the truest computer of Times, and the best Arithmetician that euer breathed, and he reduceth thy dayes into a short number." So Richard Brathwaite, in 1613, in the earliest English-language use of computer the OED has been able to find. The truest computer of times, in Brathwaite's sentence, is God. The best arithmetician that ever breathed is also God. The word at that point meant a person — or, by extension, an entity — who performed mathematical calculation. It would not mean anything else for the next 333 years.¹

The Royal Observatory at Greenwich began employing computers in 1836, in a formal job classification that lasted until 1937. The Astronomer Royal George Airy organized them in the Octagon Room into a production line: pre-printed computing forms, standardized procedures, supervisor double-checks, division of labor. The work was the reduction of astronomical observations into the data tables that allowed ships at sea to calculate longitude. Airy's computers were sometimes as young as fifteen and were initially boys; by 1865 women were entering the profession; by the early twentieth century the Royal Observatory's computing staff was the largest part of its total workforce. The same job structure was reproduced at the Nautical Almanac Office, the Bureau du Cadastre in Paris, the Astronomical Computing Bureau at Columbia under Wallace Eckert, and the Harvard College Observatory under Edward Pickering, where women like Henrietta Leavitt and Annie Jump Cannon classified hundreds of thousands of stellar spectra and established the period-luminosity relation that gave Edwin Hubble the distance ladder for the universe.^1,4

By the early 1940s, "computer" was a job title appearing in U.S. Civil Service classifications, on Aberdeen Proving Ground payroll records, on hiring memos at NACA Langley (where the African-American "West Computers," including Katherine Johnson, Dorothy Vaughan, and Mary Jackson, did the trajectory and wind-tunnel calculations later popularized by Hidden Figures), and in the recruitment materials for the Theoretical Division at Los Alamos. None of these usages referred to a machine. The reason is straightforward: there were no machines in the modern sense. There were desktop calculators, IBM punched-card accounting machines, and a few experimental devices like the Atanasoff-Berry Computer at Iowa State. The actual computational labor of the world was done by people.^5,6

T-5 Group, Los Alamos, 1943-44

The Theoretical Division at Los Alamos was organized in spring 1943 under Hans Bethe. Stanley Frankel and Eldred Nelson, both physicists who had come from the University of California at Berkeley with experience using Marchant electromechanical desk calculators, set up the first hand-computing operation. They were joined in summer 1943 by Donald "Moll" Flanders, a mathematics professor from NYU, who took over and standardized the work as Group T-5.^3,7

The recruits were, in part, women with mathematics or physics degrees. They were also, in larger part, the wives of male Manhattan Project scientists who had accompanied their husbands to the mesa and were under direct pressure from General Leslie Groves to take up war work — Groves having decided that supporting "civilians" who were not productive was a waste of scarce Los Alamos housing. Among the women in the Group T-5 hand-computing pool were Mary Frankel (Stanley Frankel's wife), Josephine Elliot, Beatrice "Bea" Langer, Augusta "Mici" Teller (Edward Teller's wife), Jean Bacher (Robert Bacher's wife), and Kay Manley (John Manley's wife). Most were college-educated; several held graduate degrees; some, like Mici Teller, had backgrounds in physics in their own right.³

The architecture of the work is what matters. Frankel and Nelson would take a physics problem — typically a partial differential equation describing a hydrodynamic process inside the implosion lens of the plutonium bomb — and decompose it into a sequence of arithmetic operations: add this, multiply by that, divide by something else, interpolate linearly between two values. Each step was recorded on a pre-printed calculating sheet with explicit cell references for input values and result values. Each computer at her Marchant calculator was assigned a single operation in the sequence. She would receive an index card carrying the input values, perform her assigned operation, write the result on a new index card, and pass it to the next computer in line. Five or six computers, each handling one step, working in pipeline.^2,3,7

Each result was checked twice by the computer who produced it; the supervisor checked again before the calculation passed to the next stage. The error rate was low. The throughput was, for hand calculation, remarkable: in March 1944, when Frankel, Nelson, and Feynman tested the proposed IBM-machine workflow against the human computing pool as a dry run, the women's pipeline operated at speeds the punched-card equipment had been predicted to achieve.⁷

This is parallel computing, in the actual sense. It is dataflow architecture, in the actual sense. The index cards are the messages between processors. The pre-printed calculating sheet is the program. The Marchant is the arithmetic logic unit. The supervisor is the error-correcting layer. None of these terms existed in 1944, but the architecture they would later describe was already running on the desks of the T-5 women.

What the calculations actually were

The Fat Man bomb required an implosion lens — a precisely shaped explosive shell that, when detonated, would symmetrically compress a sub-critical sphere of plutonium into super-critical density. The shock wave dynamics had to be calculated in detail because asymmetric compression would produce a fizzle, not a detonation. The underlying physics was the equations of hydrodynamics — coupled nonlinear partial differential equations — discretized in space and time and stepped forward numerically. Each time step required, at every spatial grid point, a sequence of arithmetic operations on the previous time step's values. The Group T-5 hand-computing pipeline did this, by hand, on Marchants. The arrival of the IBM Punched Card Accounting Machines in April 1944 mechanized the same workflow without changing its structure. The 1952 IBM 701 at Los Alamos and the MANIAC I that Nicholas Metropolis built in 1952 mechanized it further. By the 1960s the same hydrodynamics calculations were running on CDC mainframes. Today they run on GPU clusters in Memphis. The architecture is continuous; only the substrate has changed.^2,3,7

April 1944: the race

The Los Alamos IBM Punched Card Accounting Machines — eight core machines including 513 reproducing punches, 075 card sorters, 601 multipliers, and a 405 tabulator — arrived on 4 April 1944. Because of the project's classification, IBM had not been told the destination address and could not send installation engineers. Frankel, Nelson, and Feynman assembled the equipment themselves from the wiring blueprints, three days before the IBM technician (a draftee named John Johnston, transferred to Los Alamos by the Army on IBM's recommendation) arrived to do the final tuning.^2,7

What happened next is one of the cleanest data points in the history of human-versus-machine computation. Feynman, never one to leave a comparison untested, organized a head-to-head: the same hydrodynamics calculation, performed in parallel, by the women of Group T-5 on their Marchants and by the freshly assembled IBM PCAM installation. For the first day, the women were faster. For the second day, the women remained competitive. By the third day, the IBM machines pulled ahead — not because they had become faster but because the women had become tired. The PCAMs, as Nicholas Metropolis put it in his oral history, did not get tired. They also did not go home at five. By the end of the test the IBM installation had completed the calculation. The women had stopped for the night. The machines kept running.^2,6

The conclusion was not that the human computers were inadequate. The conclusion was that for sustained, repetitive arithmetic, machines could be run continuously and humans could not. Group T-5 continued operating through the war, but the strategic balance shifted. By VE Day, the IBM facility was doing the bulk of the implosion calculations. By the time MANIAC I came online at Los Alamos in 1952, the human computing pool had been disbanded. The job title "computer," which had been used in formal Los Alamos personnel records to describe people, transferred — over roughly a decade between 1946 and 1956 — to the machines.^2,3

Naomi Livesay and the bridge

The transition had a specific person at the center of it. Naomi Livesay arrived at Los Alamos in February 1944. She had a master's degree in mathematics from the University of Iowa, had taught secondary school, and — uniquely among the early Los Alamos staff — had taken formal IBM training in the operation of punched-card accounting machines. Hans Bethe, learning of her background, arranged her transfer into Frankel and Nelson's group. When the IBM equipment arrived in April 1944, Livesay was the only person at Los Alamos qualified to design and implement the plug-board programs that converted the abstract calculation flow into machine-executable operations. She was, in a precise sense, Los Alamos's first electronic-computer programmer — though the language of "programmer" did not yet exist.⁷

Livesay supervised the IBM PCAM operation for the rest of the war. The hand-computing pool continued under Flanders, doing smaller-scale and verification work; the punched-card facility under Livesay (and, later, Feynman during Nelson's hospitalization for a skiing accident in spring 1945) did the large-scale implosion simulations that fed directly into the Trinity test. Livesay's name appears in the Los Alamos technical reports of the period as "assistant scientist." It does not appear in most of the popular Manhattan Project histories. Her role in the human-to-machine transition was not recovered into the published historical record until the 2010s, primarily through the Los Alamos National Laboratory's own internal-history work culminating in LA-UR-21-20164 (Archer et al., 2021).^7,8

The semantic shift

By the time ENIAC was demonstrated at the Moore School in February 1946, the linguistic situation was unstable. ENIAC was operated, primarily, by six women — Kathleen "Kay" McNulty, Frances "Betty" Snyder, Marlyn Wescoff, Ruth Lichterman, Betty Jean Jennings, and Frances Bilas — who had been hired as human computers at the Moore School during the war and were retrained to operate the new electronic machine. They were called computers in their personnel records. The machine they operated was also, increasingly, called computer. For a few years both meanings coexisted; in oral histories from the period, "computer" can mean the woman, the machine, or both, depending on context.⁹

The disambiguation moved through the language between roughly 1948 and 1955. By 1960 "computer" in American English overwhelmingly meant a machine. The women who had been computers were retitled as "operators," "programmers," "coders," or — increasingly — squeezed out of the field entirely as the work transitioned from war service to professional discipline and was reclassified as engineering. Most of the original ENIAC women were not invited to its formal dedication ceremony in February 1946; one of them, Jean Bartik, gave an oral history in the 1990s in which she remembered being mistaken for a hostess.⁹

The naming priority is therefore unambiguous, even if the historical-recovery work is still in progress. The word computer belonged to people for 333 years. It transferred to machines in approximately a decade. The transfer was completed, in American English, by about the time UNIVAC I was delivered to the U.S. Census Bureau in March 1951. Everything after that — the priority disputes between ENIAC, ABC, Colossus, and Z3 that occupy the previous installment in this series — is an argument about which machine first qualified for a name that already belonged to people who had been doing the job for centuries.

The continuity to Memphis

This bears directly on the present series. The Memphis facility now rented out to Anthropic is named Colossus, presumably after the Bletchley codebreaker, possibly also after the 1970 film. The orbital constellation it is intended to feed is being designed to host AI inference workloads at multi-gigawatt scale. What those workloads actually do is statistical computation over high-dimensional state spaces — the same general computational pattern that the T-5 women were running on Marchants in 1944, that Feynman's PCAMs ran in 1944-45, that MANIAC I ran in 1952, that the CDC 6600 ran in 1965, that the Cray-1 ran in 1976, that ASCI Red ran in 1997, that Roadrunner ran in 2008, and that the GPUs in the Memphis Colossus are running today.^2,3

The substrate has changed by ten orders of magnitude in operations per second. The architecture of the work — decompose into arithmetic primitives, distribute across parallel processing elements, accumulate intermediate results, check for errors, integrate forward — has not changed at all. The Memphis Colossus is, in functional terms, Mary Frankel and Bea Langer and Mici Teller and Jean Bacher and Naomi Livesay, scaled up by a factor of ten billion and run continuously without rest.

This is the lineage Stephen's instinct picks out. The human computers were the original, the work was theirs, and the machines that took the name later — Colossus, ENIAC, UNIVAC, MANIAC, the rest — are downstream. Tommy Flowers's Colossus ran a statistical search across Lorenz wheel-setting hypotheses at electronic speed; the women of T-5 ran a statistical search across implosion-lens parameters at hand speed; the GPUs in Memphis run statistical searches across the parameter space of a transformer model at GPU speed. The continuity is not metaphorical. It is the same job, done by progressively faster instruments, all of which inherited their name from the people who did it first.

What this might mean for the orbital naming

The previous companion piece in this series argued for naming the orbital constellation after Tommy Flowers, or William Tutte, or one of the other Bletchley engineers. There is a stronger argument: name it after one of the human computers. Mary Frankel. Naomi Livesay. Henrietta Leavitt, whose period-luminosity relation is the reason we know how big the universe is. Annie Jump Cannon, who classified more stars than any other person in history. Katherine Johnson, whose trajectories got Apollo to the Moon. Mici Teller, who computed the implosion-lens hydrodynamics for the bomb that ended the Pacific war. Any of them would honor a longer and stronger lineage than the one running through D.F. Jones's 1966 novel.^4,5,7

It is worth noting, finally, that the conventional history that gives priority to the machines also gives priority to the men who built them — Eckert, Mauchly, Atanasoff, Zuse, Flowers, Aiken, von Neumann. The list of men is real and their contributions are real. But the lineage they enter is older than they are, and it has different names attached. Brathwaite's 1613 sentence — "the truest computer of times" — refers to a person. The 333 years of computing history that followed it refer, almost without exception, to people who are not in the priority disputes. They were the work itself. The machines were the substrate they were eventually replaced by. the claim that the women at Los Alamos who used their fingers might be the real first digital computers is correct on a more technical reading than it might first appear: they were the first computers in the literal sense the word had carried for three centuries, and the machines that succeeded them inherited the name without ever earning it from scratch.

---

Sources

1. "Computer (occupation)." Wikipedia, retrieved 7 May 2026. https://en.wikipedia.org/wiki/Computer_(occupation)  |  Brathwaite, R. The Yong Mans Gleanings. London, 1613. OED first attestation of "computer" as an English-language noun denoting a person who computes.
2. Atomic Heritage Foundation / Nuclear Museum. "The Human Computers of Los Alamos." https://ahf.nuclearmuseum.org/ahf/history/human-computers-los-alamos/  |  "Computing." https://ahf.nuclearmuseum.org/ranger/tour-stop/computing/   Includes oral histories from Jean Bacher, Kay Manley, Nicholas Metropolis, and Peter Lax.
3. Howes, R. and Herzenberg, C.L. Their Day in the Sun: Women of the Manhattan Project. Temple University Press, 1999. Standard reference on the women of the Theoretical Division and the T-5 hand-computing pool.
4. Royal Observatory Greenwich. "The post of Computer." https://www.royalobservatorygreenwich.org/articles.php?article=1000  |  Belteki, D. "Music and Murder under the Midnight Sky: the lives of the assistants and computers of the Royal Observatory at Greenwich." British Association For Local History, 25 November 2020.
5. Shetterly, M.L. Hidden Figures: The American Dream and the Untold Story of the Black Women Mathematicians Who Helped Win the Space Race. William Morrow, 2016. The standard popular treatment of the NACA Langley West Computers including Katherine Johnson, Dorothy Vaughan, and Mary Jackson.
6. Grier, D.A. When Computers Were Human. Princeton University Press, 2005. The definitive academic history of human computing from the Royal Observatory through the Manhattan Project. The phrase "the age of female computers" derives from Grier's framing.
7. Archer, B., et al. "The Los Alamos Computing Facility During the Manhattan Project." Nuclear Technology, 2021. LA-UR-21-20164. https://www.tandfonline.com/doi/full/10.1080/00295450.2021.1940060   Includes detailed account of Naomi Livesay's role and Feynman's "race" between hand computers and PCAMs.
8. Archer, B., et al. "Trinity by the Numbers: The Computing Effort that Made Trinity Possible." Nuclear Technology, 2021. https://www.tandfonline.com/doi/full/10.1080/00295450.2021.1938487
9. Light, J.S. "When Computers Were Women." Technology and Culture, vol. 40, no. 3, July 1999, pp. 455-483. The foundational scholarly recovery of the ENIAC women — McNulty, Snyder, Wescoff, Lichterman, Jennings, Bilas — and the linguistic transition of "computer" from person to machine.