POST17 REVIEW — “The Man Who Caught the Computer Disease”
POST17 REVIEW — “The Man Who Caught the Computer Disease”
Post file: _posts/2026-04-16-The-man-who-caught-the-computer-disease.md
Research file: research/people/Frankel.md
Review date: 2026-04-08
ERRORS FOUND
1. Margaret Hamilton credited for the chaos discovery — WRONG
Blog (line ~148): “One day in the winter of 1961, Lorenz wanted to rerun a simulation…” and the blog credits “with the assistance of Margaret Hamilton” for this work.
Correct: The assistant present during and after the key 1961 discovery was Ellen Fetter, not Margaret Hamilton. Hamilton programmed Lorenz’s model earlier (she arrived in 1959), but left for another project in the summer of 1961. She hired and trained Ellen Fetter as her replacement. The 1963 paper “Deterministic Nonperiodic Flow” acknowledges: “Special thanks are due to Miss Ellen Fetter for handling the many numerical computations.” Ellen Fetter also noticed the truncated printout that explained the divergence.
The blog actually names Hamilton correctly in a single sentence (“with the assistance of programmer Margaret Hamilton”), which is partially true for the earlier period — but Hamilton had already left by the time the crucial rerun discovery was made. The work that culminated in the 1963 paper was carried out with Fetter.
Fix: Change “with the assistance of programmer Margaret Hamilton (later famous for Apollo guidance software)” to “with assistance from programmers who operated the LGP-30 — first Margaret Hamilton, who arrived in 1959, and then Ellen Fetter, whom Hamilton hired and trained as her replacement in summer 1961.” Clarify that it was Fetter, not Hamilton, who was present for the chaos discovery itself, and that Lorenz’s 1963 paper explicitly acknowledges Fetter. The blog’s parenthetical attribution of the discovery moment specifically to Hamilton is inaccurate.
2. “A Million Punch Cards” — section heading conflicts with body and research file
Blog (section heading, p. ~55): Section is titled “A Million Punch Cards.”
Blog (body, line ~68): “A million punch cards of data flowed through the machine.”
Research file: States “half a million punch cards of data (some sources say ‘a million’ — the discrepancy may reflect different stages of the multi-run computation).” The research file explicitly flags the “million” figure as the weaker of the two sourced claims.
Web search findings: Multiple primary sources are split. The APS history article (December 1945) and the Bit by Bit (Haverford) source describe “a million punch cards.” The ENIAC 50th Anniversary Penn Almanac account says “half a million.” The research file treats “half a million” as the more precise figure.
Fix: The blog should either use “half a million” (the more precisely documented figure) or hedge: “half a million to a million punch cards.” The section heading “A Million Punch Cards” should be changed to “Half a Million Punch Cards” or a neutral variant if you want to stay consistent with the research file. Do not present “a million” as a settled fact.
3. Security clearance timing — “early 1950s” in blog vs “early 1949” in research file
Blog (line ~84): “In the early 1950s, Stanley Frankel lost his security clearance.”
Research file: States clearly that Frankel lost his security clearance in early 1949 — five years before Oppenheimer’s 1954 revocation. The research file flags this date specifically and notes it predates Oppenheimer by half a decade.
Fix: Change “In the early 1950s” to “In early 1949.” This is a significant factual error: the blog places the clearance revocation a full four years too late. It also changes the chronological logic of the narrative — the clearance was revoked in 1949, while the post-war Monte Carlo work described in the preceding paragraph (Manchester Mark 1, 1950; JCP paper, 1955) took place after the revocation, not before.
4. “Group T-5” identified as the IBM computing group — INACCURATE
Blog (line ~43): “Frankel created Group T-5: teams of women … organized into assembly-line workflows using Marchant and Friden desk calculators.”
This is partially correct — T-5 was indeed the human computer desk calculator group. However, the blog elsewhere (implicitly) connects T-5 to the broader IBM tabulating machine operation, which is wrong.
Research file and web search (arXiv paper on Los Alamos Computing Facility): The human desk-calculator group became T-5 under mathematician Donald Flanders when he arrived in late summer 1943. The IBM tabulating machine operation was a separate group, T-6 (“IBM Computations”). Frankel and Nelson organized T-5 first; the IBM work led to T-6. Feynman took over T-6, not T-5.
Fix: The blog’s description of T-5 as a desk calculator group is correct, but the blog needs to distinguish T-5 (desk calculators, human computers) from T-6 (IBM tabulating machines). When Feynman “took over the IBM group,” he was taking over T-6, not T-5. The narrative as written blurs these into one entity.
5. “Post-doctoral work under Oppenheimer” — may be imprecise
Blog (line ~34): “earned his Ph.D. at the University of California, Berkeley, and did post-doctoral work under J. Robert Oppenheimer.”
Research file: Says “Post-doctoral work under J. Robert Oppenheimer at UC Berkeley (1942).” Multiple external sources (Wikipedia, Alchetron, IT History Society, Nuclear Museum) consistently describe Frankel as a “post-doctoral student under Oppenheimer at UC Berkeley in 1942.” However, the research file also notes that at the Berkeley Summer Study of June–July 1942, Frankel is described as a “graduate student” (not postdoc) sitting with Oppenheimer, Fermi, Bethe, et al. The distinction matters because his PhD was from Berkeley — if he received it before 1942, he would be a postdoc; if in 1942, he might have been a late-stage graduate student.
Assessment: This is a minor ambiguity that the research file itself notes. The blog’s “post-doctoral work under Oppenheimer” matches the majority of sources and the research file summary. Flag as unresolved rather than an error requiring correction.
VERIFIED CLAIMS
The following claims in the blog were verified against the research file and web searches:
| Claim | Status |
|---|---|
| Born June 6, 1919, Los Angeles | CONFIRMED — all sources consistent |
| PhD at University of California, Berkeley | CONFIRMED |
| Graduate work at University of Rochester | CONFIRMED (graduate study, not PhD; PhD was Berkeley) |
| Post-doctoral work under Oppenheimer | CONFIRMED (with minor ambiguity noted above) |
| Joined Los Alamos T-Division 1943 | CONFIRMED |
| Group T-5: human computer assembly lines with Marchant/Friden calculators | CONFIRMED |
| Wife Mary Frankel worked as human computer | CONFIRMED |
| Frankel and Metropolis ran ENIAC H-bomb calculation | CONFIRMED |
| Calculation ran November 1945, results available December | CONFIRMED |
| First nuclear physics computation on all-electronic computer | CONFIRMED |
| Calculation revealed flaws in Teller’s Super design | CONFIRMED |
| ENIAC had 17468 vacuum tubes, 70000 resistors, 10000 capacitors | CONFIRMED (some sources say ~18000; 17468 is the precise figure) |
| ENIAC described as 30 tons, filling entire room | CONFIRMED |
| MINAC used 113 vacuum tubes | CONFIRMED |
| Germanium diodes from Hughes Aircraft | CONFIRMED |
| Magnetic drum memory | CONFIRMED |
| LGP-30 cost $47000 | CONFIRMED |
| LGP-30 weighed approximately 800 pounds | CONFIRMED |
| Over 500 units sold | CONFIRMED (~500 units, some sources say “approximately 500”) |
| LGP-30 plugged into standard wall socket | CONFIRMED |
| Joint venture with Royal McBee Corporation | CONFIRMED |
| James Cass was graduate student who helped productize MINAC | CONFIRMED |
| Lorenz used LGP-30 at MIT | CONFIRMED |
| Lorenz’s model had 12 variables | CONFIRMED |
| Discovery of sensitive dependence on initial conditions — rounding from 6 to 3 decimal places | CONFIRMED |
| Lorenz 1963 paper “Deterministic Nonperiodic Flow” in Journal of the Atmospheric Sciences | CONFIRMED |
| Lorenz died 2008 | CONFIRMED |
| Frankel died May 1978, age 58 | CONFIRMED (specific date: 2 May 1978, Los Angeles) |
| Feynman “computer disease” quote | CONFIRMED — quote in blog is accurate and consistent with the Caltech Archives transcript |
| LGP-30 called “first personal computer” | CONFIRMED as a common characterization |
| Frankel designed LGP-21 as transistorized successor | CONFIRMED |
SUGGESTED FIXES (PRIORITY ORDER)
Priority 1 — Factual errors requiring correction
-
Margaret Hamilton → Ellen Fetter (paragraph about Lorenz rerunning the simulation): The assistant credited for the 1961 chaos discovery moment and the 1963 paper is Ellen Fetter, not Margaret Hamilton. Hamilton preceded Fetter and left in summer 1961. Fix: identify both women by name and be clear about who was present when.
-
“Early 1950s” → “early 1949” for clearance revocation: Change one phrase. The error is significant because it misrepresents the timeline by four years and affects the narrative logic of the post-war section.
-
“A Million Punch Cards” → “Half a Million Punch Cards” (or hedge): The research file identifies “half a million” as the more precisely sourced figure. The blog uses “a million” without qualification, which the research file warns against. Either change the heading and body text, or add a note that some sources say a million.
Priority 2 — Accuracy improvements worth making
- T-5 vs T-6: When Feynman “took over the IBM group,” it was T-6, not T-5. Consider adding one sentence clarifying that the desk-calculator group was T-5 and the IBM tabulating group became the separate T-6.
Priority 3 — Minor / low-stakes
- Post-doctoral vs graduate student ambiguity: The blog’s phrasing (“post-doctoral work under Oppenheimer”) matches the majority of sources and the research file. No correction needed unless you want to add a caveat matching the research file’s uncertainty note.
NOTES ON THE FEYNMAN QUOTE
The blog quotes Feynman as saying Frankel had caught “the computer disease – the delight you get from seeing how much you can make a machine do.”
The research file contains the extended verbatim transcript from the Caltech Archives (“Los Alamos From Below”). The blog’s condensed version is a paraphrase of the spirit of the passage, not a direct quotation. The actual Feynman text does not contain the phrase “the delight you get from seeing how much you can make a machine do” as a direct quote — that is a paraphrased gloss on his longer description. The actual quote is the longer passage in the research file.
Assessment: The blog presents a condensed version in quotation marks that is not the verbatim Feynman text. Consider either using the full verbatim passage (as in the research file) or removing the quotation marks and making the condensed version clearly a paraphrase. As it stands, the quoted phrase “the delight you get from seeing how much you can make a machine do” may not be Feynman’s exact words.
Fix: Either quote the actual Feynman passage verbatim (the longer text in the research file), or rephrase as: Feynman diagnosed it as “the computer disease” — describing it in his own talk as the irresistible urge to see how much you can make a machine do.