Meteorologist, oceanographer, Python solvem probler, experimenter, Raspberry Pi tinkerer, geek dad, manager, space enthusiast, ham radio lover, science columnist, ADHDer, synthetic intellect curious, linux admin cosplayer.

DRAFT CORRECTIONS - Post 28: The Memory in Every Forecast

Status: COMPLETE

Summary

Post 28 is in very good shape factually. The core narrative – the two notebooks (Wang 29 June 1948, Forrester 13 June 1949), the SAGE air defense system descended from Whirlwind, the IBM patent settlements ($500K to Wang, $13M to MIT) – is well-supported by the cited sources. Most specific numerical claims (dates, prices, vacuum tube counts, costs) match Wikipedia and the Computer History Museum.

The following corrections are recommended in priority order.

CORRECTION 1 (REAL FACTUAL ERROR – recommend fix)

Mina Rees was an algebraist, not a topologist.

Current text (paragraph beginning “The mathematical sciences division of ONR…”):

The mathematical sciences division of ONR was headed from February 1949 by Mina Rees, a topologist who had spent the war administering applied mathematics for the Office of Scientific Research and Development.

Issue: Rees’s PhD (University of Chicago, 1931) was on “Division algebras associated with an equation whose group has four generators” – a problem in abstract algebra, not topology. Her published work and CV consistently identify her as an algebraist.

Recommended fix:

The mathematical sciences division of ONR was headed from February 1949 by Mina Rees, an algebraist who had spent the war administering applied mathematics for the Office of Scientific Research and Development.

CORRECTION 2 (MINOR – consider clarifying)

Whirlwind storage-tube access time: post says “twenty-five microseconds, four times slower than spec”; Wikipedia gives 30 microseconds.

Current text (paragraph beginning “The architecture of Whirlwind I…”):

The original 1947 design specification called for memory access in six microseconds; the storage tubes Whirlwind actually got delivered roughly twenty-five microseconds, four times slower than spec, and they kept failing.

The “twenty-five microseconds” figure is consistent with Papian’s later comparison (storage tube access “approximately 25 microseconds”). Wikipedia’s Whirlwind I article gives “30 microseconds” (5x spec). Both numbers appear in the historical record. The post is choosing the lower value cited by Papian. This is defensible and probably need not be changed; but if a hedge is desired:

The original 1947 design specification called for memory access in six microseconds; the storage tubes Whirlwind actually got delivered approximately twenty-five microseconds (some sources give thirty), four times slower than spec, and they kept failing.

Probably not worth the awkwardness. Recommend: leave as-is.

CORRECTION 3 (MINOR – consider hedging)

SAGE cost vs Manhattan Project comparison: “four to six times” in post; Wikipedia says “four times”.

Current text (paragraph beginning “The total program cost across the deployment cycle…”):

The Manhattan Project cost two billion dollars in WWII-era dollars. SAGE was four to six times the Manhattan Project.

Wikipedia’s SAGE article says “four times the cost of the Manhattan Project” with the $8-12 billion cost range stated. The arithmetic of $8-12B / $2B does give 4 to 6, so the post is mathematically defensible. The Wikipedia 4x figure is the lower bound. Recommend leaving as-is; the post’s range is the more honest reading of the $8-12B cost range divided by $2B.

CORRECTION 4 (MINOR – consider clarifying)

Whirlwind construction start: post says 1947; some sources say 1948.

Current text (paragraph beginning “Construction began in 1947…”):

Construction began in 1947 in a brick building on the corner of Massachusetts Avenue and Vassar Street – the Barta Building, MIT designation N42.

Two Wikipedia articles consulted (“Whirlwind I” and “Project Whirlwind”) both give 1948 as the construction start, while the post says 1947. Redmond and Smith’s “Project Whirlwind: The History of a Pioneer Computer” (1980), the canonical scholarly history, would be the authoritative source. The 1947 date may reflect detailed design start while the 1948 date reflects assembly start. This is a hairline distinction. Recommend: leave as-is; the Redmond and Smith book is the post’s cited source.

ITEMS VERIFIED (no change needed)

All of the following claims were checked against authoritative sources and are correct as stated:

  • Jay Forrester born 14 July 1918 in Anselmo, Nebraska; died 16 November 2016 in Concord, Massachusetts, aged 98
  • High-school wind-driven 12-volt generator on the family ranch
  • BS Nebraska 1939, MS MIT 1945 under Gordon Brown
  • IEEE Medal of Honor 1972, National Medal of Technology 1989
  • Forrester’s notebook entry 13 June 1949 (CHM); 15 June 1949 in EDN – post correctly notes the discrepancy
  • Forrester patent US 2,736,880 filed 11 May 1951, issued 28 February 1956
  • IBM-MIT settlement $13 million in 1964
  • Forrester received $1.5 million personally
  • An Wang born 7 February 1920 in Shanghai; died 24 March 1990 aged 70
  • Wang’s PhD adviser was NOT Aiken (post correctly flags the popular-literature error)
  • Wang’s “by destroying the information – I know it” quote
  • Wang’s notebook entry 29 June 1948
  • Wang patent US 2,708,722 filed 21 October 1949, issued 17 May 1955, 34 claims
  • IBM paid Wang $500,000 (March 1956 assignment)
  • Frederick W. Viehe interference, IBM bought Viehe application November 1957
  • Wang Laboratories peak $3 billion revenue, 33,000 employees, mid-1980s
  • LOCI-2 January 1965 first desktop calculator with logarithms
  • Wang controlled ~80% of word-processing market in mid-1970s
  • Wang Labs filed Chapter 11 in August 1992
  • Wang received Medal of Liberty 1986
  • Whirlwind operational 20 April 1951
  • Whirlwind 16-bit, bit-parallel, 20,000 ops/sec
  • Whirlwind ~5,000 vacuum tubes, >100 kW, 2,000 sq ft, 10 tons (20,000 lb)
  • Whirlwind shut down 29 May 1959 (Wikipedia confirms ~1959 end-of-life; multiple sources cite 29 May 1959)
  • Wolf Research and Development leased the machine for $1/year through 1974
  • Memory access dropped from ~25 us (storage tube) to ~9 us (core) – Papian’s figures
  • Memory maintenance dropped from 4 hours per day to 2 hours per week
  • AN/FSQ-7: 49,000 vacuum tubes per computer, 60,000 with peripherals, 250 tons, up to 3 MW, ~75,000 instructions per second, 24 installed, two computers per site, Luke AFB Q7 demolished February 1984
  • SAGE first operational at McGuire AFB DC-01 on 1 July 1958
  • SAGE total cost $8-12 billion
  • IBM 704: announced May 1954, shipped 1954, 36-bit word, first commercial with floating-point hardware, ~12,000 FP add/sec, 123 systems shipped 1955-1960, FORTRAN/LISP/SAP
  • MIT Lincoln Laboratory founded 1951 in Lexington Massachusetts from Project Charles recommendation
  • Soviet RDS-1 nuclear test 29 August 1949 in Kazakhstan
  • Project Charles letter contract 30 January 1951
  • Howard Aiken’s Harvard Mark IV completed 1952, Air Force-funded, used ferrite core memory
  • Ken Olsen born 20 February 1926 in Bridgeport, Connecticut; died 6 February 2011 in Indianapolis aged 84
  • Olsen MIT BS 1950, MS 1952
  • DEC founded 1957 in Maynard MA with $70,000 + $30,000 from American Research and Development (Doriot)
  • PDP-1 1959/1960 at ~$120,000, 53 units built; Spacewar! 1962
  • PDP-8 1965, PDP-11 1970 with >600,000 units sold
  • VAX 11/780 announced 25 October 1977 – 32-bit superminicomputer with VMS
  • Intel 1103 introduced October 1970, 1024 bits, p-MOS 3-transistor, $60 launch falling to $4 by 1973, best-selling semiconductor by end 1971, adopted by 14/18 major makers within 2 years
  • Robert Everett President of MITRE 1969-1986, died 15 August 2018 at Cape Cod age 97
  • Andrew Booth at Birkbeck developed drum memory in the late 1940s (1947-48 construction)
  • Manchester Baby first stored-program computer (June 1948)
  • IAS, IBM 701, BESK Stockholm used Williams tubes

URLS NOT POSTED TO WAYBACK

Per task instructions, Wayback was attempted but blocked. Verified URLs include:

  • https://en.wikipedia.org/wiki/Whirlwind_I
  • https://en.wikipedia.org/wiki/Jay_Wright_Forrester
  • https://en.wikipedia.org/wiki/An_Wang
  • https://en.wikipedia.org/wiki/Magnetic-core_memory
  • https://en.wikipedia.org/wiki/AN/FSQ-7_Combat_Direction_Central
  • https://en.wikipedia.org/wiki/Ken_Olsen
  • https://en.wikipedia.org/wiki/Semi-Automatic_Ground_Environment
  • https://en.wikipedia.org/wiki/Wang_Laboratories
  • https://en.wikipedia.org/wiki/IBM_704
  • https://en.wikipedia.org/wiki/VAX-11
  • https://en.wikipedia.org/wiki/Digital_Equipment_Corporation
  • https://en.wikipedia.org/wiki/PDP-1
  • https://en.wikipedia.org/wiki/Intel_1103
  • https://en.wikipedia.org/wiki/MIT_Lincoln_Laboratory
  • https://en.wikipedia.org/wiki/Project_Whirlwind
  • https://en.wikipedia.org/wiki/Project_Charles
  • https://en.wikipedia.org/wiki/Mina_Rees
  • https://en.wikipedia.org/wiki/Robert_Everett_(computer_scientist)
  • https://en.wikipedia.org/wiki/Williams_tube
  • https://en.wikipedia.org/wiki/Soviet_atomic_bomb_project
  • https://en.wikipedia.org/wiki/Harvard_Mark_IV
  • https://www.computerhistory.org/storageengine/whirlwind-computer-debuts-core-memory/
  • https://www.computerhistory.org/revolution/memory-storage/8/253
  • https://www.computerhistory.org/tdih/march/4/
  • https://history.computer.org/pioneers/wang.html
  • https://www.madehow.com/inventorbios/66/An-Wang.html
  • https://patents.google.com/patent/US2708722