Wednesday, June 8, 2016

Picture of One of the World's First Computers

at JPL in 1953. Each woman is a switch. The program was a set of instructions each was given: receive the number from the woman to your left, divide by seven, and hand the result to the woman behind you. That sort of thing.


  1. The computers' jobs were nowhere near as simple as "receive the number from the woman to your left, divide by seven, and hand the result to the woman behind you."

    They were solving equations and doing numerical integration, interpolation and extrapolation. They computed rocket and spacecraft trajectories and orbits. Besides pencil and paper, they used slide rules, books of mathematical tables -- many of them computed under Works Progress Administration sponsorship during the Depression -- and electro-mechanical Friden desk calculators. Many later programmed punched-card calculating machines and electronic computers. Some of them had bachelor's degrees in, or with minors in, math or a hard science or engineering.

    See, for example, Rise of the Rocket Girls, by Nathalia Holt.

    Holt has a serious technical ignorance problem; her work desperately needed a competent technical review it did not get. She writes nonsense like (page 176) "While Helen could translate the punches in the cards into the equations they stood for, the IBMs could not. She had to run all the cards through a special machine called a compiler." Urk? The compiler was a "special machine"??? In fact, as all know, a compiler is Just Another Program that runs on a general-purpose machine, usually (not quite always, but absolutely certainly in the case of FORTRAN on an IBM 1620 or 704/709/7090/7094 at JPL) on the same machine on which the compiled program will run.

    Nevertheless, I doubt she seriously botched her reporting of the academic histories of the women she wrote about.

    1. Sure, and I didn't mean to imply they only did arithmetic; only that they executed what today would be single steps in a program. Feynman ran such a computer room earlier at the Manhattan project.

    2. Milutin Milankovitch, a Serbian scientist ran one much earlier. His lady "computers" (We should refer to electronic computers as females like we do with ships) helped him describe orbits and axial tilts over tens of thousands of years. See a youtube with Dan Britt about 24.19/55.49 called "Orbits and Ice Ages" (h/t Jerry Pournelle)

  2. Seems to me their role was more like the role of cores in a multi-core processor, or of nodes in a modern cluster computer, than of executing "what today would be single steps in a program". Each had her own multi-step program to run and her own local cache memory, and communicated with other cores or nodes over a local network ("sneakernet").



  Teodorq sunna Nagarajan has joined an expedition tasked with evaluating the"serving tray" as a suitable observation post for the...