Slashdot Mirror

← Back to Stories (view on slashdot.org)

Krawtchouk's Mind

Posted by Hemos on from the piecing-together-the-beginning-of-computers dept.

A reader writes: "Central Europe Review is running an article on a gulag-condemned Soviet scientist whose contribution to the first computer is virtually unknown because of the Cold War mentality that infected much of society on both sides of the Iron Curtain. The story tells of how in 1937, American digital computer pioneer John Atanasoff came across a Myhailo Krawtchouk paper on a new method for finding approximate solutions to differential equations. Atanasoff tried sending a letter to him, but received no response. Krawtchouk had been attainted for giving a favorable review of the work of "enemies of the people" and shipped to Siberia for 20 years of gold mining, where he died four years later. Krawtchouk's biography gives a more detailed account of how Krawtchouk was labeled a "Polish spy" and "Ukrainian nationalist," stripped of his Academy of Sciences membership, and forced to sign a confession -- that he later retracted -- under torture and threats upon his family. "

2 of 220 comments (clear)

  1. Actually... by LeoDV · · Score: 5, Interesting

    I think the first computer was invented in 1936 by a German scientist, Konrad Zuse, who later had to flee to Switzerland because of the war... At least that't About.com claims.

    You know, it's really funny how things can be invented in several places at the same time... Like the modern guitar as we know it was come up with in China, the Middle East and Spain at the exact same times (and not chronologically, implying that the invention would have traveled)... Or how Pythagores, Zarathustra, Buddha and Lao-Tse, who each pioneered philosophy in their own continent, were contemporaries.

  2. Re:Not Turing Machines? by pmc · · Score: 4, Interesting
    Colossus was a Turing machine, just. It had the (rough) equivalent of "computed goto" where the result of an operation could determine what instruction was executed next, which is enough.


    This, its "Turingness", came about almost by accident - in breaking the Lorenz codes it ran a computation step where it worked out some property of captured cipher text against generated enciphering text. This produced a potential deciphered text, which an operater would look at to see if it made sense. The second generation machine was designed to calculate some statistical properties of the text, which could tell resonably well if it had been broken properly. It was when they were building the capability of doing this that the computed goto snuck in (which gave them the ability to do conditional branching), and the equivalent of "if Text looks like German then Stop".


    After the war the runners of the machine tried to program it to do base 10 arithmetic, but clock speed was against them, so it never quite worked (not that they spent a lot of time on it as shortly after the war almost all the colossi were scrapped).


    It was an odd machine - extremely fast at what it was built to do, with the bonus that it could do anything (given enough vacuum tubes!). Different to the ABC and Zuse machines in that these were non-Turing machines (I think that's true of the Atanasoft-Berry machine too - having looked at the information available it doesn't seem Turing complete, but it is fairly sketchy).