Herman Goldstine and the IAS Machine: Unveiling the Modern Computer
How one man helped shape the development of modern computers
Sometimes even men from the future need a little help.
It’s quite hard to escape Oppenheimer at the moment. So it’s a good opportunity to go back to the 1940s and consider the contribution of someone who knew and worked with members of the Manhattan project.
I recently came across a film tribute to John von Neumann, made in 1966 by the Mathematical Association of America. Grainy and noisy in its current YouTube incarnation, it’s nonetheless riveting, with footage of von Neumann himself and contributions from friends and colleagues on the Manhattan Project including Stanislaw Ulam, Eugene Wigner and Edward Teller.
The whole film is worth watching, but one moment grabbed my attention. Two-thirds of the way through, one of the contributors visits a store-room in the Smithsonian Institution in Washington and pulls the cover off a large object stored there.
The object is the IAS Machine and the man pulling off the covers is Herman Goldstine. The IAS Machine was an early ‘stored program’ computer, named after the Institute for Advanced Studies at Princeton, where it was built over 1946-51 under von Neumann’s supervision.
The IAS Machine was organised according to the approach von Neumann had first described in his 1945 paper ‘First Draft of a report on the EDVAC’. It was this report that first set out what would become known as the von Neumann architecture, which, of course, utterly dominates computing today.
The EDVAC (Electronic Discrete Variable Automatic Computer) referred to in the ‘First Draft’ report was intended to be the successor to the ENIAC (Electronic Numerical Integrator and Computer) the world’s first programmable, electronic digital computer1
And it was Herman Goldstine who revealed von Neumann’s ideas to the world.
Goldstine had studied maths at the University of Chicago and then went on to teach at the University of Michigan. He joined the US Army in 1942 and was soon employed at the Aberdeen Proving Grounds in Maryland working on calculations to support the aiming of artillery.
There Goldstine made his first vital contribution to the development of the modern computer, by convincing the U.S Army to fund development of the ENIAC.
He first talked with von Neumann in August 1944 whilst waiting for a train, although he’d previously heard him lecture at a symposium in Michigan.
Goldstine later recalled:
I had never met this great mathematician. It was therefore with considerable temerity that I approached this world-famous figure, introduced myself, and started talking. Fortunately for me von Neumann was a warm, friendly person who did his best to make people feel relaxed in his presence. The conversation soon turned to my work. When it became clear to von Neumann that I was concerned with the development of an electronic computer capable of 333 multiplications per second, the whole atmosphere changed from one of relaxed good humor to one more like the oral examination for a doctor’s degree in mathematics.
Goldstine was soon working closely with von Neumann. He would be part of a small group including J. Presper Eckert and John Mauchly, who would work together on the development of ENIAC and explore ideas coming out of its development.
Out of these explorations came von Neumann’s handwritten ‘First Draft’ report. It was Goldstine who had the report typed up and then helped to distribute it, actions that would have far-reaching consequences.
Goldstine’s comments in the film neatly sum up both the importance, and simplicity, of the ideas in von Neumann’s paper (my emphasis):
Narrator:
During the period he was working in Los Alamos von Neumann became intensely interested in computers Dr Herman Goldstine, now of the IBM corporation, was one of his earliest collaborators in this work. In the spring of 1966 Dr Goldstine was invited to visit a warehouse of the Smithsonian Institution in Washington where the original Princeton computer is stored. He was accompanied by Dr Utta Mertzebach of the Smithsonian staff.
Goldstine:
Well these things would have fascinated Johnny. He used to love toys. When he worked on the on his automata theory at the beginning, he started since he had to make three dimensional objects, he brought the biggest tinker toy set that he could find, and he made the objects out of these until somebody we showed him one time how to make these two dimensionally. So he was able to give the tinker toys to Morgenstern's son.
Mertzebach:
Oh really?
Goldstine:
Yes. Well as Johnny would have said, mercy here it is. All these years. May I move the curtain? Sure be glad to let's see what it looks like, if I can do it. Very good, I succeeded.
Yes. All right thank you.
This machine was the concrete embodiment of von Neumann’s very great ideas and contributions which he has made to the electronic computer field.
In 1946 Johnny asked me if I would join him at the end of the war in Princeton and help him to carry out in concrete form the ideas which he had been working on in 1944 and 1945.
Of course I jumped at the chance. We rushed to Princeton and got started the machine that eventuated from that is the one you see here and it contains essentially those things which the modern computer has in it, although in somewhat primitive form.
This machine has stored program concept as its major feature and that in fact is the thing which makes the modern computer revolution possible. The older machines required one to clumsily perform hand pluggings of connections which took hours indeed days. It meant that programming was an art, in fact a very black art and furthermore it meant that the total number of instructions one could write were comparatively small. This new concept has been carried so far today that programs are written involving tens of millions of instructions whereas in those days of course nobody dreamed of such complexity. But Johnny's idea made this basically possible.
What is the stored program concept? Well it's the notion that you can describe in a finite number of words in fact a fairly small number of words in a fairly simple language exactly and unequivocally the description of a problem. And that this description is then translated into binary digits and stored in the memory of the computer exactly as numbers are stored. This was the discovery by Johnny.
You may say what's so remarkable about that. Well the only thing I can tell you an answer to that is it's just like the wheel. What's so remarkable about the wheel when you look at it. You can't conceive how anybody would not have known that there was one. Indeed it must have been that the moment somebody mentioned the wheel or somebody mentioned the stored program everybody around us obviously knew that this was the way to do it and in fact we accepted it immediately.
It was not one of these inventions or discoveries which is enormously complicated and few people can understand. It's tremendously simple. It immediately hit hits a person and he knows that's it.
It’s notable that Goldstine unequivocally attributes the idea of the stored program computer to von Neumann. J. Presper Eckert, who had been chief engineer on the ENIAC project, would later dispute that attribution.
After the war von Neumann and Goldstine would work together at Princeton’s Institute for Advanced Studies, where Robert Oppenheimer was director. There they developed the IAS Machine.
Construction of the IAS Machine took longer than expected. William Aspray, in John von Neumann and the Origins of Modern Computing, writes:
The engineering challenges, especially in the design and construction of an acceptable primary memory unit, were more difficult than originally anticipated. The computer took almost twice as long as expected, nearly six years to complete.
This delay would mean that the IAS Machine wasn't the first electronic computer built using von Neumann’s approach.
von Neumann’s ‘First Draft’ paper was distributed widely, including across the Atlantic. In Manchester in England, the ideas were central to the design of the ‘Manchester Baby’ which ran its first program in June 1948.
The Manchester Baby was followed by the EDVAC itself, referred to in the title of von Neumann’s report, which was delivered in 1949, but only became operational in 1951.
The IAS Machine was highly influential, though, and plans for its design were distributed widely, partly due to von Neumann’s insistence that advances be kept in the public domain. Many computers were built using the IAS Machine as a starting point, including the first IBM Mainframe, the IBM 701.
Goldstine himself went on to become IBM’s technical director, overseeing the development of many of the IAS machines successors.
Thanks to the modern ‘von Neumann machines’ at YouTube, we can watch him unveil von Neumann’s own embodiment of the modern computer again, twenty years after he played a key role in making the world aware of von Neumann’s ‘tremendously simple’ invention.
And to return to Oppenheimer, the photo with Oppenheimer and von Neumann at the top of this post, has them posing in front of, of course, the IAS Machine.
But we can put Goldstine back in the picture, where he surely belongs, in another photo with Oppenheimer and von Neumann, again with the IAS Machine in the background.
Herman Goldstine played a key role in the invention the modern computer. Twice he recognised the importance of the ideas of others, and then helped to ensure that those ideas came to life.
If you value The Chip Letter, then please consider becoming a paid subscriber. You’ll get additional weekly content, learn more, and help keep this newsletter going!
Note that the first version of the ENIAC doesn’t count as a stored program computer although it was programmable. It was altered so as run a program stored in manually altered rotating switches and this version ran before the Manchester Baby making it - according to some - the first operational stored program computer. The later version wasn't a von Neumann architecture machine as we use the term today.
Love von Neumann! 🤩