Herman Goldstine and the IAS Machine: Unveiling the Modern Computer (2023)
thechipletter.substack.com
thechipletter.substack.com
Right. This is, again, the memory hardware problem. Eckert had been building computing machines since the 1930s, and his programs were on cams. In 1949, IBM started selling a "memory unit" the IBM 941, that stored 16 10-digit numbers, using about a cubic meter of relays.[1] Multiple memory units could be plugged into an IBM Card Programmed Calculator. It was possible to store short programs in the relays.[1] So it was a very limited "stored program" machine.
It was recognized by workers in the field for years before the IAS machine that stored program computers were desirable, if only someone could build something to store the programs. Something with an affordable cost per bit. That was the limitation.
"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."
(You can quibble about whether Zuse or Turing described stored-program approaches earlier, but these were significantly more hypothetical / non-practical than the designs being discussed for the EDVAC. This is kind of a theme: there are many sideshows in early computing which were flirting with the right ideas, but the ENIAC and subsequent Moore school lectures stand out as the ancestors of everything that actually went anywhere. The cambrian explosion started there, which is why people get strung out about who gets the credit.)
Von Neumann and Goldstine are clearly at fault in not crediting Eckert and Mauchly at all in the 'first draft'. The precise role of the players in developing the concept is disputed though. The 'first draft' is the first complete description of a stored program computer - there is no comparable document from Eckert and Mauchly - and it's inconceivable that Von Neumann didn't contribute significantly to the design.
And if we are focused on practical designs, then Eckert and Mauchly weren't even the first to implement a working stored program computer using the approach set out in the first draft.
We can disagree whether distributing the 'first draft' was morally right. My personal view is that doing so and hence invalidating the patents was a huge net good.
I hear "this put the computer in the public domain" a lot. But it was patented, and that didn't stop the digital revolution one bit.
Eckert and Mauchly filed a patent (on ENIAC) and it was believed to be valid for the next 25 years. That doesn't mean anybody tried to create a monopoly. IBM quickly traded some punch card patent rights with the owner, Remington Rand, for the (still pending!) ENIAC patent. Once it was finally granted Remington Rand pressed for royalties from the other computer companies.
Evidently they asked for too much, and that fomented Honeywell's suit to invalidate the patent, which succeeded. That, technically, put the ENIAC in the public domain, but by then there were thousands of other computer related patents on the books. They annoy, especially the bad ones, but business goes on.