von Neumann documentary clip

Brian L. Stuart blstuart at bellsouth.net
Thu Feb 26 11:50:47 CST 2015


> I was alluding to the never-ending dispute/question as to
> the origins of the stored-program concept.
 
The thing to remember about this question is that in January 1944,
Eckert wrote a short document titled "Disclosure of Magnetic
Calculating Machine" that discussed using disks or drums for
primary storage reflecting initial discussions related to the EDVAC.
In it, he includes the statement:

"If multiple shaft systems are used a great increase in the
available facilities and for allowing automatic programming
of the facilities and processes involved may be made, since
longer time scales are provided.  This greatly extends the
usefulness and attractiveness of such a machine.  This
programming may be of the temporary type set up on alloy
discs or of the permanent type on etched discs."

How much earlier than January 1944 this idea of programming
was floating around the Moore School is not known, but it's
clear that at least by early 1944, there was a recognition that
storing programming on the same media as data had substantial
advantages.  There is every indication that von Neumann
first learned of the work at the Moore School in July 1944 at
a chance encounter with Herman Goldstine, when Goldstine
told him of the ENIAC.  At least by August 1944, von Neumann
visited the Moore School to see the ENIAC, at which time
they were testing a two-accumulator configuration.

The bottom line is that there's not much question that by the
time von Neumann was involved with the EDVAC project,
about which his "first draft" was written, the idea of the stored
program was already established at the Moore School.

However, the situation is not as simple as this.  The idea of
the stored program would not have been in any way alien
to von Neumann, and it would not surprise me at all if he
did make significant contributions to the conversation about
what form it would take.  This is because von Neumann was
well aware of Turing's "On Computable Numbers."  Remember
that Turing did his PhD at Princeton after the publication of
that paper.  Although the IAS isn't really a part of the university,
there can be no doubt that there was a lot of contact among
people like Church, Turing, Godel, and von Neumann.  Further,
in von Neumann's examinations of cellular automata, he
specifically mentions the Turing Machine.

We also have no record of when Turing first started thinking
about embodying the universal machine of his paper in physical
form.  He didn't really write much of anything about that during
the war.  However, because he moved so quickly into working
with NPL after the war, it seems very likely that he was already
thinking about it at Bletchley.

Ultimately, trying to assign an individual and a name to the
concept is unproductive.  The ideas themselves are much
more interesting and important than credit.

BLS



More information about the cctalk mailing list