time he was creating and refining his
abstract design of EDVAC.
More interestingly, the lecture gives
us a sense of the contexts in which von
Neumann did, and did not, find Turning machines relevant. They appear in
a tutorial role, introducing an abstract
model of computation to make a point
about the equivalence of different possible computers with respect to the
computations they could perform “
given all eternity.” That resembles the use
of the concept by modern computer
science popularizers, though its coupling with discussion of neurons and
the creation of machines able to learn
reminds us that for von Neumann, as
for other founders of cybernetics, enthusiasm for theories of computation
was bound up with much grander visions. The example of Mooers suggests that von Neumann successfully
communicated this cluster of ideas to
at least one recipient.
While Turing machines do not appear in von Neumann’s work on computer architecture and logical control,
they are prominent in his later work
toward a “general and logical theory of
2 His lecture at the 1948 Hix-on symposium showed, in the words of
historian William Aspray, that he “had
in mind the McCulloch-Pitts and Turing contributions as the foundation
for his new theory.” Von Neumann
developed these ideas further over the
next few years, culminating in a major
work on cellular automata and a series
of lectures on the computer and the
brain, both of which remained unfinished at the time of his death.
Von Neumann freely acknowledged
the contribution of Turing’s 1936 pa-
per to his work on automata theory but
made no such connections in his dis-
cussion of computer design. Now that
we know how clearly and concisely von
Neumann could explain the universal
Turing machine in 1945, its absence in
his other reports, lectures, and letters
of the 1945–1946 period speaks loud-
ly, like the dog that Sherlock Holmes
realized had failed to bark during the
night. Von Neumann deployed it in an
introductory lecture, but not for other
purposes. He made no reference to any
feature of the Turing machine in the
First Draft or his other 1945–1946 de-
tailed writings on computer architec-
ture and instruction sets. Neither did
he mention Turing’s demonstration
of universality (or related work by Post
and Church) in his other speeches and
letters lobbying for the construction of
EDVAC-like computers.m In contrast
the neuron notation and cybernetic
language did make it into the First
Draft, though as a convenient way to
describe digital logic rather than as a
source of architectural inspiration. The
abstraction from constraints of space
and time that eventually made Turing
machines so useful for computer sci-
entists looking to lay theoretical foun-
dations for their new field made them
irrelevant to people trying to design the
first real electronic computers.
m For example, a lengthy 27 August 1945 report
to the Navy Ordnance Department on “
Computer Services” (HHG-APS, box 9). See Priestley14
(2018, 92-3) for a discussion of the term JvN used
instead, namely “all-purpose,” and its difference from Turing’s notion of “universal.”
1. Akera, A. Calculating a Natural World: Scientists,
Engineers, and Computers During the Rise of U. S. Cold
War Research. MIT Press, Cambridge, MA, 2007.
2. Aspray, W. John von Neumann and the Origins of
Modern Computing. MIT Press, Cambridge, MA, 1990,
3. Birkhoff, G. Computer Developments 1935–55, as Seen
from Cambridge, USA. In A History of Computing in the
Twentieth Century, N. Metropolis, J. Howlett and G.-C.
Rota, Ed., Academic Press, New York, 1980, 21–30.
4. Church, A. Review of Turing “On Computable
Numbers…” Journal of Symbolic Logic 1 (1937), 42–43.
5. Davis, M. Computers and Unsolvability (Dover, 1958).
6. Davis, M. Engines of Logic: Mathematicians and the
Origin of the Computer. Norton, New York, NY, 2001.
7. Galison, P. The Ontology of the Enemy: Norbert Wiener
and the Cybernetic Vision. Critical Inquiry 21, Autumn
8. Haigh, T. Actually, Turing did not invent the computer.
Commun. ACM 57, 1 (Jan. 2014), 36–41.
9. Haigh, T., Priestley, M., and Rope, C. ENIAC In Action:
Making and Remaking the Modern Computer, 2016.
10. Hodges, A. Alan Turing: The Enigma. Simon and
Schuster, New York, NY, 1983.
11. Kline, R. The Cybernetics Moment, Or Why We Call Our
Age the Information Age. Johns Hopkins University
12. McCulloch, W.S. and Pitts, W. A logical calculus of
the ideas immanent in nervous activity. Bulletin of
Mathematical Biophysics 1943, 115–133.
13. Mooers, C. N. The computer project at the Naval
Ordnance Laboratory. IEEE Annals of the History of
Computing 23, 2 (Apr.–Jun. 2001), 51–67.
14. Priestley, M. Routines of Substitution: John von
Neumann’s Work on Software Development, 1945-
1948. Springer, Cham, Switzerland, 2018.
15. Priestley, M. and Haigh, T. The media of programming.
In Exploring the Early Digital. Springer, Cham,
Switzerland, 2019, 135–158.
Thomas Haigh ( firstname.lastname@example.org) is a Professor of History
at the University of Wisconsin—Milwaukee and Comenius
Visiting Professor for the History of Computing at Siegen
University in Germany. He has written widely on the history
of computing—read more at www.tomandmaria.com/tom.
Mark Priestley ( email@example.com) is Research Fellow
at the U.K.’s National Museum of Computing, Bletchley Park,
U.K. See www.markpriestley.net for more information.
Copyright held by authors.
his return to Washington, Mooers re-
quested a photostat copy of Turing’s pa-
per from the library but, unlike the First
Draft, he never recorded reading it.k
Mooers’ experience of von Neu-
mann spontaneously launching into
similar material in October builds our
confidence in a mid-1945 date for prep-
aration of the lecture text. Von Neu-
mann gave many lectures and assisted
many computing groups, undoubtedly
recycling material between then. He
obviously hadn’t prepared the text
specifically for this occasion, not least
because it is coy about the details of
projects he had already discussed with
Mooers. The material von Neumann
followed it with on that occasion, de-
scribing his new approach to logical
control, would have been the obvious
content for a “lecture 4” to solve the
problem posed at the start of lecture 3.
Mooers did not reference Turing in
his efforts to design a computer and
instruction set. In March 1946, however, he did track down the work of McCulloch and Pitts and used the neuron
notation and cybernetic terminology
to sketch out plans for a “thinking machine.” He discussed it with Pitts that
summer, noting in his diary that he told
him “how by use of a magnetic edvac
type machine a device could be made
which would trace through a nervous
net. Showed him how a ‘Turing Machine’ (which it is) can be elaborated
to do the job.”l Not long after Mooers’
boss, John V. Atanasoff, ordered him to
work on more useful matters.
Even before learning of the lecture
notes on “High Speed Computing” we
believed that von Neumann was almost
certainly familiar with Turing’s 1936
paper prior to beginning work on the
First Draft. His later remarks showed
that he fully understood its use as an
abstract model of computation, and
there was no reason to believe he suddenly developed this understanding
after 1945. The new evidence confirms
that von Neumann had read and fully
understood Turing’s paper around the
k “NOL Notebook 3,” in box 28 of the Calvin R.
Mooers Papers, Charles Babbage Institute,
l Lengthy extracts from Mooers’ diary are in box
28 of his papers at CBI.