CARPENTER DORAN | 227
it in Germany, in the form of his Plankalkül language, without having read Turing’s paper. The
next four ideas in Table 22.1 were also to be found in the EDVAC report which Turing had
recently read. This might suggest that they derived indirectly from various American sources
via von Neumann. However, since binary implementation using standard electronic elements
had been used in Colossus by Tommy Flowers, it seems unlikely that Turing derived his knowl-
edge of this from American sources, and of course the idea of binary working was already
there in ‘On computable numbers’. Colossus also had an elementary conditional branch, as Max
Newman emphasized. The circuit notation in the ACE proposal derived (via von Neumann)
from the famous American pioneers of computational neurophysiology, Warren S. McCulloch
and Walter Pitts,^11 who in turn had cited Turing’s ‘On computable numbers’.
All the other concepts in Table 22.1 appeared in Turing’s report but not in von Neumann’s.
They were without published antecedents, and one should not overlook the fact that they pre-
ceded the influential Moore School lectures in Philadelphia by several months. We now know
about Turing’s experiences at Bletchley Park, and about his knowledge of Colossus and of the
work of Jacquard, Babbage, and Lovelace in the previous century, as well as his discussions with
Claude Shannon during the war (see Chapter 31). While it is very likely that some of Turing’s
ideas were not brand new in the ACE report, it nevertheless remains quite startling to find them
all in one place at such an early date. Other pioneers took a few more years to reach a similar
point.^12
what happened to Turing’s ideas?
Next we consider in turn the formative ideas in Table 22.1 that were apparently unique to Turing
in 1945. All of these resurfaced over the following fifteen years or so. The question is: how much
of that was rediscovery, and how much was unacknowledged reuse?
Turing’s descriptions of these ideas were not in the open literature. The 1945 ACE report,
mimeographed in a limited number of copies, was out of stock by 1948 and vanished from view
until 1972. The Pilot ACE was well known in itself, but with little mention of Turing’s contribu-
tions, although two reports by James Wilkinson on the Pilot ACE were widely circulated.^13 The
Cambridge computer design team (see Chapter 21), especially Maurice Wilkes, never admitted
to much influence from Turing—although Stanley Gill, for one, worked alternately on the Pilot
ACE and the Cambridge EDSAC, but in both cases after the main design choices were fixed.
The University of Manchester team mainly followed the EDVAC line (see Chapter 20).
Yet all of the ACE ideas showed up in later designs. We aim to give the flavour of Turing’s
insights and a feel for why they were so prophetic in 1946. It is difficult to convey the breadth
and depth of Turing’s originality in the ACE design without going into technical detail and
using modern jargon, but we have kept the technicalities to a bare minimum.
Several of Turing’s ideas were bound up with the notions of ‘words in memory’, ‘addresses’,
and ‘registers’. Computer memory is divided up into small pieces, known as ‘words’, and in the
case of ACE these consisted of thirty-two bits each, a word size still commonly used today.
The position of a given word in memory is simply a number (0 for the first word, 1 for the next
word, and so on) and this is called its ‘address’. When the content of a word in memory has to
be processed by a computer instruction it is often copied into a temporary storage device called
a ‘register’, which is part of the computer’s central processing unit. Registers in ACE each held
thirty-two bits.