COPElAND | 209
the Princeton design. Apart from Kilburn’s idea of using subtraction as the single basic arith-
metical operation, Baby’s instruction set was virtually identical to what was being proposed at
Princeton:^78
Princeton. In the von Neumann plan the machine consists essentially of a memory and a static
accumulator. All transfers from the memory to the accumulator are essentially additions. The
orders used are of the following type:
(1) Clear A This clears the accumulator.
(2) x to A Adds the number in position x in the memory to the contents of the
accumulator.
(3) A to x Transfers the number in the accumulator to position x of the memory.
(4) C to x Transfers control to x; i.e., the next order to be obeyed is in position x in the
memory.
(5) CC to x Conditional transfer of control; i.e., control is transferred to x in the memory if the
number in A is negative. Otherwise, control obeys consecutive orders in the memory.
Baby’s originality certainly did not lie in its logical design, but in its cathode-ray tube mem-
ory and its electronic engineering. Baby is regarded as a British triumph, a world first for British
computing. Yet one of computing’s greatest ironies is that, thanks to Good and his Princeton-
based instruction set, the logical design of Baby was imported from America—seemingly with-
out Kilburn or Williams ever realizing that this was so.
Von Neumann and ACE
Von Neumann also had some influence on Turing’s design for ACE. Von Neumann’s first paper
on computer design, now known universally as ‘First draft of a report on the EDVAC’, was cir-
culated as early as June 1945 and was widely read and used.^79 Turing’s great friend Robin Gandy
recalled that Turing ‘became excited’ when he first read von Neumann’s paper.^80 He studied it
closely—although, with his focus on squeezing as much speed as possible from the hardware,
he went on to design a very different, decentralized, computer. Turing expected the readers of
his own design paper ‘Proposed electronic calculator’ (completed later the same year) to be
familiar with the ‘First draft’, and he recommended that his paper be read ‘in conjunction with’
von Neumann’s.
In fact, the design that Turing set out in ‘Proposed electronic calculator’ was much more
concrete than the proposals in the von Neumann paper, where EDVAC was described at a high
level of abstraction: von Neumann hardly mentioned electronics at all. Huskey, the engineer
whose job it was to draw up the first detailed hardware designs for EDVAC, said that he found
von Neumann’s paper to be of ‘no help’.^81 Turing’s design paper, on the other hand, gave detailed
specifications of the hardware and also included sample programs in machine code. Turing was,
though, perfectly happy to borrow some of the more elementary material in von Neumann’s
report. For example, his diagram of an adder is essentially the same as von Neumann’s dia-
gram (Figure 10 of Turing’s ‘Proposed electronic calculator’ shows an adder identical to the
one depicted in Figure 3 of von Neumann’s ‘First draft’), although in general Turing’s logic
diagrams—which set out detailed circuits for the arithmetical parts of the calculator and also