CAmPBEll-kElly | 221
In a way, the demise of ACE vindicated Wilkes’s original objections to Turing’s design. Wilkes
argued that delay lines were an ephemeral technology, that true random-access memories
would eventually come along, and that with a conventional design it would be relatively easy
to replace one memory technology with another. Indeed, this is exactly what happened with
EDSAC 2, which was designed for a mercury delay-line memory but switched to a random-
access core memory midway through the project. This was not possible with ACE. Turing’s
viewpoint was undoubtedly coloured by the fact that he was blessed with the manipulative
skills of a first-class mathematician. He believed that the trade-off between a little extra effort
with programming was amply justified by the superior performance of the ACE design, though
he underestimated the difficulty ordinary users would have in programming the machine.
It is perhaps fair to say that Turing was right in the short term, and Wilkes was right in the
long run. But in the early 1950s the short term counted for a lot.
figure 21.2 The full-scale ACE in 1958.
© Crown Copyright and reproduced with permission of the National Physical Laboratory.