The future of automatic digital computers

Before discussing the future I want to talk about the past, which I find a source of disappointment. One of the things which prompted me to give the lecture its title was that a number of people who should know better have recently given tongue and pen to the statement that “new, second generation computers are with us” and that these machines are better than anything which has been thought of before. The first part of this lecture will attempt to disabuse you of this idea for, far from thinking that any second generation computer exists, I think that we are only just seeing the growing up of first generation computers. To justify this statement I remark that, in 1946-7, I worked with the late John von Neumann on the logical and physical design of computing machines. Von Neumann wrote, at the Institute for Advanced Study at Princeton, two reports [1, 2] on aims and objectives to which we may look to see the type of computer which was envisaged at that time. This computer was in fact a machine having, in retrospect, certain rather interesting characteristics. The most obvious of these is speed of operation and this was desired to be such that a 40 bit addition or subtraction would take about 10 micro-seconds. There is no computing machine commercially available in this country which achieves this addition time. The multiplication time of the von Neumann machine varied from 400 micro-seconds for the crudest scheme to 50 micro-seconds for a more sophisticated logical device which still made no use of the steam-roller electronics to be seen in at least one machine of the present day, which achieves a rather worse performance. So much for second generation speed!

View this article in the ACM Digital Library.