"That nothing made of matter can transmit or process information faster than 1047 bits per gram per sec may seem of small practical importance. In fact, many of the processes that have been proposed for machines with artificial intelligence require transmissions far in excess of this limit. Examples are given to show that large-scale processes of combinatorial richness run into the limit only too easily.
Not only are our machines so restricted, but the scientist's brain, made of matter, is also so restricted. Thus, our personal knowledges, our philosophies, and our science are also limited to the same degree.
Some of its consequences in science are discussed. If our science is to be realistic, our theories must be structured so that this limit becomes an integral part of them. A suggestion is made of one way in which this incorporation might be achieved." 13