8/17 AN INTRODUCTION TO CYBERNETICS 



What was important in the argument about t/'s feedback to T 

 is that what U feeds back to T is highly correlated with what is in T, 

 for each U feeds back into the particular T that acted on it a step 

 earlier, and no other. The argument thus demands an accurate 

 treatment of the correspondences between the various T's and t/'s. 



The arguments of the previous few sections have shown that 

 though the matter can be treated in words in the simple cases (those 

 just considered), the attempt to handle complex cases in the verbal 

 form is likely to lead to intolerable complexities. What is wanted is 

 a symbolic machinery, an algebra, which will enable the relations 

 to be handled more or less mechanically, with the rules of the 

 symbolic manipulation looking after the complexities. It seems 

 likely that the theory of sets, especially as developed by Bourbaki 

 and Riguet, will provide the technique. But further research is 

 needed into these questions. 



8/17. Interference. If acid and alkali are passed along the same 

 pipe they destroy each other; what will happen if two messages 

 are passed along the same channel? — will they interfere with, and 

 destroy, each other? 



Simple examples are quite sufficient to establish that the same 

 physical channel may well carry more than one message without 

 interference, each message travelling as if the others did not exist. 

 Suppose, for instance, that a sender wanted to let a recipient know 

 daily, by an advertisement in the personal column of a newspaper, 

 which of 26 different events was being referred to, and suppose he 

 arranged to print a single letter as the coded form. The same 

 channel of "one printed letter" could simultaneously be used to 

 carry other messages, of variety two, by letting the letter be printed 

 as lower case or capital. The two messages would then be trans- 

 mitted with as little interference as if they were on separate pages. 

 Thus, if ten successive messages were sent, NKeSztyZwm 

 would transmit both nkesztyzwm and 1 10 10 10 

 completely. It is thus possible for two messages to pass through the 

 same physical thing without mutual destruction. 



As an example of rather different type, consider the transformation 

 of Ex. 2/14/11, and regard the position of, say. A'" as a coded form 

 of that of A (with B'" similarly as the coded form of B). Thus 

 treasure might be buried at A and weapons at B, with recording 

 marks left at A'" and B". Now a change in the position of B 

 leads to a change of A'", so 5's value plays an essential part in the 

 coding o^ A to A'" (and conversely of /I on B"')\ so the two messages 



158 



