AMPLIFYING REGULATION 14/4 



Obviously, both points of view are right. Designers of cranes 

 should be well aware that they are not really amplifiers, but the 

 users of them should think of them as if they were. 



14/3. We can now see how we should view the question of amplify- 

 ing regulation. During the designing (in this chapter) we shall have 

 to be clearly aware that the designer is really achieving only a 

 supplementation, by robbing some easily available and abundant 

 source of it. When he comes to use it, however (a matter for the 

 future), he should forget the fact, and should know only that he is 

 now like a workman equipped with power-operated tools, able to 

 achieve tasks impossible to the unaided workman. 



14/4. Regulation and selection. In S. 13/ 10 we started to consider 

 what would get the regulator (previously assumed to be given) into 

 actual existence, either as a formula for behaving, contained within 

 the organism {Q) that wants the regulation, or as a material machine 

 built by the organism to act for him. We saw that the quantity of 

 design that goes to it can be measured (by the amount of selection 

 necessary) and we saw (S.13/18) that selection can, in a sense, be 

 amplified. To make the matter clearer, let us consider more directly 

 the relation between regulation and selection, especially so far as 

 the quantities of variety or information are concerned. If the dia- 

 gram of immediate effects is 



Designer 



Regulator 



' C 



we want to know how much variety or information the channel 

 between them will have to carry. 



To get a regulator made, selection is essential. Here are three 

 examples : 



The first regulator we discussed (S.11/3) led to our identifying 

 it as 



1 2 3 



pay 



and this particular transformation (the regulatory) had to be selected 

 from the set of all transformations possible, which numbered, in this 

 case, 27 (cf. Ex. 7/7/8). Here the regulator is "made" by being 

 unambiguously specified, i.e. distinguished from the others. 



In S. 13/ 12 another method was used, and a machine, which might 

 be a regulator, was "designed" by a particular value being selected 

 from the set of possible input-values. 



267 



