REGULATING THE VERY LARGE SYSTEM 13/7 



13/5. Constraints. What this means, by the law of Requisite 

 Variety, is that the variety in the disturbances D is not really as 

 large as it seems; in other words, by S.7;8, the disturbances show a 

 constraint. 



Thus the case we are led to is the following: D has many compo- 

 nents, each of which shows variety. The first estimate of D's 

 variety puts it too high, and we are in danger of deducing (if the 

 regulator's capacity is given) that regulation of £" to a certain degree 

 is not possible. Further examination of D may, however, show that 

 the components are not independent, that constraint exists, and 

 that the real variety in D is much lower than the first estimate. It 

 may be found that, with Ws capacity given, this smaller variety can 

 be regulated against, and full regulation or control achieved at E. 

 Thus the discovery of a constraint may convert "regulation impos- 

 sible" to "regulation possible". If Ks capacity is fixed, it is the 

 only way. 



We are thus led again to the importance and usefulness of dis- 

 covering constraints, and to yet another example of the thesis that 

 when a constraint exists it can be turned to use (S.7/14). 



Let us then consider the question of what constraints may occur 

 in the disturbances that affect very large systems, and how they may 

 be turned to use. The question is of major practical importance, 

 for if i?'s capacity is not easily increased and the other methods are 

 not possible, then the law of Requisite Variety says that the discovery 

 of a constraint is the would-be regulator's only hope. 



13/6. As was said in S.7/10, constraints do not fall into a few 

 simply-described classes. Having indicated some of the more 

 interesting possibihties in Chapter 7, I can only continue to mention 

 those classes that are of peculiar interest to us now. With this 

 brief reference I shall pass by a vast subject, that comprises a major 

 part of all human activity. 



Accordingly we shall study one particular form of constraint. It 

 is of great interest in itself, it will illustrate the thesis of the last 

 chapter, and it is of considerable practical importance in the regula- 

 tion of the very large system. 



REPETITIVE DISTURBANCE 



13/7. Though little reference has been made to the fact in the last 

 few chapters, many disturbances (and the corresponding regulatory 

 responses) are repetitive, especially if the system is viewed over a 

 long time. The cough reflex is regulatory and useful not merely 



247 



