David Cole judges that my views are infected with 'élitism and generational 
prejudice'.  There is not much point in responding in kind.  More ad hominem 
rhetoric will not clarify the issues involved here; but I do want to plead 
guilty to élitism of a sort, not certainly to élitism of some silly 
socio-economic sort but to intellectual élitism, yes.   (Socio-economic elitism 
is based upon the dubious premise that there is credit to be garnered for 
having chosen one's grandfather well.)
 
G. H. Hardy, a convinced socialist as well as a great mathematician, famously 
observed that since most people cannot do anything well, it does not much 
matter what they do; and in this sense it does not I suppose matter that some 
dullards elect to become systems programmers or, now, system administrators.  
 
Whether they should be encouraged to do so is another question about which I 
have a predictable opinion; but I do not think their needs should shape IBM's 
publications policies.   
IBM manuals anciently addressed why as well as how.  Increasingly, they are 
being turned into tutorials for the unprepared, who want only to know how at a 
very low level of generality.  
 
The PrOp was and remains an honorable exception.  It is not a perfect 
publication, whatever such a thing may be; but it has evolved over a now very 
long period of time to meet one class of needs of one class of people; and it 
does what it was anciently supposed to do very well.  
 
Detailed criticisms are appropriate.  I register them frequently through the 
standard IBM channels, which are, as I have already had occasion to observe, 
responsive.  What I am objecting to is an effort to strike down this great work 
of time, replacing it with one more howto for the unprepared.
 
That there is a generational component in all this I am also ready to concede.  
I recently had occasion to explain the C switch statement to a very able tween. 
 She did not understand
 
"The switch statement is a multi-way decision that tests whether an expression 
matches one of a number of constant integer values , and branches accordingly."
 
This is understandable since the switch statement, which uses a branch table, 
does not test anything and would be of little interest if it did.  I suppose 
that Kernighan and Ritchie, who of course know what is involved here, judged 
that to explain it would be too complicated and that an explanation would not 
be understood anyway.  They were wrong; but 'their' view, which may be that of 
some editor, is pernicious and will shortly be pervasive.  I have no illusions 
that I can stop such changes; but, to borrow a phrase from Quine, I can drag my 
feet. 

John Gilmore Ashland, MA 01721-1817 USA                                         

Reply via email to