D. M. MACKAY 
education, and to revolutionize the academic standards 
attainable with present human resources. 
In science itself routine aspects of information-retrieval are 
rapidly passing beyond the capabilities of human librarians, 
and great efforts are being made to codify knowledge for 
machine-handling before it is too late. Here, however, 
we meet an interesting difficulty inherent in the labile 
character of science itself; for the basis of any automatic 
filing or retrieval system is a systematic method of determining 
the likely relevance of each item to others and to all possible 
enquiries, present and future; and it is of the essence of scien- 
tific advance that such criteria are in principle hard to find, 
and even harder to maintain unaltered for any length of time. 
It seems unlikely that any advance in machine technology can 
circumvent this difficulty; but since without machines the 
difficulty is no less and the situation even more unmanageable, 
the mechanization of library routine is high on the priority list 
at the present time. 
What then of still more complex functions such as the 
diagnosis of disease, or the administration of justice? Both 
have their mechanical aspects—looking-up relevant descriptions 
of syndromes, or of Acts of Parliament, for example—in which 
something like an electronic computer might undoubtedly be 
used to save drudgery. Developments to this end are already 
under way. 
Can we then expect that in due course these whole functions 
will be mechanized? Doctors and lawyers are not optimistic; 
and their most basic objection has nothing to do with any 
ignorance of the capacities of machines. They point out that 
what a patient or a litigant wants is finally a personal judgment, 
based partly on experience which the doctor or the judge may 
not be able, even in principle, to make explicit. It is not the 
explicit and deductive, but the implicit and creative aspects 
of human judgment that they believe to be irreducibly a 
of the would-be mechanizer. 
Is this mere obscurantism? I do not think so. True, there 
now exists at least one standard procedure whereby any pattern 
160 
