D. M. MACKAY 
the appropriate band waggon. As Popper points out, there 
are many human situations where such extrapolating is logically 
impossible, since the attitude of the predictor himself to his 
prediction is one of the data needed to complete that prediction. 
In other words, no matter how scientific the basis of an alleged 
prediction of the course of history, it may still be possible for 
an individual or society confronted with it to make nonsense 
of it, by taking the opposite attitude to the one assumed when 
the calculation was made. 
SELF-VALIDATING DESCRIPTIONS 
The mechanically-minded reader may now be impatient. 
“Surely”? he may ask “‘it must be possible, with a sufficiently 
complex predictive model, to trim the prediction in such a 
way that when published it ensures its own fulfilment?” ‘There 
are, indeed, circumstances in which it is possible to ‘‘adjust”’ 
a poll result, for example, so that its publication brings about 
its own fulfilment. But in the field of human attitudes such 
circumstances are exceptional. 
There is, moreover, an obvious objection to such a procedure 
at a more fundamental level. Suppose we assume that the 
adjusted forecast is confirmed by events, so that nobody who 
believed it feels himself to have been deceived. But suppose it had 
not been published? Then (ex hypothest) the result would have 
been different—perhaps even reversed. ‘Thus publication here 
was not primarily informative, but manipulative. And although 
a large computer may not always be essential to this end, the 
more powerful the predictive apparatus used, the more subtle 
and wide-ranging will be the manipulation of social attitude 
possible under the guise of scientific prediction. 
Here, I believe, is a threat to society far more serious than 
any from the growth of automation. If any future use of 
computers wants watching on behalf of mankind, it is this; 
for our society’s insatiable thirst for information about itself 
and its future has now laid it wide open to the most subtle 
bondage of all, in which major decisions can in principle be 
taken for it (wittingly or otherwise) by those whom it asks to 
164 
