DISCUSSION 
They fit together because there are a great many of them, all 
seeking where they are going to fit; and when they find the fit, 
and only then, they are stable. This is an example of the 
transition from mechanistic to statistical thinking. Physicists 
have come to terms with this; they now know at what levels 
it is necessary to talk of individual quantum jumps, and at 
what levels it is good enough to consider only large integrations. 
But biologists are still struggling to learn this. Szent-Gydrgyi 
is still frightened by the detailed arithmetic which seems to be 
required. I remember being just as frightened in 1950 when 
I started some new work in electronics. How would I ever master 
the detail? And then I found that the young men who came to 
me could write wiring diagrams for electronic circuits as easily 
as other people write music. They knew how to think in units 
of the right level of organization. In the same way, biologists 
must now learn to think in the right statistical units. 
Let me adda statistical footnote, Szent-Gy6rgyi has calculated 
that altogether 10° people are going to die from existing doses 
of fall-out. That must be just about as many people as die in 
the world now in any 24 hours. I think that every individual 
life is precious; but I think we falsify our own values if we 
flourish 10° deaths as if they were a statistically large number. 
Szent-Gyérgyt: The stability you have talked about may be 
the driving force I was aiming at, and I think about it a great 
deal. I should add that I would not condemn a hundred thous- 
and people to death just because that many die anyway in a day. 
Mackay: ‘There is a missing element in our discussion which 
disturbs me. “Evolution”’ as an explanatory principle is rather 
like “‘money’’. We may (or may not) believe that “‘money can 
buy anything’’; but if we want a new car we have also to ask 
whether the money we have is sufficient. There is a need to 
ask similar questions in any discussion of the adequacy of 
evolutionary explanation. 
Thus in Szent-Gyodrgyi’s combinatorial problem, he stressed 
the complexity of the parameters that have to be “‘just right”’ 
for reproduction. But from the standpoint of information theory 
what one wants to ask is, how redundant is this complexity? 
242 
