September 3, 1955

Dr. John von Neumann
Atomic Energy Commission
Washington 25, D.C.

Dear Dr. von Neumann:

Thank you for your letter of August 15. I was away from Madison at
that time and have taken the first opportunity to reply.

As a result of this discussion,and others with Szilard and others,
I think I have been able to clarify ay views slightly, at least enough
to appreciate better what I do not understand. I am enclosing a revision
of p. 39 of the original draft of that "Growth" paper, which I had sent
you befpre; you may keep or discard these as you papfer, when you have read
them. I think the revision better points up the rather obvious ideas I was
trying to get over. On the whole, I thought it would be better to delete
any specific reference to your models, until I better underateod their appli-
cation. "Beg the question" in the first version was an unfortunate phrase,
perhaps presumptuous as well. Everything that I had in mind by that is
expressed in the revision.

I think the root of our trouble is that we are working at very different
planes. The propagation, and evolutionary elaboration, of complexity is self-
evident, or rather very evident, to a biologist, and we are now concerned with
realistic working models of reproduction. It is by now reasonably obvious how
one might design some such models on an electromemhanical basis; a good chemical
analogue to a punch-card reproducer, if we had an equal knowledge of its parts,
would take us a long ways towards the experimental initiation of life. But -
outside living systems, we have not learned how to string autéchtalytic xheckaxk
molecules together, in an autocatalytic system, 4n such a way as to signlate
a@ punch card reproduction with more than one ar a few bits on it. And none of
the chemical machines that we can now devise gives a product of anywhere near
the complexity of an organism. I can see that you have been looking for the
foundations of an amlomatic theory of reprodiietion, and that I had been needlessly
reading my own mechanical interpretations into it. I would have to ask you
what gaferial interpretations are feasible, Without knowing some of the other
theorems of your system, I could not begin to say whether they would be helpful
in genetic analysis. How might A and A' be understood in biclogical terms, for
example? I can see that written correspondence will be too cumbeszome to
let us get very far. If I can sustain your interest, could I make an appointment
to discuss these matters at closer hand? I will be in Washington at the end
of October (28-29) for a panel meeting at the National Science Foundation, and
could arrange time on the 27th, or perhaps the 29th. The evening of the 27th
would be the most convenient on my part. If you are not too busy, and are disposed
to belabor this subject further, I would be indebted to you for the occasion.

f /
Yours fsincerely,
Aidt EEE A a Fe

Joshua Lederberg

Prnfarany af flanatiaan