Nice to see some evidence for the hypothesis that (unlike Hoare with QuickSort) von Neumann wasn't trying to develop merge sort as a new sorting algorithm, but was merely trying to see if he could, in software, do the part of the job that had up until then been done, in hardware, on a separate unit record machine.
> He never actually gets around to coding
the entire sorting routine in this document;
only the merging process is described.
> He formulated the merging method as
follows (based on a procedure then used
with the IBM collator):
I highly recommend this book if you want to know more about John Von Neumann and his contribution to some of the most influencing topics of today's world including Game Theory, Quantum Physics, Electronic Computer, Hydrogen bombs etc.
This book is also great for history lovers since it also delves into the period of post World War II when the best of the brains started migrating to US from Europe and influence major national defense and security projects.
It's amazing to see how Von Neumann is able to move from one hard topic to another and make an impact.
> He never actually gets around to coding the entire sorting routine in this document; only the merging process is described.
> He formulated the merging method as follows (based on a procedure then used with the IBM collator):