HOUSE_OVERSIGHT_016854.jpg

there is a constant structure with some activity going on inside it. When von Neumann’s abstract machine reproduced, it made a copy of itself in another region of the plane. Within the “machine” was a horizontal line of squares which acted as a finite linear tape, using a subset of the finite alphabet. It was the symbols in those squares that encoded the machine of which they were a part. During the machine’s reproduction, the “tape” could move either left or right and was both interpreted (transcribed) as the instructions (translation) for the new “machine” being built and then copied (replicated)—with the new copy being placed inside the new machine for further reproduction. Francis Crick and James Watson later showed, in 1953, how such a tape could be instantiated in biology by a long DNA molecule with its finite alphabet of four nucleobases: guanine, cytosine, adenine, and thymine (G, C, A, and T).13 As in von Neumann’s machine, in biological reproduction the linear sequence of symbols in DNA is interpreted—through transcription into RNA molecules, which then are translated into proteins, the structures that make up a new cell—and the DNA is replicated and encased in the new cell.

A second foundational piece of work was in a 1945 “First Draft” report on the design for a digital computer, wherein von Neumann advocated for a memory that could contain both instructions and data.14 This is now known as a von Neumann architecture computer—as distinct from a Harvard architecture computer, where there are two separate memories, one for instructions and one for data. The vast majority of computer chips built in the era of Moore’s Law are based on the von Neumann architecture, including those powering our data centers, our laptops, and our smartphones. Von Neumann’s digital-computer architecture is conceptually the same generalization—from early digital computers constructed with electromagnetic relays at both Harvard University and Bletchley Park—that occurs in going from a special-purpose Turing Machine to a Universal Turing Machine. Furthermore, his self-replicating automata share a fundamental similarity with both the construction of a Turing Machine and the mechanism of DNA-based reproducing biological cells. There is to this day scholarly debate over whether von Neumann saw the cross connections between these three pieces of work, Turing’s and his two. Turing’s revision of his paper was done while he and von Neumann were both at Princeton; indeed, after getting his PhD, Turing almost stayed on as von Neumann’s postdoc.

Without Turing and von Neumann, the cybernetics of Wiener might have remained a dominant mode of thought and driver of technology for much longer than its brief moment of supremacy. In this imaginary version of history, we might well live today in an actual steam-punk world and not just get to observe its fantastical instantiations at Maker Faires!

My point is that Wiener thought about the world—physical, biological, and (in Human Use) sociological—in a particular way. He analyzed the world as continuous variables, as he explains in chapter 1 along with a nod to thermodynamics through an overlay of Gibbs statistics. He also shoehorns in a weak and unconvincing model of information as message-passing between and among both physical and biological entities. To me, and from today’s vantage point seventy years on, his tools seem woefully

13 “A Structure for Deoxyribose Nucleic Acid,” Nature 171, 737–738 (1953).
14 https://en.wikipedia.org/wiki/First_Draft_of_a_Report_on_the_EDVAC#Controversy. Von Neumann is listed as the only author, whereas others contributed to the concepts he laid out; thus credit for the architecture has gone to him alone.

51

HOUSE_OVERSIGHT_016854