ULTRASTABLE: THE SLEEPING MACHINE
In the first part of her essay on self-organization in machines, Evelyn Fox Keller discusses the “Homeostat” of W. R. Ashby (1903-1972), which he designed in 1948 to explore the possibilities of self-organization in machines. The apparatus (which the British Library, wonderfully, calls a “baroque and bulky machine” in its post on Ashby’s papers) consisted of four circuit blocks, each connected to a thin metal vane immersed in small water tanks. Displacing the vane would trigger the block to produce an electrical current. With the vane set to the middle position, any given block would produce no current; disturb the vane’s position and the block would send current to the other blocks, causing their vanes to move. Overall, the apparatus was disposed to seek the equilibrium state—vanes set to their neutral positions—dynamically, in response to changes in the “environment.”
Ashby considered his machine to be very good for thinking with—a position not universally agreed, as Fox Keller discloses:
This was a machine for doing nothing, or, as Grey Walter put it, a machina sopora, a sleep machine. But to Ashby this was to miss the central point of the device. And that was that a reorganization (rewiring, in this case) would be automatically triggered every time the system departed from its range of stability, thereby guaranteeing that it would always reach a stable state no matter how serious the perturbation of the inputs. It was not merely stable, it was “ultrastable.”
This notion of ultrastability was of great importance to Ashby. It marked a cru- cial distinction between the capacity of his device and that of others being of- fered up as candidates for self-organizing machines… here was a machine that could automatically change its organization in a positive direction—it could spontaneously shift from a “bad” organization, or way of behaving, to a “good” one. It could adapt…. This, he claimed, was precisely what his homeostat did. It showed that a properly designed machine could exhibit autonomous, self-organizing behavior of just the kind that animals displayed (Fox Keller 68).
Although Walter’s description of the Homeostat as a “sleeping machine” was meant as a dig, there’s something sublime in it: no matter the perturbation, the vanes seek their quietude, and the machine finds its way to a state of “ultrastability” akin to slumber—or enlightenment?
The Homeostat is a striking example of design being used to explore a philosophical problem (Ashby, in his journals, described himself as an artist foremost). The apparatus became an early exemplum of cybernetics; Ashby demonstrated it at the 9th Macy conference, and Norbert Wiener extolled its virtues and implications. The degree to which contemporary theories of animal cognition directly drove the cybernetic imaginary is striking. At the same time, it’s interesting to note that Ashby developed the circuit blocks from surplus British bomb-control devices. At its start, cybernetics thought the thinking machine by joining war and the animal.