About two years ago, I had enough of only writing code to make work. As an experimental filmmaker, I started writing my own animation software in Lisp, Forth and Assembler on Atari computers in the 1990's, later moved to super-8, and then worked for about a decade on 16mm and eventually 35mm celluloid in combination with various optical and mechanical contraptions. Many of these image-generating devices were midi-controlled, and at some point a gradual transition happened towards using code to control image-generators that were also bits of code. After fifteen years of only working with digital computers, and initially triggered by RSI, I started to look for other ways of working and thinking. Something more connected to the physical world and involving a larger scope of bodily activities than only looking at a screen, pressing keys and waving magical wands of various kinds.

So while meditatively soldering the first circuit boards of what was going to become an analog video-synthesizer, I remembered a device that I had seen when I was studying electronic music. I used to spend one afternoon a week in the analog studio of the Royal Conservatoire in The Hague and one day, a large machine appeared, consisting of a bunch of controls and a removable patchpanel divided into modules. It looked very similar to the rest of the studio, except that I could not make any sense of what these modules were supposed to do. Many years after this fleeting and frustrating initial encounter, I still associated the 'analog computer' - since that is what fellow students told me what it was - with an idea of baffling and intriguing otherness.

When I started researching the history of analog computing, it struck me how strange it was that this history is almost completely forgotten. Electronic analog computers were crucial in developing supersonic flight, helped put man on the moon, calculated nuclear power plants and flood defenses, and were prevalent in engineering and scientific simulation labs from the early fifties onwards. And still, in most histories of computing, the only reference to analog computing tends to be to the mechanical analog computers developed in the thirties that were in comparison extremely rare, short-lived, did none of these things and certainly never developed into an industry. It took until the mid seventies until digital computers could do what analog computers were used for; after that they quickly disappeared and the cultures and practices associated with them were absorbed into the culture around digital simulation.

To give a feel of how the relationship between analog and digital computing was around 1955-56, here a quote from a description of 'Project Cyclone', one of the early, large, defense-funded US analog computing projects: “In keeping with project goals to improve and to verify the accuracy of results, problems were routinely re-run on digital computers. In a typical application – the simulation of a guided missile in three dimensions – the average run time for a single solution on the electronic analog computer was approximately one minute. The check solution for this problem by numerical methods on an IBM CPC (Card Programmed Calculator) took 75 hours to run.”

Most of the electronic analog computers referred to here were so-called 'functional' analog computers. These consist of modules that can add, invert, integrate and multiply voltages, often include comparators and switches, and sometimes also have modules that produce logarithmic, exponential or trigonometric curves. Initial conditions and variables are set by using a potentiometer for each value. By connecting these modules with patch cables, electronic circuits can be constructed that behave according to systems of mathematical equations. It is a type of computation that is inherently parallel, has no memory, no clock, and where precision and speed depend on the noise level and signal bandwidth of the hardware. Depending on the type of problem, the end result of the computation can be a single value read from a voltmeter or can be plotted as a set of curves. In fact, since the very first beginnings of this type of computing in the late thirties, these machines were designed to operate on two speeds: one slow, standard speed intended for drawing curves on paper, and a fast speed for what was called 'repetitive operation'. In 'rep-op' mode, calculations are restarted ten to fifty times a second and the resulting curves are displayed on an oscilloscope, so that a human operator can explore the influence of each variable by turning knobs, making the curves change in real-time.

It is this human-in-the-loop interactive exploration and simulation that is an important characteristic and perhaps the most important legacy of analog computing. Physical simulations involving humans were one of the areas in which analog computing survived the longest, simply because it took until the eighties before digital computers became affordable that were fast enough. Also this aspect has the most obvious links to interactive exploration in the electronic arts; the analog computer is the direct ancestor of both the analog audio synthesizer as well as the various synthesizers that were developed for video.

Going back a bit more in the history of analog computing, one encounters another category of devices that were at the time referred to as 'direct analog' computers and that embody the origin of the word 'analog' before it came to mean 'non-digital'. These encompass a wild and fascinating fauna of extremely different types of set-ups that each explore an equivalence between two types of physical systems. One easy to understand example would be the analogy between electricity and water, in which voltage is analogous with water level, electrical current with water flow and electrical capacitance with the size of a water reservoir. Since the formulas describing water flow and electrical current are essentially the same, one can make an electrical model of a river system using a few resistors, capacitors and a voltage source. Water levels can then be 'computed' by setting some input voltages and measuring the output at some points in the circuit, the advantage being that it is a lot easier to change a resistor than to vary the width of a river bed. Devices exploring other analogies would consist of grids of resistors and coils, of resistive paper or even of threedimensional reservoirs filled with a liquid with a known electrical resistance. Not to mention a host of fascinating non-electronic devices involving soap bubbles, heat or rubber sheets.

Interesting about this part of the history of analog computing is that it shows how incredibly standardized - one could argue impoverished - our ideas about computation have become. Perhaps the most fascinating aspect of this for me is that here is a decades-long culture of computation and simulation in which the word 'information' is practically absent and that shows none of the implicit and naive platonism that is so rampant in digital culture, probably because computations are so clearly embodied and the analogies are so explicitly formulated.

These investigations of mine into the history of analog computing have so far led to one finished video piece and a number of experiments and ongoing projects that will occupy me for the years to come and that will undoubtedly feed into other projects. As a very hands-on way to learn about analog computing (and about the basics of analog electronics) I had the pleasure to find and restore an EAI TR-48 analog computer from 1963. This was one of the first transistorized analog computers, marketed as one of the first 'desktop' computers, weighing in at a mere 400 pounds. I have been using it so far as a generator of control voltages for much faster oscillators that can generate signals up to 60 MHz, the bandwidth of the analog signals going into a HD monitor via the VGA connector. Also it serves as the model for a high-frequency analog computer that I am in the process of building, with the aim to synthesize HD video signals using the 'language' of analog computing. This device will essentially be an analog video synthesizer in which the basic modules are more fundamental than what is typically found in a synthesizer; so no sine or triangle wave oscillators, but integrators, adders, inverters and multipliers.

In these projects I am certainly not interested in reviving any kind of debate about analog versus digital (debates that seem to be repeating the same boring tropes since at least the time of Charles Babbage), but I am interested in the productive energy of trying to integrate a different paradigm. I find it stimulating to think of ways to compose images as signals in a system that has no memory. Programming by connecting physical objects is interestingly different from writing code, leading to different types of resistances in the creative process and different ideas that emerge as solutions to these resistances. And in the longer term, I am looking for ways to make moving image compositions that have characteristics that are somehow specific to analog: chaotic oscillators are an interesting and perhaps obvious choice, but I am also thinking of what it means to electronically integrate or differentiate signals and how it would be possible to build compositions from this essential building block. And finally I am also looking at early analog models of neurons that have very interesting temporal properties and that would be great to use as a set of oscillators. It is perhaps in the realm of 'direct analogs' that the physicality of these machines can its most interesting and fundamental voice.

Joost Rekveld is an artist and experimental filmmaker who lives and works in Amsterdam and Ghent. He is interested in how humans can learn about the world through a dialog with their machines and tools. Since 1990 he has realized a small oeuvre of abstract animated films and installations, and collaborated with many composers and theatre makers. Joost is also active as a curator and has for many years been teaching interdisciplinary arts at the intersection of art, science and technology. He is currently a researcher at the School of Arts, University College Ghent.

Joost Rekveld
Analog Computing
arts.codes Vol 1
May 20th 2017