“Automation Divine”: Early Computer Music and the Selling of the Cold War
Matthew Guerrieri dives deep into something particular about the early days of computer music in the United States. It got its start, quite literally, in the off-hour downtime of the military-industrial complex.
It was a love song—not what viewers expected, perhaps, who tuned into a July 1956 episode of Adventure Tomorrow, a science documentary program broadcast by KCOP, channel 13, out of Los Angeles. But, then again, it was a love song to a computer. Push-Button Bertha. Sweet machine. What a queen. Jack Owens, the lyricist (and, on that July 1956 episode, the performer), had taken his inspiration from the tune’s composer: a Datatron 205, the room-filling flagship computer of Pasadena-based ElectroData, Inc.
Bertha’s not demanding
Never wants your dough
Just flip a switch and she’ll go
Just that month, ElectroData had been acquired by the Burroughs Corporation; Burroughs, an adding-machine manufacturer, was buying a ready-made entry into the computer business. The Datatron had been programmed by Martin L. Klein and Douglas Bolitho, a pair of engineers. (Klein also moonlighted as Adventure Tomorrow’s on-air host.) “Push-Button Bertha” wasn’t Datatron’s magnum opus, but rather one of thousands of pop-song melodies the program could spit out every hour. Its inspiration was purely statistical.
In fact, it was a perceived deficit of inspiration that supposedly prompted the project. Klein explained: “[W]e set out to prove that if human beings could write ‘popular music’ of poor quality at the rate of a song an hour, we could write it just as bad with a computing machine, but faster.” Klein and Bolitho went through the top one hundred pop songs of the year, looking for patterns. They came up with three:
1. There are between 35 and 60 different notes in a popular song.
2. A popular song has the following pattern: part A, which runs 8 measures and contains about 18 to 25 notes, part A, repeated, part B, which contains 8 measures and between 17 and 35 notes; part A, again repeated.
3. If five notes move successively in an upward direction, the sixth note is downward and vice versa.
To those principles were added three more timeworn rules:
4. Never skip more than six notes between successive notes.
5. The first note in part A is ordinarily not the second, fourth or flatted fifth note in a scale.
6. Notes with flats next move down a tone, notes with sharps next move up a tone.
The six rules were then put to work via the Monte Carlo method, which had been developed around the speed and indefatigability of the newly invented computer, harnessing the wisdom of a crowd of countless, repeated probabilistic calculations. Fed a stream of random, single-digit integers (which limited the number of available notes to 10), Datatron would test each integer/note against its programmed criteria. If it met every guideline, it was stored in memory; if not, it was discarded, and the program would move on to the next integer. After a few dozen iterations, presto: another prospective hit. Or not. Klein and Bolitho never admitted the program’s rate of success; out of Datatron’s presumably thousands of drafts, only “Push-Button Bertha” saw the light of day.
The song took its place in a small but growing repertoire of computational compositions. 1956 was a banner year for statistically designed, computer-generated music. A team of Harvard graduate students, including Frederick P. Brooks, Jr. (who would go on to lead the design of IBM’s famed System/360 mainframes), programmed Harvard’s Mark IV computer to electronically analyze and then generate common-meter hymn tunes. (“It took us three years to get done,” Brooks later remembered, “but we got stuff you could pass off on any choir.”) Lejaren Hiller and Leonard Isaacson used the University of Illinois’ ILLIAC I machine to create a string quartet, its movements moving through music history from basic counterpoint to modern speculation; portions of the Illiac Suite were premiered on August 9, 1956, only a month after the TV debut of “Push-Button Bertha.”
“Push-Button Bertha” was a curiosity, but it reveals something particular about the early days of computer music in the United States. The Harvard and Illinois efforts were research and experimentation that were at least nominally driven by curiosity and the prospect of expanding academic knowledge. But all three were, in part, justifications of more hard-nosed concerns. When Owens sang of Bertha never wanting your dough, he shaded the truth a bit; Bertha wanted quite a bit of dough indeed. A Datatron 205 computer cost $135,000, and that didn’t include necessities such as a control console, punched-card input and output equipment, or magnetic tape storage. Nor did it include desirable extras, such as the capability to do floating-point calculations—that alone required an additional $21,000, more, at the time, than the median price of a house in the United States. The Burroughs Corporation needed to justify the price tag of its newest product line. “Push-Button Bertha” was putting a cloak of high-minded research around that most hallowed of American art forms: a sales pitch.
Off the air, Martin Klein wasn’t employed by ElectroData or Burroughs; at the time, he worked for Rocketdyne, North American Aviation’s rocket and missile division, based in the San Fernando Valley. The Brooklyn-born Klein had initially pursued music—as a teenager, he composed and copyrighted (though did not publish) a piano-and-accordion number entitled “Squeeze-Box Stomp”—but instead took up science. He did graduate work at Boston University, earning a master’s and a Ph.D.; his master’s dissertation, especially (on methods of using optical refraction to measure the flow of air around high-speed objects—supersonic planes or missiles, say), foreshadowed his work at Rocketdyne, which was largely devoted to designing circuits to convert rocket-engine test-firing data into forms that a computer could analyze.
But that hint of a performing career would eventually resurface. In 1956, Klein began to spend his weekends on television. Saturdays brought Wires and Pliers, in which Klein and his North American Aviation colleague Harry C. Morgan (with the help of electrical engineer Aram Solomonian) showed viewers how to assemble simple electronic circuits and gadgets. (The show’s sponsor, the Electronic Engineering Company of California, conveniently sold kits containing the necessary components for each project.) On Sundays, Adventure Tomorrow promoted technological optimism by way of the latest advances from California’s rapidly expanding electronics and defense industries. Burroughs needed a showcase for its technology; Klein needed technology to showcase.
Klein had demonstrated a flair for technologically enhanced promotion. His first efforts with the Datatron were intended to spotlight Pierce Junior College, where Klein was an instructor. In December 1955, Klein had the computer predict the winners of New Year’s Day college football bowl games; it got four out of five correct. News reports made sure to mention that Klein was teaching computer design at Pierce, “one course of a whole program in electronics offered by the college preparing men for occupations in this industry, so vital to our country’s defense.” By April, Klein, under the auspices of Pierce and backed by several electronics-industry sponsors (including ElectroData), was on the air every week. Wires and Pliers didn’t last long, but Adventure Tomorrow did. From the beginning, Adventure Tomorrow was a cheerleader for the latest military technology—“the wondrous world of missiles, jets, and atomic projects,” as a later advertisement for the program put it. It was in that spirit that Klein and Bolitho went to work extracting a bit of publicity-friendly frivolity from the Datatron 205.
If Klein knew how to engineer attention, Bolitho’s specialty was wrangling the machines. Early computers were a forest of hard-wired components, fertile ground for capricious behavior. Bolitho’s rapport with the finicky beasts was legendary. His ability to coax computers into reliability eventually led him to be tasked with leading prospective customers on tours of Burroughs’ Pasadena plant. “He had some kind of magical quality whereby he could walk up to a machine that was covered with cobwebs and dust and turn it on and that thing would work, even if it had been broken for years,” a fellow engineer remembered.
Depending on his audience, Klein would oscillate between extolling computers as inhumanly infallible and comfortingly quirky. Explaining the basics of the new tools to readers of Instruments and Automation magazine, he lauded “the advantage of automatic control over control by human operators where human forces are constantly at work to disrupt the logical processes.” But, recounting the genesis of “Push-Button Bertha” for Radio Electronics—a magazine aimed more at hobbyists and amateurs—Klein struck a more whimsical note, echoing (deliberately or not) the Romantic stereotype of the sensitive, temperamental artist:
The words “electronic digital computer” immediately conjure up a picture of a forbidding, heartless device. Those of us who design computing machinery know this isn’t true. Computing machines have very human characteristics. They hate to get to work on a cold morning (we call this “sleeping sickness”). Occasionally, for unexplainable reasons, they don’t work the same problem the same way twice (we say, then, that the machine has the flu).
Klein’s joke turns a little more grim knowing that, by 1961, the United States military was using no fewer than sixteen Datatron 205 computers at twelve different locations—including the Edgewood Arsenal at the U.S. Army’s Aberdeen Proving Grounds in Maryland, where the technology that produced “Push-Button Bertha” was instead used to calculate simulated dispersal patterns for airborne chemical and biological weapons.
All the composing computers, in fact, were military machines. ILLIAC, for instance, was a copy of a computer called ORDVAC, built by the University of Illinois and shipped to the Aberdeen Proving Grounds to calculate ballistics trajectories. Hiller and Isaacson had first learned their way around ILLIAC and the Monte Carlo method trying to solve the long-standing problem of determining the size of coiled polymer molecules—a problem of more than passing interest to the United States government, which funded the research as part of a program to develop and improve synthetic rubber production. (It was Hiller, who had coupled his studies of chemistry at Princeton with composition lessons with Milton Babbitt, who realized the same mathematical technique could be applied to musical composition.)
Harvard’s Mark IV was the last in a group of computers designed by Howard Aiken. The Mark I had helped work out the design of the first atomic bombs; Marks II and III were built for the U. S. Navy. The Mark IV, which had produced all those hymn tunes, had been funded by the U. S. Air Force; it worked out guided-missile flight patterns and helped design lenses for the U-2 spy plane. The Harvard computers, it turned out, ran more reliably if they were never turned off; Aiken duly assigned Peter Neumann, a music-loving graduate student, to watch over the Mark IV from Friday night until Monday morning. Student projects—hymn-tune-generation included—happened on the weekends. Computational composition in the United States got its start, quite literally, in the off-hour downtime of the military-industrial complex.
For a few years, American computer-music researchers may have looked with jealousy across the Atlantic, to Paris and Cologne and the fledgling, dedicated electronic-music studios that had blossomed under the aegis of government-supported radio stations. But there are suggestions that those European efforts, too, emerged out of a nexus of technology and defense.
The origin story of the famous WDR electronic-music studio in Cologne, for instance, starts with an American visitor. In 1948, American scientist Homer Dudley visited Germany, bringing along his invention, the vocoder; the device made a crucial impression on Werner Meyer-Eppler, who would later help create the WDR studio—and whose students would include the studio’s most famous denizen, Karlheinz Stockhausen.
Dudley was an employee of Bell Labs, one of the great 20th-century American research-and-development shops, a hive of telecommunications innovation. The vocoder had originally been developed as part of investigations into shrinking the bandwidth of telephone signals, in order that more messages might travel over the same wires. But, especially with the onset of war, the work at Bell Labs was increasingly aligned with the desires of government. The vocoder had been pressed into wartime service as the backbone of SIGSALY, the Allied system that successfully masked high-level phone conversations from German eavesdropping, and which practically introduced numerous features of the modern digital communications landscape: compression, packet-switching, electronic key encryption. (The keys for SIGSALY were stretches of electronically generated random white noise, pressed onto matched pairs of phonograph records, each pair being destroyed after a single use.)
One wonders if Meyer-Eppler had been targeted for recruitment into the development of SIGSALY’s sequels; after all, so much of the WDR studio’s work seemed aligned with and adaptable to the sort of research that Bell Labs was pursuing in the wake of its wartime work. Think of one of the WDR studio’s most celebrated productions, Stockhausen’s Gesang der Jünglinge; if the work’s combination of a transmitted human voice and electronic noise recalls SIGSALY, the way it deconstructs, processes, and reassembles that voice, the way it filters the sounds through various statistical screens—it practically outlines a research program for next-generation voice and signal encryption.
At the very least, the new music triangulated Dudley’s sonic manipulation with two other innovations, the transistor and Claude Shannon’s new information theory; all three had emerged from Bell Labs, which would also birth Max Mathews’s pioneering MUSIC software—all the while pursuing numerous military and defense projects. In later years, Bell Labs would consult on the formation of IRCAM, Pierre Boulez’s hothouse of computer music in Paris.
But IRCAM, envisioned as a seedbed, was instead an endpoint, at least in terms of the sort of institutional computing that, in its interstices, had provided a home for early computer music. Already the future was in view: a computer in every home, a chip in every device, casual users commandeering the sort of processing power that the builders of the UNIVACs and the ORDVACs and the Datatrons could barely imagine. (The year IRCAM finally opened, 1977, was the same year that the Apple II was introduced.) Even the output of those institutions—for instance, Max/MSP, the descendant of an IRCAM project—was destined for laptops.
Surrounded by the surfeit of computation, it is hard to imagine the scarcity that led those first computer musicians to a marriage of convenience with the military and national-security bureaucracies—a marriage convenient to both sides. But to understand that give-and-take is to understand something about the nature of music in the middle of the 20th century, the technocratic faith that came to inform so many aspects of the culture. The sounds of the post-war avant-garde were never far, in concept or parentage, from the technological needs of the Cold War.
And what of the composer of “Push-Button Bertha”? Even as it became obsolete, the Datatron 205, with its blinking console and spinning tape drives, enjoyed a long career as a prop in movies and television, lending a technologically sophisticated aura to everything from Adam West’s television Batcave to Dr. Evil’s lair in the Austin Powers movies. That, too, may have been a result of Klein and Bolitho’s public-relations stunt. Only a few months after the Adventure Tomorrow premiere of “Push-Button Bertha,” producer Sam Katzman, a veteran impresario of low-budget genre movies, gave the 205 its big-screen debut, going to the Datatron’s Pasadena factory home to film scenes for a science-fiction production called The Night the World Exploded. In the movie, the 205—mentioned prominently, by name, in dialogue and narration—is used to determine just how long before a newly discovered and volatile “Element 112” works its way to the earth’s surface and destroys the planet. The Datatron had returned from its pop-song holiday to a more familiar role for the era’s computers: calculating the end of the world.
1. The founder of the Burroughs Corporation, William Seward Burroughs I, was the grandfather of the Beat writer William S. Burroughs. In a 1965 interview, the younger Burroughs gave his opinion of computational art:
“INTERVIEWER: Have you done anything with computers?
BURROUGHS: I’ve not done anything, but I’ve seen some of the computer poetry. I can take one of those computer poems and then try to find correlatives of it—that is, pictures to go with it; it’s quite possible.
INTERVIEWER: Does the fact that it comes from a machine diminish its value to you?
BURROUGHS: I think that any artistic product must stand or fall on what’s there.”
(See Conrad Knickerbocker, “William Burroughs: An Interview,” The Paris Review vol. 35 (1965), p. 13-49.)
4. F. P. Brooks, A. L. Hopkins, P. G. Neumann, W. V. Wright, “An experiment in musical composition”, IRE Trans. on Electronics Computers, vol. EC-6, no. 3 (Sep. 1957). See also Grady Booch, “Oral History of Fred Brooks,” Computer History Museum Reference number: X4146.2008 (http://archive.computerhistory.org/resources/access/text/2012/11/102658255-05-01-acc.pdf, accessed September 18, 2018).
5. Datatron prices from Tom Sawyer’s Burroughs 205 website (http://tjsawyer.com/B205prices.php, accessed September 10, 2018). In 1957, the median home price was approximately $17,000, as calculated from Robert Shiller’s archive of historical home prices (http://www.econ.yale.edu/~shiller/data/Fig3-1.xls, accessed September 10, 2018).
6. Martin L. Klein, The Determination of Refractive Indices of Dynamic Gaseous Media by a Scanning Grid, M.A. Thesis, Boston University (1949). Klein’s doctoral thesis was on zone plate antennae—forerunners of the modern flat versions used for HD television.
9. Richard Waychoff, Stories about the B5000 and People Who Were There (1979), from Ed Thelen’s Antique Computers website (http://ed-thelen.org/comp-hist/B5000-AlgolRWaychoff.html, accessed September 10, 2018).
10. Martin L. Klein, Harry C. Morgan, and Milton H. Aronson, Digital Techniques for Computation and Control (Instruments Publishing Co.: Pittsburgh, 1958), p. 9; Klein, “Syncopation by Automation,” p. 36.
11. For the 205’s usage within the military, see Martin H. Weik, A Third Survey of Domestic Electronic Digital Computing Systems (Public Bulletin no. 171265, U.S. Department of Commerce, Office of Technical Services, 1961), p. 145. For the 205 at Edgewood, see Arthur K. Stuempfle, “Aerosol Wars: A Short History of Defensive and Offensive Military Applications, Advances, and Challenges,” in David S. Ensor, ed., Aerosol Science and Technology: History and Reviews (RTI Press: Research Triangle Park, NC, 2011), p. 333.
12. See, for instance, F. T. Wall, L. A. Hiller, Jr., and D. J. Wheeler, “Statistical Computation of Mean Dimensions of Macromolecules. 1” The Journal of Chemical Physics, vol. 22, no. 6 (June 1954), pp. 1036-1041; F. T. Wall, R. J. Rubin and L. M. Isaacson, “Improved Statistical Method for Computing Mean Dimensions of Polymer Molecules,” The Journal of Chemical Physics, vol. 27, no. 1 (January 1957), pp. 186-188. The University of Illinois had received $135,000 from the National Science Foundation for research into synthetic rubber, the largest such grant given to a university under the NSF’s synthetic rubber program; Special Commission for Rubber Research, Recommended Future Role of the Federal Government with Respect to Research in Synthetic Rubber (National Science Foundation: Washington, D. C., December 1955), p. 9.
13. As it turned out, Aiken’s conservative design left the Mark IV significantly slower than other computers; James G. Baker, who ran the Harvard group researching automated lens design, grew frustrated with the speed of the Mark IV (and his access to it), eventually switching to an IBM mainframe at Boston University. See Donald P. Feder, “Automated Optical Design,” Applied Optics vol. 12, no. 2 (December 1963), p. 1214; Gregory W. Pedlow and Donald E. Welzenbach, The Central Intelligence Agency and Overhead Reconnaissance: The U-2 and OXCART Programs, 1954-1974 (Central Intelligence Agency: Washington, D.C., 1992), p. 52 (declassified copy at https://www.cia.gov/library/readingroom/docs/2002-07-16.pdf, accessed September 12, 2018).
15. For an historical and technical overview of SIGSALY, see J. V. Boone and R. R. Peterson, “SIGSALY—The Start of the Digital Revolution” (2016) (at https://www.nsa.gov/about/cryptologic-heritage/historical-figures-publications/publications/wwii/sigsaly-start-digital.shtml, accessed September 17, 2018).
16. Robin Maconie has speculated on the implications of the Bell Labs connections in a pair of articles: “Stockhausen’s Electronic Studies I and II” (2015) (at http://www.jimstonebraker.com/maconie_studie_II.pdf, accessed September 17, 2018), and “Boulez, Information Science, and IRCAM,” Tempo vol. 71, iss. 279 (January 2017), pp. 38-50.
17. For the 205’s film and TV history, see the Burroughs B205 page at James Carter’s Starring the Computer website (http://starringthecomputer.com/computer.php?c=45, accessed September 12, 2018). The Night the World Exploded, written by Jack Natteford and Luci Ward, and directed by Fred F. Sears, was released by Columbia Pictures in 1957.