Lois Cook (Leurgans) was a programmer, operator, and problem analyst. She is pictured here next to the MANIAC’s arithmetic unit in the 1950s. Photo Courtesy LANL
BY NICHOLAS LEWIS
National Security Research Center
Seventy years ago this spring, the first fully electronic computer at what is today Los Alamos National Laboratory came online. Called the MANIAC, it was constructed when electronic computing and its associated industry were still in their infancy. Early computer centers like Los Alamos had to build their own systems and invent the methods for using them. This pioneering effort gave Los Alamos and the few places like it an outsized influence on the emerging electronic computing field, shaping the computing technologies and methods we experience today.
One of the first electronic computers in the world, the MANIAC represented a substantial human investment at Los Alamos to develop the computing methods and tools capable of serving the Lab’s complex problems in weapons science. This meant that people with a variety of backgrounds and skill sets had to invent concepts and practices that are commonplace today, but were new and profound in their implications in the 1950s.
The MANIAC soon gave way to machines from an emerging computer industry, but that was the goal behind creating it. It provided Los Alamos with a mission-critical resource that no one else at the time could provide, while demonstrating to the world the extraordinary potential of electronic computing as a scientific tool.
The MANIAC’s chassis under construction in 1950. Photo Courtesy LANL
That legacy, however, began when computers were humans, performing calculations by hand. By the early 20th century, most computers in the United States were women, many with degrees in the sciences and mathematics. They often became computers because of discrimination in their chosen fields.
When Los Alamos opened in 1943 as a secret wartime Lab to create atomic bombs, the first computers here were women, some of whom had volunteered from the townsite, and others who were members of the Women’s Army Corps.
Because of the time-sensitive work at the wartime Lab, numerical methods took the place of many live weapons experiments. The human computers and a collection of IBM punched-card machines conducted the numerical simulations that made it possible to field two unique weapon designs in only 27 months and in a perceived race against Nazi Germany.
After the war ended in the fall of 1945, Los Alamos continued its hand-computing and punched-card operations. It also borrowed time on the few computing machines available across the U.S., including the ENIAC, an acronym for Electronic Numerical Integrator and Computer, which was the first electronic, general-purpose computer. Computing machines helped to offset postwar staff departures. They also made possible richer investigations of atomic weapons, and the development of fusion, or thermonuclear weapons, often called H-bombs.
The MANIAC’s arithmetic unit nearing completion in 1952. Photo Courtesy LANL
As part of this effort, Los Alamos made breakthroughs in computer-modeling techniques over the late 1940s, including the Monte Carlo method of simulation through statistical sampling, now used widely in myriad fields, such as climatology, computational biology, and even finance.
By 1948, largely due to the complexity of thermonuclear problems, Theoretical Division Leader J. Carson Mark recognized that Los Alamos needed an electronic computer of its own, but there was effectively no commercial electronic-computing industry at that point. The Lab would have to build its own system. Mark enticed the mathematical physicist and Manhattan Project veteran Nicholas Metropolis to return from the University of Chicago with the promise that he could build a computer at Los Alamos.
Rather than start from scratch, Metropolis partnered with mathematician and longtime Lab consultant John von Neumann, who was already constructing a computer at the Princeton Institute for Advanced Study. Duplicating an existing design would save time, and let Los Alamos learn from the IAS computer project’s mistakes.
MANIAC project leader Nicholas Metropolis (standing) and the MANIAC’s chief engineer Jim Richardson in 1953. Photo Courtesy LANL
What’s in a name?
With his wry sense of humor, Metropolis decided on MANIAC as the computer’s name. He joked that it might discourage the trend of naming computers with acronyms, like ENIAC. The MANIAC’s full name, a retronym, was Mathematical Analyzer, Numerical Integrator And Computer. Although George Gamow, a Lab consultant and friend of Metropolis, joked that it stood for “Metropolis And Neumann Invent Awful Contraption.”
Construction of the MANIAC began in mid-1949, with the project divided into hardware and software teams in the new group created to ensure Los Alamos remained at the cutting edge of electronic computing. Construction of the so-called arithmetic unit, where the calculations took place, went quickly because the Los Alamos team avoided mistakes made on the IAS project.
Marjory Jones (Devaney), a mathematician and programmer, is shown here in 1952, punching a program onto paper tape to be loaded into the MANIAC. Photo Courtesy LANL
Meanwhile, the software team was made up of about 10 mathematicians and physicists, about half of whom were female human computers. They included Mary Tsingou (Menzel), Verna Ellingson (Gardiner), Elaine Felix (Alei), Lois Cook (Leurgens) and Marjory Jones (Devaney). They knew how to break down complex problems into their constituent components for efficient processing, a needed skill with both electronic and manual computing. The software team created a subroutine library that saved time when writing programs for the MANIAC and also coded simulation programs for Lab scientists.
Parts of the MANIAC entered testing in late 1951 and became fully operational in March 1952. For good luck against “gremlins,” Metropolis hung a horseshoe next to the computer. If that didn’t work, he hung an abacus in a glass case on the wall. A sign said, “In case of emergency, break glass.” A fire extinguisher sat nearby, as early electronic computers were very good at catching on fire.
The MANIAC itself was about 7-feet tall and 9-feet wide, with about 3,000 vacuum tubes. It weighed around 1,000 pounds.
The MANIAC also had a collection of peripheral equipment, including paper-tape punches and readers to input programming, a reel-to-reel tape drive and a printer for output, an operator’s station to monitor the computer’s status, a bank of car batteries that supplied the computer with power, and a large metal hood that hung over top to provide ventilation.
Capable of 10,000 operations per second, the MANIAC was easily the most powerful computer in the world upon its debut. Fast and possessing a large memory, it opened up a broad range of problems that had previously been intractable to scientists.
Programming the MANIAC meant breaking down a problem diagram into a collection of system instructions, then punching them onto a paper tape. Results were either printed or punched onto a new tape.
The MANIAC’s operators knew the machine so well, they could move data around the computer manually, using alligator clips. They could also hear when the MANIAC made an error, using a radio to listen to the interference patterns the machine generated.
Operators are pictured here in 1952 in front of the MANIAC. The horseshoe on the pillar on the right was hung for luck. Photo Courtesy LANL
MANIAC goes to work
The Lab immediately put the MANIAC to use on thermonuclear problems. The H-bomb was, after all, the main reason for building a computer onsite.
Although most of its time was spent on classified problems, the MANIAC found many unclassified applications. The scientist and programmer Tsingou, for example, created and wrote the program for the world’s first experiment conducted entirely on a computer. Now called the Fermi-Pasta-Ulam-Tsingou Problem after its creators, the problem modeled the vibration of a string, and was an early step in the development of chaos theory. Gamow, who had jokingly suggested the alternate meaning for “MANIAC,” used the computer for early genetic modeling.
The MANIAC was also the first computer to beat a human opponent at chess. Called “Los Alamos chess,” the board was simplified to six-by-six squares, with the bishops removed to reduce computer-processing time. The MANIAC required 20 minutes to make a single move, but it beat an opponent who had learned the game only a week before.
The MANIAC also provided a platform for developing the Metropolis Algorithm. Written mostly by physicist Arianna Wright Rosenbluth, the Metropolis Algorithm greatly improved on the Monte Carlo method, making Monte Carlo considerably more useful and prevalent in many computer-modeling applications up to the present.
Companies like IBM and Remington Rand had initially resisted developing electronic computers, because it was not clear that a demand existed, or that such machines could be made useful.
Because systems like the MANIAC demonstrated a demand for computing, IBM created the 701 computer, which eventually supplanted the MANIAC as the Lab’s main production system, thanks to IBM providing software and support services. This meant that Los Alamos no longer needed to build its own computers. However, the Lab constructed one last system, the MANIAC II, which entered service in 1957 as a computer science R&D platform. The goal to keep Los Alamos on the cutting edge of computing continued with the MANIAC II. The computer underwent continual change and modification before it was decommissioned in 1977.
The original MANIAC was donated in 1958 to the University of New Mexico, where it was refurbished and became a research tool until 1965. Pieces of the MANIAC still exist, and can be found on display in the Lab’s Nicholas C. Metropolis Center for Modeling and Simulation, and at the Bradbury Science Museum, which unveiled an updated supercomputing exhibit last month.
Paul Stein and Nicholas Metropolis play “Los Alamos” chess against the MANIAC. “Los Alamos” chess was a simplified version of the game, with the bishops removed to reduce the MANIAC’s processing time between moves. The computer still needed about 20 minutes between moves. The MANIAC became the first computer to beat a human opponent at chess in 1956. Photo Courtesy LANL