Charles Shorb, an ISU graduate student, works on a nearly exact replica of the world's first electronic digital computer. (Photo by Gary Fandel, The Register)
by Thomas R. O'Donnell
of The Register's Ames Bureau
August 1997
Charles Shorb, an ISU graduate student, works on a nearly exact replica of the world's first electronic digital computer. (Photo by Gary Fandel, The Register)
Charles Shorb says getting a replica of the world's first electronic digital computer to work is a little like the time he rebuilt the engine of his 1980 Volkswagen Jetta. But at least there were directions for the Jetta.
There are no similar instructions for assembling and tuning up the Atanasoff-Berry Computer. Shorb, fueled by Coca-Cola and Hostess Fruit Pies, has spent the last two months in the basement of Spedding Hall on the Iowa State University campus, sorting out a maze of wires, capacitors, resistors, vacuum tubes and rotating drums.
He's nearly finished the job.
"The logic runs, and we can demonstrate it and the memory works. . . . That proves the concept," said John Gustafson, who is overseeing Shorb's work. Gustafson is an ISU professor and researcher at the Ames Laboratory, a federal installation operated by the university.
The success so far is largely thanks to Shorb, Gustafson said. "Charles produces the work of three engineers."
The pleasure's all his, said Shorb, who turns 27 this week. The ISU computer science graduate student told computer chip maker Intel it will have to wait until September for him to start his new job so he can spend the summer working out the computers kinks. If Intel hadn't delayed his start date, Shorb said, he would have turned down the job in a second. "How many times in my life am I going to be able to work on the first computer?" he asked. When the replica goes on display, "I'm going to be able to say I made that thing work."
What Shorb is fixing is a nearly exact replica of the computer built in the late 1930s by ISU mathematics professor John Vincent Atanasoff and graduate student Clifford Berry. The complex, desk-sized machine was designed to solve 29 simultaneous equations with 29 unknowns, but in actual operation solved just five equations with five unknowns.
Atanasoff and Berry never patented the machine, and its parts and plans were scattered. But in 1973 a judge declared it to be the foundation of later computers.
Two years ago, a group of Ames Lab researchers and technicians set out to build a working replica of the machine. With more than $300,000 in donated money, they scrounged parts and pieced it together, sometimes guessing how it all fit. The machine was turned over to Gustafson, Shorb, and Ames Lab technicians John Erickson and Gary Sleege this spring.
"My job is to get the machine working. That is my sole goal," Shorb said.
The job has proved tougher than expected. Shorb had to dismantle much of the device to correct errors made last summer, when workers rushed to get it ready for display.
Shorb has spent hours cleaning up faulty soldering and minute electrical shorts. The almost complete absence of schematics and blueprints made the job even more difficult.
"There are hundreds of wires running around, all colored black," he said.
Shorb's work is paying off. When he turns it on, the computer purrs like a sewing machine as electrode-studded drums spin and lights go on. And there was a breakthrough late one night when the computer successfully solved a simple equation.
"It was almost like Atanasoff and Berry were in there, it was so exciting," Shorb said.
So far, he's been unable to duplicate that feat, but he insists the computer is nearly ready. "We could put the machine together tonight . . . and it could be working. Or it could elude us for another two weeks."
Of course, whether the machine is working depends on how one defines working. If it successfully solves five unknowns, it would do as well as the original. But Shorb hopes to get the computer to reach its capacity and solve for 29 unknowns and one constant--something that would take the machine an entire day.
"I want to light that sucker up and see how much power it drains," he said.
Gustafson said there are tentative plans to unveil the working computer at the National Museum on Natural History in Washington, D.C., on October 8, 1997. The machine probably will go on tour for at least a year before returning to ISU for permanent display.
Atanasoff's son, John V. Atanasoff Jr., chief executive of a Colorado medical device manufacturer, will get a look at it this weekend, when he visits ISU.
"They're here. This is them," he said, gesturing toward the device. "They're the cornerstone of computing."