How Alan Turing Set the Rules for Computing

Reconsidering Computations

At the time Turing sketched out his ideas, the world had plenty of pretty sophisticated adding machines that would allow someone to perform simple calculations. What Turing offered was the idea of a general-purpose programmable machine. "You would give it a program and it would do what the program specified," Kahn explained.

In the next decade, another polymath, John von Neumann, at the Princeton Institute for Advanced Study, started working on an operational computer that borrowed from Turing's idea, except it would use random access memory instead of infinite tape to hold the data and operational programs. Called MANIAC (Mathematical Analyzer, Numerator, Integrator, and Computer), it was among the first modern computers ever built and was operational in 1952. MANIAC used what is now called the Von Neumann architecture, the model for all computers today.

Returning to Britain after his time at Princeton, Turing worked on another project to build a computer that used these concepts, called the Automatic Computing Engine (ACE), and pioneered the idea of a stored memory machine, which would become a vital part of the Von Neumann architecture.

As well as sparking the field of computer science, the impact his work had on cracking encryption may ultimately have also saved Great Britain from becoming a German colony. People have argued that Turing's work defining computers was essential to his success in breaking the encryption generated by Germany's Enigma machine--work that helped bring World War II to an end.

"By today's definitions, the Enigma was an analog computer. What he [and his team] built was much closer to [the operations] of a digital computer," Rensselaer's Hendler explained. "Essentially he showed the power of digital computing in attacking this analog problem. This really changed the whole way that the field thought about what computers could do."

AI Pioneer

Having defined computational operations, Turing went on to play a fundamental role in defining artificial intelligence -- or computer intelligence that mimics human thinking. In 1950, he authored a paper that offered a way to determine if a computer possessed human intelligence. The test involves a person having an extended conversation with two hidden entities, a computer and a man pretending to be a woman. ("In both cases he wanted pretending," Hendler explained.) If the person can't determine which party is the computer, the machine can be said to think like a human.

"He wanted to put human and computing on equal footing," Hendler said. "Language is a critical skill for humans because it requires understanding and context. If a computer showed that level of understanding then you wouldn't notice the difference."

Google honored Turing with a doodle for his 100th birthday.

The test "has the advantage of drawing a fairly sharp line between the physical and the intellectual capacities of a man," Turing wrote in the original paper.

As IBM's Brown noted, Turing's legacy is still strongly felt today. In his mathematics work, he showed that "there exists problems that no decision process could answer," Hendler said. In terms of computers, this means, "You could never prove for all complicated computer programs that they are correct," Hendler said. "You could never write a computer program that could debug all other computer programs."

But far from restricting progress of computer science, the knowledge of such inconclusiveness paved the way for building previously unimagined technologies. It allowed engineers to create immensely helpful services such as Internet search engines, despite knowing that the answers such services were to provide would not always be complete.

"You have people who say we should never build a computing system unless we can prove it is secure. Those of us who understand Turing say, 'Well, you can't.' So you must start proving some approximation of secure, which starts a very different conversation," Hendler said.

And despite numerous attempts to beat the Turing Test, it still hasn't been done, except within the most limited of topics. That means we will likely be working to meet Turing's benchmarks for years to come.

"You can't say, 'Siri. How are you today?' and expect it to go on from there in any interesting way," Hendler said.

Joab Jackson covers enterprise software and general technology breaking news for The IDG News Service. Follow Joab on Twitter at @Joab_Jackson. Joab's e-mail address is Joab_Jackson@idg.com

Subscribe to the Today in Tech Newsletter

Comments