Computing needs a Grand Challenge
Sir Tony Hoare lays down the gauntlet
Sir Tony Hoare - British computing pioneer and senior scientist at Microsoft Research - believes the computer industry needs a "grand challenge" to inspire it. In the same way that the lunar challenge in the 1960s sparked a decade of collaborative innovation and development in engineering and space technology, or the human genome project united biologists around the globe, so too must computer scientists pull together on such a scale to take their industry to its next major milestone.
Speaking last Tuesday at an open day at Microsoft Research's lab in Cambridge, Hoare told the audience of around 60 journalists and analysts that there are seven such challenges facing researchers today. Significantly, these are not purely computational challenges, but involve a mix of disciplines from biology and psychology, right through to quantum physics. This reflects how much other areas rely on and use IT to support their research, but also the changing nature of computer science itself.
By 2020, Hoare predicts, the world will contain 100 times as many computers as it does now, each with 100 times as much power and memory, all interconnected. And to best understand this world, he says, we should not think of it as containing many discrete computing devices, but as a global ubiquitous computer (GUC).
He argues that in this world, the classical theory of computation, based on Turing's description of a single, localised machine sequentially executing a deterministic program to completion, no longer applies. One of the grand challenges, then, is to re-write the basic foundations of the science, to find a theory of computation that is "more realistic than the Turing model, and can take into account the discoveries of biology, and the promise of the quantum computer".
"Computations carried out in nature, for example in the brain and body of a living organism, are nothing at all like that. They are widely distributed over space and over time; they essentially involve massively parallel operation; they involve continuous interaction with their environment; and they are highly non-deterministic," Hoare says. In this way, the global ubiquitous computer is much more like a living organism than the Turing machine.
Wanted: life models
The links between the computing and biological sciences don't stop there. Perhaps the grandest of the grand challenges for computing are about modelling life, in particular, developing a model of an organism that will make predictions that will be experimentally testable.
To make progress, Hoare suggests the project "will probably concentrate on the same simple organisms that are the subject of widest biological experiments, for example the thale-weed Arabidopsis Thaliana".
But Hoare has grander plans still.
"An ultimate joint challenge for the biological and the computational sciences is the understanding of the mechanisms of the human brain, and its relationship with the human mind," he says.
"A single human brain has about a hundred million million nerve cells...and a computer program that throws light on the mind/brain problem will have to incorporate the deepest insights of biologists, nerve scientists, psychologists, physiologists, linguists, social scientists, and even philosophers. This challenge is one that has inspired Computer Science since its very origins, when Alan Turing himself first proposed the Turing Test as a still unmet challenge for artificial intelligence."
The Cambridge facility is one of five facilities run by Microsoft Research, an independent sub-section of the software giant. The other labs are based in China and the US, with the bulk of the research coming out of Redmond. The group's remit is to research pure computer science, rather than to develop products.
Hoare, who was knighted in March 2000 for his services to computer science, joined the lab when he retired from Oxford University. Among his achievements in his career in industry and academia, is the development of the first commercial compiler for the programming language Algol 60.
He argues that the approach to Grand Challenges, in any discipline, is driven primarily by scientific curiosity and idealism and a desire to understand basic phenomena, in this case computational phenomena.
"It is easy to predict that some of the discoveries of research directed towards Grand Challenges - but only the most unexpected ones, and at the most unexpected times - will be the basis of revolutionary improvements in the way that we exploit the power of our future computing devices." ®
That list of Grand Challenges in full.
Sponsored: RAID: End of an era?