VX Heaven

Library Collection Sources Engines Constructors Simulators Utilities Links Forum

Can Computer Viruses Be Domesticated to Serve Mankind?

John Markoff
The New York Times
ISSN 0362-4331
October 1991

[Back to index] [Comments]

Ideas and Trends

Biologists have learned to harness viruses to create vaccines and, in recent years, to reprogram faulty chromosomes by using viruses to smuggle new genes into cells. Now a small but growing group of computer scientists is examining the possibility of designing computer viruses and similar programs called worms to burrow into computer networks and set in motion a whole range of beneficial activities.

Many computer users have been the victims of malicious virus programs propagating through networks and erasing data or causing the whole system to fail. But now some researchers are suggesting that it is possible to harness the subtle power of computer viruses to perform useful tasks.

They imagine self-propelled programs gliding through networks of hundreds of thousands of computers, automatically updating software, diagnosing hardware problems, seeking out information stored in vast data banks or doing routine garbage collection -- mopping up the residue of leftover calculations that can clog up memory banks.

"Although malicious viruses certainly must be brought under control, it is also imperative that the fear of viruses not keep computer scientists from exploring the many benefits they could bestow," Fred Cohen, the scientist who coined the term "computer viruses," wrote in the current issue of The Sciences, published by the New York Academy of Sciences.

Mr. Cohen has offered a $1,000 annual prize for the most useful computer virus. Many computer security experts think this is one of the worst ideas they have ever heard.

"It's like publicizing in the paper that anyone who comes up with the best use for a handgun should be given an award for firing it within the city limits," said Eugene Spafford, a Purdue University computer scientist. Computer security experts like Mr. Spafford say they are not opposed in principle to computer virus research. But they argue that in a world increasingly dense with computer networks, software designers must rethink the potential hazards of releasing autonomous programs into the electronic milieu. "It's just as if we're doing biological research," Mr. Spafford said. "We wouldn't encourage somebody to work on recombinant DNA in their garage."

The power of computer viruses for both constructive and destructive uses is growing. Last May, for example, the United States Army Signals Warfare Laboratory issued a $550,000 contract for the study of viruses as military weapons.

Mr. Cohen was one of the first academic researchers to openly study computer viruses. As a graduate student at the University of California at Los Angeles in the early 1980's he conducted experiments showing that viruses could be useful. But when he set loose a virus to demonstrate to campus administrators that their systems were vulnerable, their response was to shut their machines to further experiments.

In his article, Mr. Cohen described a virus-based bill collecting program he has invented. In a traditional bill collecter, a single program, often cumbersome and complex, monitors all accounts and sends out overdue notices. In Mr. Cohen's system a simple virus program is spawned each time a customer makes a transaction. Every month the virus checks to see what its customer owes and dispatches a bill. Once the account is settled, the collector deletes itself.

Useful as such devices might be, other experts are uneasy about the idea of distributed systems, in which friendly viruses wander through a network, performing maintenance chores or borrowing computer time here and there to solve especially difficult problems. Mr. Spafford says there is always an ethical question raised when a program is written to run on somebody else's computer.

"There is no easy answer to this problem," said Peter Neumann, a computer security expert at SRI International, a research center in Menlo Park, Calif. "Distributed computing has enormous potential but you can use it destructively. It's a two-edged sword."

Computer viruses actually started out as helpers rather than invaders. The precursor of modern software viruses and worms is probably a simple game called Darwin, invented in 1962 by two Bell Laboratories computer scientists, Doug McIlroy and Victor Vyssotsky. The two researchers designed a game in which programs competed within the computer's memory. The winner was the one that could kill all its opponents by erasing them.

The research was carried a step further by scientists at Bolt, Baranek & Newman, a computer company in Cambridge, Mass. In 1971 a scientist there, Bob Thomas, was experimenting with software for an air traffic control network. The idea was to monitor a plane as it flew across the country and automatically move control of the flight from one computer to another. In a simulation of the system, Mr. Thomas wrote a program that crawled through the network and popped up on each computer screen, leaving the message "I'm creeper! Catch me if you can!" Later another program, called a reaper, was written to move through the network and kill creepers.

A decade later two Xerox computer scientists began doing similar work at the company's Palo Alto Research Center in California. John Shoch and Jon Hepp wrote a series of computer programs that would scurry around the laboratory's network performing helpful tasks. They named their programs worms after similar programs that had been described in a science fiction novel.

One of their most powerful programs was called a vampire worm. It was designed to hide until the human users had left for the night and then come to life and parcel out a large calculation to computers all over the network. In the morning, as the computers were reclaimed, the vampire program would shrink away and wait until the machines were freed again the next evening.

Carrying the biological metaphor even further, scientists today are experimenting with systems in which software that would be too complex to be easily designed by human minds arises by a process akin to Darwinian evolution. They start with a seed program that is changed slightly each time it runs. Most of these random mutations don't help, but every once in a while the system hits upon a slight improvement. After millions of these tiny adaptations, a complex program is born.

Danny Hillis, the cofounder of Thinking Machines Corp., a maker of supercomputers in Cambridge, Mass., has used the company's machine to evolve powerful sorting programs. Mr. Hillis reasons that in the future it may be possible for computer operating systems to evolve that are more resistant to viruses.

But he suggests that there may be hidden risks in such research. Programs evolving in computers may function in a manner not easily recognizable to human programmers. And such creations could be dangerous if let loose in a computer network. "It's a little bit like trying to study useful weeds," Mr. Hillis said. "As long as you do it in your own garden it's O.K., but you shouldn't do it in other people's gardens."

By accessing, viewing, downloading or otherwise using this content you agree to be bound by the Terms of Use! aka