As web companies and the government decide what to do about the recent Trojan horse-type virus residing on major Web sites, I am reminded of a dilemma the world faced in the early 1960s as population growth outstripped the ability of the food supply to keep pace. Agricultural researchers responded by creating strains of various grains that proved far more productive than the native strains. But the so-called Green Revolution spawned a new problem as the engineered strains drove the native strains to the brink of extinction. Geneticists warned that depending on one strain of a productive grain was dangerous because a single virus could kill everything and again put the world at risk of starvation.
Today, our Net-centric world faces the same dilemma. The recent virus is only the latest outbreak, made worse by the market dominance of just one or two operating systems, browsers, word processors and e-mail programs. Almost weekly there is news of a hacker virus spreading quickly through millions of computers that use the same software. And commercial antivirus programs deal with each virus as it is introduced. That's like battling fire ants by squashing them underfoot as they stream in under the door.
As we move to 24-hour, seven-day-a-week connections to the Internet via cable or DSL, our dependence on just one or two software strains should worry us. Computer viruses will be able to rip across the Internet at 300 kbits/second or more.
To be sure, companies are tackling the problem. IBM's Immune System for Cyberspace is a global distributed system that can find and capture an unknown virus. The virus is sent over the network to an automated analysis center. There, cure information is extracted and the antidote delivered to all others on the net in minutes.
Symantec and McAfee plan to commercialize similar approaches. But I believe an industrywide effort is needed.
When will it happen? I see three possibilities: One of the antivirus vendors may beat the others into submission and become the de facto standard; the industry may get together and agree to a common, coordinated approach (yeah, right); or the government will step in and say "Do something or else we will"-a likely possibility, considering how integral Internet-based commerce and communications are becoming and how panicked Washington is.
Until then, I think software balkanization is the safest bet. And maybe it is time to put a hold on the idea of a language such Java or ActiveX that writes once and can run everywhere. The thought makes me shudder.
Bernard Cole is EE Times' Managing Editor for Embedded Design and Netcentric Computing.
RSS
RSS| 
Digital| 
Mobile
