Tech's Reliability Gap

Scattering employees may keep them safe, but networks aren't up to the task

It sounded like a reasonable response to the devastating attacks of September 11. Companies talked of reducing air travel, increasing their use of technologies such as videoconferencing and collaboration software, and further decentralizing operations through networks such as the Internet. In all of these measures, there's an assumption that technology is the key to making businesses not only safer but more stable and resilient.

Boy, I'm not so sure. When is the last time you said, "Wow, I just can't believe how dependable my personal computer is!"?

There's no doubt that computers, software, global networks, and the like have made us more efficient as corporations and as individuals. Tom DeMarco, a fellow with tech consultant Cutter Consortium's Business Technology Council, notes that some security measures, such as dispersing people and computers and using networks to connect them, have made us even more dependent on information systems. But all too many of these technologies are not only wide open to attack by hackers and fanatics, they're also absurdly unstable and unreliable. Relying on them still more, without making huge improvements in how they're designed and used, seems certain to compound the disruptions we've already seen.

The root problem is the very nature of technology development, notes Richard D. Pethia, director of the Computer Emergency Response Team Centers at Carnegie Mellon University's Software Engineering Institute. "Technology evolves so rapidly that vendors concentrate on time to market, placing a low priority on the security of their products," he said in recent testimony before a House subcommittee.

Even more worrisome is the low priority placed on basic reliability. One of the biggest challenges to distributed corporate operations may be bad software. According to information technology consultant Standish Group International Inc., bad software last year cost about $175 billion worldwide in lost business due to downtime, more than double the amount two years ago. It makes you wonder if companies are ready to push more people into remote offices or even home to telecommute, where technical support is shaky or nil. I can't tell you how many times my PC has inexplicably crashed, requiring on at least two occasions that it be sent to techies across the country to fix.

To the Internet's credit, key services such as e-mail and instant messaging held up amazingly well in the hours after the terrorist attacks. But the very strengths of the Internet may well increase the likelihood of widespread disruptions in the future. The nearly unstoppable global connections it forges among some 110 million computers online make it far easier to spread viruses and software bugs from a few computers to millions in an instant. Private networks may be better equipped to limit the impact of attacks and glitches. But even there, remote and small-office setups often use PC technologies that are more susceptible to viruses and just plain-old breakdowns.

This shaky situation suddenly makes cyber-attacks much less farfetched than they once seemed. Viruses set loose by bored teenagers continue to bring corporations to their knees. It's not hard to imagine the havoc that a dedicated group of fanatics could unleash on the systems that we intend to depend on more than ever. Moreover, says Standish Group Chairman Jim Johnson, disaster recovery plans at all but two of 56 companies he surveyed recently are out of date, typically allowing them to return only 10% of their critical software applications to operation quickly. Says Johnson: "Disaster recovery is a disaster."

Most of these problems require big changes not only in technology, but in our thinking. Companies must make technology reliability a top priority and demand the same of their suppliers. Let's not fool ourselves: None of this can be fixed overnight. But I can't think of a better time to start.

By Robert D. Hof

    Before it's here, it's on the Bloomberg Terminal.