Friday, Jul 25, 2008
So what can we do?
"While stories of bad software killing people are still rare, they exist and may portend the future," James Morris, dean of the Carnegie Mellon School of Computer Science, recently argued in an essay proposing a joint research and education push by universities, government and industry to improve the dependability of computer systems.
He suggested calling this the High Dependability Computing Consortium (HDCC), warning that "as we entrust our lives and livelihoods to computers, many systems will effectively become critical ... Even a simple word processor can become mission critical if it crashes a few minutes before the courier pickup deadline for a proposal submission. It is vital that even everyday, seemingly non-critical applications be raised to a higher level of dependability to replace the enormous hidden costs their unreliability levies on businesses and individuals."
But set against the practices of the personal computer industry, Morris' perspective might easily belong to another galaxy.
Common sense would seem to suggest that measuring defects is a vital first step toward eliminating them. Incredibly enough, no one collects (un)reliability statistics for personal computers and software. "There's no independent body today collecting numbers and the companies themselves certainly won't give out any information about the performance of their products," says Gartner's Knox.
Common sense would also seem to hold that knowing the root causes of breakdowns is essential in preventing recurrences. Yet, in a short paper Knox sent me, he wrote, "PC problems are very rarely diagnosed. Rather, ad hoc solutions are immediately applied." The accepted industry practice, he said, is "fix rather than diagnose ... For example, rather than diagnosing and fixing a specific system-level problem, the [computer company] might immediately apply a new motherboard assembly and BIOS, never determining what the root cause of the problem was."
It's as if Kawasaki's "don't worry, be crappy" advice about the development of products had become a license for slipshod work in every sphere of computer companies' operations; carelessness that freely wastes these firms' own resources.
Even a mere computer-user can see this. Let me explain. Twice, between May and October of this year, Dell, America's biggest direct seller of personal computers to consumers, had to replace the modem in a $3,600 portable machine still less than two years old that replaced its predecesor, a lemon, within six weeks of my original purchase. (I will leave the reader to imagine what weeks of pointless troubleshooting, assisted by Dell's telephone tech support -- first on behalf of one computer and then each of two modems -- did to deadlines for my own work.)
After my second Dell modem failed, the company was supposed to send its on-site service tech to my house with a new motherboard as well as a third modem. But someone at a Dell warehouse somewhere in Texas failed to put the motherboard on the plane, so that the same service tech had to wait for another expensive overnight shipment and pay me another call two days later. When changing both components did no good and I was asked to let Dell fly my machine to its repair shop in Tennessee, my local Airborne Express truck driver also had to make a second trip to my house because the reference number a Dell employee gave the air-freight company did not match the one she had given me. And at the end of all this, when I finally had a working computer again, no one at Dell could tell me why the cause of the latest episode of trouble was essentially unknown. ("Fix, don't diagnose.")
One more example of monumental inefficiency and waste. In the early autumn, two separate faults in my Hewlett Packard printer -- slightly older than a year -- led to hours-long tech support calls over several days, at the end of which the company's tech support staff told me I needed the latest version of its software driver, and that I'd have to exchange my printer for a new one they would send out.
The replacement printer arrived the next day, shipped by Fed Ex even though it was perfectly useless without the new driver, for which I was told I'd have to wait a week. A kind HP tech support lady, Joann Osier, took pity on me when -- after three weeks in which I had two printers but could print nothing -- I still hadn't received the promised software CD. She said she would simply Fed Ex me a sample copy she happened to have sitting on her desk, and she did. I thought of that good Samaritan with special gratitude nearly two months later, when the software promised nearly three months earlier finally appeared on my doorstep: It had been shipped by (priority overnight) Fed-Ex.
Perhaps the biggest irony in all this is that the shoddiness of high-tech products means that people don't use more than a very small fraction of the innovations developed at breakneck speed that are supposed to justify high-tech sloppiness. For a start, many of these have more style than substance -- what computer scientists are calling "feature-itis," "glitzware" or, in a pointed reference to the products of late-1950s Detroit, "tail fins" and "flashy chrome bumpers."
But that's not the worst of it. In a syndicated newspaper column published in the Los Angeles Times on Nov. 27, Gary Chapman, director of the 21st Century Project at the University of Texas and an authority on the social implications of new developments in information technology, noted that "repeated experiences with software glitches tend to narrow one's use of computers to the familiar and routine. Studies have shown that most users rely on less than 10 percent of the features of common programs as Microsoft Word or Netscape Communicator. It takes a high toleration for frustration and failure to explore beyond the boundaries of one's own comfort level ... It also calls into question how much money and energy we spend on new software features that most people don't use or even know about."
In the essay by Carnegie Mellon's James Morris on computer dependability, he wrote, "In the 1950s, what was good for the car industry was good for the U.S. ... As with car quality in the 1950s, it is widely argued that it is a disservice to stockholders to make software more reliable ..." I e-mailed him a few days ago to ask what he thought it would take before this state of affairs ends. In his reply, he mentioned that a number of Silicon Valley companies, as well as NASA and his university, would be launching his brainchild, the High Dependability Computing Consortium, on Monday. But he didn't promote his consciousness-raising effort as the most likely agent of change (and he gave no answer at all to my supplementary question about whether American computer or software companies -- as opposed to universities and government agencies -- had committed serious sums of money or resources to the project.)
"Not until the consumers demand [quality] and get it from overseas will the reigning companies believe," he e-mailed me. "American computer and software companies are making too much money in the current environment to care."
Get Salon in your mailbox!