I LIVE with a psychotic answering machine. It has a habit of occasionally and unpredictably answering phone calls on the first ring. It鈥檚 always polite, true 鈥� as soon as I answer the phone, it hangs up, loudly saying 鈥淭hank you for calling鈥� in smug, synthesised tones. Later it sulks, and doesn鈥檛 answer calls at all 鈥� even on the fourth ring, when it鈥檚 supposed to. Lord knows what it gets up to when I鈥檓 not there. A contact at Panasonic says only that these are 鈥渆xcellent鈥� answering machines.
I was thinking about this the other day when I read that Intel, the company whose chips are the brains (if that鈥檚 the word I鈥檓 looking for) of most PCs, has finally agreed to issue a replacement for its flawed Pentium microprocessor chip to anyone who asks for one (Technology, 7 January).
This decision came after about a month of increasingly bad publicity for Intel, which all began when Thomas Nicely, a professor of mathematics at Lynchburg College in Virginia, checked some calculations and found errors he traced specifically to his Pentium machine. According to the Associated Press, Nicely sent an e-mail to several colleagues telling them what he had found. A colleague posted the letter on the CompuServe on-line service, where someone else distributed it on the Internet, the global computer network, and over the following few weeks, to the international news media (Technology, 10 December 1994).
Advertisement
Several software companies came up with fixes that involved turning off the section of the processor called the floating-point unit, slowing things down. Intel eventually admitted the flaw was there, but said it should only show up once every 27 000 years for people doing 1000 divisions a day. The risk area is long division calculations for about 1700 combinations of numbers (out of billions) with between 4 and 19 digits to the right of zero 鈥� hence Intel鈥檚 insistence that it would affect only a few scientists and mathematicians. In fact, contacts tell me that the financial world has been worried for some time by the reliability of the chips in its calculating machines. And at least one computer graphics person is convinced that the Pentium is causing some animation routines to crash.
Intel then said it would replace chips for users who could show they needed them for high-precision mathematical and engineering work. IBM carried out some tests and then stopped using Intel鈥檚 chip, saying the error was likely to occur as often as once every 24 days (Technology, 24/31 December 1994). Rumour then spread that IBM might be trying to promote the rival PowerPC chip which it is developing with Motorola and Apple, while criticising Intel鈥檚 stance. But an independent test carried out by two American magazines, PC Week and PC Magazine, concluded that the chance of error was more like once every two months to ten years, depending on usage.
When Intel finally gave in and agreed to replace the chip for anyone who asked, most commentators said that was what it should have done in the first place. But a few people complained that the decision set a poor precedent as usres would now have unrealistic expectations of technology. It鈥檚 a subtle point: until this case, the industry had computer users nicely trained to believe that nothing would ever work 鈥渁s advertised鈥�. Now, users would expect products to perform as claimed.
Some cynics might say that it would have been more typical of the computer industry to follow its standard operating procedure of fixing products that don鈥檛 quite work by selling customers an upgrade. They might also point out that when Sharp received complaints about the screen and word processor on its touch-screen personal organiser 鈥� sold in Britain as the IQ-9000 and in the US as the Wizard 9600 鈥� it released a new model with revisions and twice as much memory. Disgruntled customers (in the US only) were offered a special replacement deal for about $200. Of course, software companies have sometimes been bludgeoned by the sheer fury of their customers into giving out a free 鈥渕aintenance release鈥�.
I say blame the Internet: until the massed users of any given computer product had a forum in which they could find each other and compare gripes, it was an easy matter for a company to say: 鈥淣o one else has complained.鈥� Until users turned the 鈥淧olaris forum鈥� on CompuServe into a boisterous, vituperative pit of vengeful farewells, journalists were even kind to Polaris鈥檚 Packrat 5.0, which in its early version was one of the more notoriously bug-ridden releases of 1993 (there have been maintenance releases since, and the British product is relatively stable). Similarly, it was the Net and the resulting spread of bad publicity that pushed Intel into issuing replacements.
You can use this sort of thing to your advantage: check the Net for the questions and problems of users grappling with the product you鈥檙e thinking of buying before you spend your money. Let other people find their work being delivered to a flock of piranhas cleverly hidden in the latest system software. Wait for a later release and count your blessings.
Even if the Pentium flaw is not serious, Intel鈥檚 decision may be a sign that the industry is finally growing up, and that users are becoming sophisticated enough to demand the same level of reliability that they insist on for cars or any other product. People in the industry are fond of assuming that a computer is obsolete after two years. But most consumers expect to be able to resell a machine and receive at least a few hundred pounds for it. And for consumers, that鈥檚 when a flawed chip creates a real problem.
Most of the time it鈥檚 those small, daily niggles that get to you. These are things the exact cause of which you can鈥檛 trace, with the result that you start to believe you are starring in a remake of the 1944 movie Gaslight. In it, the part of the husband, who is deliberately trying to drive Ingrid Bergman mad, is taken by the computer. For ordinary users, the Pentium flaw may still prove to be in that category 鈥� some friends running the game Doom on a Pentium are getting some strange effects where the screen suddenly starts spinning and won鈥檛 stop. They can鈥檛 tell if it is the chip 鈥�
Of course, there are lots of other things like that which don鈥檛 get replaced for free: my experience of Windows is that it runs out of resources every other day; a weird glitch in the phone system one night which led it to connect me to a phone number in US area code 617 instead of 607; the way the encryption on Sky takes whole, maddening minutes to clear when it unexpectedly interrupts the movie you鈥檙e watching. There鈥檚 really nothing you can do about things like that, except try to get out more. I鈥檓 considering replacing my answering machine with a mobile phone.