In one of the many memorable exchanges to emerge this week from Microsoft's increasingly absurd confrontation with the U.S. government, District Court judge Thomas Penfield Jackson asked a Microsoft vice president: "It seemed absolutely clear to you that I entered an order that required that you distribute a product that would not work -- is that what you're telling me?"
"In plain English, yes," the Microsoft exec responded.
Perhaps Judge Jackson might not have been so flummoxed if he understood, as millions of PC owners and users do, that distributing products that don't work is standard operating procedure in this industry anyway. You ship products that don't work to seize market share so that some other company can't get a foothold. Or to be able to tell business customers that your company offers a full suite of services. Or to placate impatient investors. "Does it work?" has always been fairly low on the computer industry's list of important questions -- as a refreshing rant this week by New York Times columnist Stephen Manes reminds us.
I thought of Microsoft's court battle and Manes' column as I flipped through my latest issue of PC Magazine. The flagship Ziff-Davis magazine has always been a reliable booster of the newest processors and gadgets. You turn to PC Magazine when you want detailed comparisons of dozens of highly similar products, like this month's survey of high-end notebooks -- not when you want answers to broader philosophical questions, like, "Why are PCs still so hard to use?"
Yet there on Page 93 was columnist Jim Seymour complaining about what he called "The New IRQ Scandal." The headline caught my eye because I have had a chip on my shoulder about IRQs ever since I tried to install my first modem in my first PC a dozen years ago.
Whenever you install new hardware in a PC -- a modem, an expansion card or a new drive -- you have to assign it to one of a finite number of IRQ addresses, or "interrupts," or it won't work. (IRQ addresses allow a PC's processor to talk to different pieces of PC hardware.) In the old days this was accomplished with great difficulty, via toying with jumpers on circuit cards; it was often difficult just to ascertain which IRQs were in use. Today's "Plug and Play" architecture theoretically streamlines this process -- though just last year I found myself plunged back into the nightmare realm of IRQs as I tried to install a new modem on a 1995-model Pentium running Windows 95. The headache of dealing with IRQs is precisely the sort of thing that Macintosh users point to in touting the advantages of their computers -- and it's one point even the most dedicated PC lover must grant them.
A 1985-vintage PC/AT made 16 IRQs available (most of which were already claimed by stuff like your keyboard and your system clock). Though the dozen years between then and now have seen vast advances in every aspect of PC hardware and software, the IRQ lineup hasn't changed a bit.
And so Seymour's column points out a hilarious new bind the PC industry finds itself in: With a proliferation of new multimedia devices (like TV tuners and DVD drives) being piled upon the aging PC infrastructure, some manufacturers are now shipping brand-new PCs with no free IRQs -- they're all in use from the moment you pull the computer out of the box. What this means is simple: There may be free drive bays in the computer, and free expansion slots on the motherboard, but there is no way to expand this computer short of disconnecting one of the devices that it came with.
IRQs are precisely the kind of boring, geeky detail that technology companies never mention as they promote their hot new products. Imagine if every ad for a PC peripheral included a warning like: "You must have at least one free IRQ on your PC to be able to install our SuperModDrive Ultra Pro." Instead, the industry follows a strict "caveat emptor" approach: Each company sees its role as making sure only that its own products work; the problem of getting stuff to work together is dumped in the user's lap. This is the way it was in 1985, and this -- despite the promise of new standards like the Universal Serial Bus -- is still the way it is today.
But except for rare outbursts like the Seymour column, you won't find many complaints about problems like the IRQ crunch. The computer industry is used to having its own way with the terminology surrounding its products. And so the IRQ situation -- which an impartial observer might well see as a case of antiquated specifications hampering system expandability -- can quickly be given a positive spin: IRQs are an example of the PC platform's resilience, an indication of the flexibility of its open hardware architecture!
Which brings us back to Microsoft and the Justice Department. As the dispute has played out in court so far, the central issue is how to define what's involved in "removing" Internet Explorer from a computer. The government maintains that "removing" IE means removing the central browser program and the user's access to the program via the icon on the desktop. Microsoft says that to "remove" IE one must also remove dozens of obscure helper files (".DLLs") that most users will never encounter -- and, the software giant warns, if you remove these files you will cripple the whole operating system.
But what most people mean when they say "Internet Explorer" is simply the browser program they use by that name -- chiefly, a program file called "iexplore.exe." The whole point of .DLLs is that they are shared libraries, pieces of code that one program can install and make available for other programs to use. The same .DLLs are routinely installed by more than one program. If you use a commercial "uninstaller" program to "remove" a program like Internet Explorer, it will usually warn you not to remove .DLLs.
Microsoft has been the leading player in the PC industry long enough now that it is thoroughly accustomed to revising its definitions on the fly according to its needs of the moment. Thus Internet Explorer can one minute be a stand-alone retail product and the next it's an integral part of the operating system. One minute, certain .DLLs are helpful utility files that many programs can access; the next, they are an inseparable part of a single program package.
Like the cagey manufacturers who ship computers with all their IRQs used up, Microsoft plainly expects that the general public is too easily cowed in the technical arena to see through this smoke screen. What's most gripping about the unfolding drama in court is that, for the first time in a long while, the company is getting called on its double-talk.
Whatever else they achieve, the government's lawyers and the court deserve a round of applause for standing up to Microsoft's arrogance and paternalism (traits it shares with much of the computer industry). From a company this powerful, "We know what's best for you" and "You wouldn't understand" are simply not acceptable responses to legitimate public inquiries.