Let me tell you a story. Long, long ago, in a development environment so far away that it might as well have been a different galaxy, I was a CP/M & MP/M jock. MS DOS (a crude CP/M subset) didn't exist except in Tim Patterson's mind. XENIX (the forerunner of SCO UNIX) hadn't been ported to a x86 environment yet. What ran on Apple IIs was, well, I should be charitable. By 1980 MP/M was multi-user, multi-tasking in its way and suitable for a business environment. There were at least 30 HW manufacturers that built CP/M machines, and most serious applications were being developed for this platform.
Soon the mainstream computer companies started making bids for this emerging market. DEC made a CP/M machine. So did NCR and I believe HP and Sperry-Rand did also. Badly designed and marketed, they were laughed off the market, and promptly discontinued by their respective companies. By late 1980 we had gone 16-bit, with the i8086 and early models of the i80286 CPUs, which were the powerhouses of that day. Then IBM announced the PC.
We looked this thing over, and laughed ourselves sick. Of all the big corporate entries into the microcomputer field, this was undoubtedly far and away the worst. It had an already obsolete 8-bit i8088 CPU, the most benighted memory model any of us had ever seen, a mere 8 HW interrupts, was limited to two serial ports, and didn't even have hard disk support in the HW or in the OS! When we were once again able to stand, we relegated this turkey to the backs of our minds, and went off to work on real computers with real OSs.
About a year later, we couldn't help but notice that, far from disappearing from the market, this gobbler was racking up some big corporate sales, and all of the main CP/M application developers had ported their products to MS-DOS. A year after that, IBM dominated. Another year, and the once mighty CP/M market just didn't exist any more.
I once heard a fellow say about his opinion on some issue, "I'm certain. It may turn out that I'm dead wrong, but I AM certain". Well, I was certain that the PC was no real threat to CP/M, and I was certainly dead wrong. Even my ego feels battered after blowing it that badly, so I took a good look at what had really happened, as opposed to what I had thought was happening.
First, no flavor of CP/M was quite the same. Each HW manufacturer sold a customized version modified to conform to his own HW configuration. Most manufacturers had slightly different diskette formats and several other small items (i.e. serial ports tended to have different addresses and different interrupts from machine to machine). No two printers were compatible, using different command sets and different interpretations of serial and parallel ports, requiring a custom cable for every new device. Stores had to stock a half dozen different versions of the same application on their shelves, and developers had to support all these permutations. Second, there was hardly any support for software developers by manufacturers, no SDKs.
IBM provided, first, an absolutely uniform HW and SW environment for developers, and second, partnering and strong support for these folks, often including bundling and IBM support. The fact that we had the best HW and OS on the market, and they had the worst, simply didn't matter. And third, and possibly the most important, they provided a solid corporate migration path for those companies who already had strong investments in CP/M, by supporting all the major CP/M applications on the PC, and by helping with conversions to the MS-DOS environment. That's why they won, and we lost. And looking back on all this from a 20-year vantage point, we richly deserved to have our butts kicked. We were arrogant fools.
Now, back to the present. I believe strongly that the key to large-scale acceptance of Linux, especially in corporate and government environments, depends on support of existing applications in those environments (which are 95% Windows) in Linux, a solid support environment for Windows application vendors (including Microsoft, especially if MS is indeed split), and an acceptable transition path for corporations changing over.
As a veteran of some rather large rollouts (MS Exchange to 150,000 desktops in Lockheed in a single month, NT to 40,000 discrete locations with perhaps 400,000 computers for the USPS, and NT and applications to 3000 Gap stores via VSAT satellite, no less), I'm quite familiar with the problems of deployment and support on this scale, and with the people who make the decisions in these environments. These folks are not wedded to Windows, or to anything else but the bottom line. I hope someday to direct the deployment of Linux on such a scale, but it's not quite there yet.
Finally back to the main point: I don't dispute anything Mr. Boorshtein says technically. I disagree strongly with his conclusions. The strength or weakness of VB from a technical viewpoint is just as irrelevant as the superiority of CP/M and S-100 machines. UNIX advocates have tried to move UNIX to the non-engineer's desktop for the last 20 years. IBM and Microsoft have swept the floor with every flavor and every company that has tried. The few UNIX survivors defend constantly eroding niche markets, mostly because Microsoft and Compaq get better returns elsewhere and need some opposition to show the anti-trust boys anyway. Don't misunderstand my position here: I'm pro-Linux and pro-Open Source. I merely have something bad to say about absolutely everybody.
I think the important path for Linux is to fully support the Windows Office Suite and all other major Windows applications, yesterday if not sooner. Support of the VB environment is crucial to this. Development of freeware and shareware clones for these applications will follow once the Linux market is established. The day someone like me can walk into a CIO's office and show him that his Excel spreadsheets and macros, his PowerPoint presentations, his databases and his groupware work under Linux, cost less to support, look and feel much the same, interact seamlessly with his existing environment, and cost him a fraction of what he's paying today, that's the day we can start replacing Windows with Linux 100,000 machines at a time.
Kenneth Broll (firstname.lastname@example.org) is a network design consultant in Silicon Valley and veteran of more than 30 years in the industry. He has done work for Sun, The Gap, NASA, Lockheed, Martin Marietta, GE, Honeywell, IBM, Apple, the USPS, Santa Clara County, and Mitre Corporation, among others. He professes to learn by the few mistakes he will admit to.