The Early Legacy Of The Network Computer

By Stuart J. Johnston

have a confession to make: I am now, and always have been, a PC bigot.

So now that I've exposed my personal biases, you can take the rest of this column for what it's worth. But while it's early yet to talk about the demise of the Network Computer, I think it's cle ar that it has accomplished what it set out to do, albeit not how its supporters thought it would.

Instead, the nagging fear that users might actually adopt NCs triggered the institutionalized paranoia at both Intel and Microsoft, driving those companies -- and, indeed, the entire industry -- to deliver much of what the NC proponents promise, but mostly on PCs running Windows.

NC advocates would like to centralize systems again and not for some evil intent like keeping pesky users in line. Instead, they want recentralization for good reasons like reining in costs and stopping users from wasting time -- their own and support personnel's -- by monkeying around with their PCs and breaking things. Additionally, installation and maintenance of applications on users' machines is expensive and time-consuming.

At the same time, don't forget why PCs became popular in the first place. The emphasis was always on the "personal" in personal computer -- a name that pre-existed IBM's PC but became widely popula r afterwards. The PC meant no more having to plead with the guys in the glass house for a report that would take them six months to deliver.

Before PCs, users were always in the supplicant's position of needing data immediately to stay competitive. When I was a consultant at Fortune 50 companies in the early 1980s, many of these workers were bringing their own machines in from home, much to the consternation of the IS staffers.

And long before there was a network they were allowed to plug into, or PC-based terminal emulation systems that let them connect to the mainframe, these workers were re-entering data by hand from printed reports that did not have the information in the formats they needed. They then wrote their own formulas for evaluating that data. After all, they were the ones whose job it was to understand what that data meant -- who better to build the spreadsheets to analyze and massage it?

In the process, they began performing tasks that had traditionally resided in the IS departmen t, thus usurping some of the IS department's power. Was that inefficient? Sure. But not as inefficient as waiting six months for an overworked IS staff to get the report done. Users with PCs drove the pace of change, escalating that pace as their competition escalated. And in doing so, those workers revolutionized the way business is conducted.

It also became clear fairly early that personal empowerment itself was as important an aspect of that revolution as the ability to respond to changing business needs faster. Still, the costs associated with PCs have always been a major concern. And the numbers, always bleak, have remained roughly stable: About two-thirds of the cost of having a PC on a worker's desk is eaten up in maintenance, administration, training, and technical support.

Today, with 80 to 90 million new PCs likely to be shipped in the next 12 months, those costs have become a gigantic issue -- and rightly so. But no one should kid themselves. The NC will never return us to the digital equiv alent of "Father Knows Best." There are too many compelling reasons to continue to decentralize -- not the least of which is the advent of the virtual office and telecommuting. That such work arrangements have become so common is why one in every three new PCs sold to U.S. corporations today is a laptop. No NC can address the needs of a mobile or remote work force. They don't function if they're not plugged into the network.

The NC, even if it doesn't become wildly popular, has served a very important purpose: It made the Wintel consortium decide to finally do something about those terrible cost ratios. Without the "promise" of a thin-client device that would cost less than $1,000, Microsoft, Intel and the PC vendors would never have endorsed so strongly the concept of a PC for less than $1,000. Now we're looking at full-function PCs that may cost less than $400, and pundits are declaring that there will be Pentium II machines for $850 by Christmas.

How can we be sure that this would not have occu rred otherwise? I have covered several events over the years where Microsoft chairman Bill Gates praised the PC hardware vendors for continuously jamming more power into the average PC while maintaining the average cost for those machines at between $2,000 and $2,500 over a decade or more.

It was a business model that provided significant power to customers along with significant profits to the vendors. But the vendors saw no need to drive prices below that magical $1,000 mark before the NC showed up and scared them, likewise with multiuser support in Windows NT. Microsoft had been happy to leave that market to Citrix Systems Inc.'s WinFrame and its variants until it looked like NCs might achieve some popularity in the PC displacement market. Now Microsoft is gearing up to ship Terminal Server in the next few weeks.

But this happened partly because the pro-NC, anti-Microsoft forces went out of their way to ignore Wintel to death. They unilaterally decided that users would not need to run Windows appli cations because users would obviously run Java applets instead. How arrogant. By purposely ignoring the customers' need to run Windows, the NC camp undermined its own message, leaving the door open for Microsoft to rush in. In the words of Michael Corleone in The Godfather Part III: "Never hate your enemies. It infects your judgment."

As reported InformationWeek a few months ago " FedEx Moves Away From NC Approach ," Feb. 16, 1998, Federal Express Corp. has basically decided that Windows-based terminals (WBTs) and a Citrix WinFrame/Terminal Server model are what it needs, even though FedEx had been an NC poster child just a year ago.

Predictably, the company that has the most success selling NCs has been IBM. After all, IBM has the largest installed base of terminals out there, and NCs make dandy terminal replacements. For instance, InformationWeek reported in March that Allstate Insurance Co. has chosen IBM to supply NCs for many of it s 45,000 or so insurance agents. But those NCs will replace existing IBM terminals and also provide access to Windows applications via NT server running on add-in boards in AS/400s. That is, it sure looks like a terminal replacement situation that also runs Windows apps.

The PC and Microsoft's dominance remain unthreatened because the NC and WBT market will continue, for the most part, to be a terminal replacement, not a PC displacement, market. Additionally, in its mad scramble to head off the NC bandidos at the pass, Microsoft came up with its Zero Administration Windows initiative, a move to keep all the user's files, configuration information, and programs on the network server. When a user logs on to a computer, the server will make that PC into the user's own machine by copying the necessary files down to it. This will provide the ability to "roam" from machine to machine, as well as to quickly replace a user's machine if it fails.

But some users worry that moving such a large volume of informa tion around on the network will create large bandwidth bottlenecks, and the programming effort for Microsoft to make ZAW rock solid as well as to deal with those bandwidth issues has cost Windows NT 5.0 more lost time. While Microsoft still claims it will get the second beta of NT 5.0 out this quarter, hardly anyone's taking bets that the company will actually ship it by the end of 1998, and many Microsoft-watchers are placing it in mid-1999.

But by scaring Intel and Microsoft, the NC vendors have accomplished what the users could not -- they've made those companies much more responsive to users' needs. Still, they have not destroyed Wintel. So at a time when IT staffs and marketers talk about using data mining to create "a market of one," my money is that the PC stays on top. And Windows remains a fact of life. The users, by and large, appear to want more freedom and power, not less.

AuthorITies Archive

Send Us Your Feedback

Top of the Page

Rich Levin: Run Time Rich fills you in on all of the latest products, issues, and trends in application development.		Stuart J. Johnston: Redmond Watch As our eyes an d ears in Redmond, Stuart gives his perspective on the latest events at Microsoft.		Charles Pelton: Eye On IT Charles explores IT management issues and strategies that business and technology managers face.