And that, in a nutshell, is the grand challenge for advocates of thin-client computing. A wave of research--and a measure of common sense--tells us that centralized management of thin-client computers, be they WBTs (Windows-based terminals) or more sophisticated NCs (network computers), will save more money than the current paradigm of a system administrator, often masquerading as an innocent technology user, at each and every desktop. Maybe it's not quite so bad in your shop--but even in the best-managed environments, that networked Wintel PC sometimes spins out of control, leaving a few casualties in its wake.

How did we get into this mess? It's really quite simple. Way back when, Microsoft Corp. chose to optimize its products for standalone computers by requiring that all system and application software be loaded from a local disk drive, thus sowing the seeds that today give us the weeds of high cost of ownership, inadequate security and diminished reliability. Speculation flourished about why Microsoft would inflict so much pain on IT managers. Perhaps Microsoft simply lacked network expertise early on, and by the time its Windows designers finally "got it," they had created such a legacy of standalone operations that it's been too difficult to change course, despite dramatic improvements in network performance that could justify a more network-centric approach. Microsoft's own NetPC and Zero Administration Windows initiatives acknowledged the need for change, though the company's commitment to this path is still uncertain.

Despite Microsoft's apparent inability to develop a computing environment system that managers can learn to love, give the crew in Redmond credit for coming up with one that makes most computer users feel comfortable. Within every organization, pockets of power users, some in senior positions, insist on having a PC on their desktop that can carry on if the network fails, even if that network has a near-perfect track record of reliability. Combine that preference with the bad memories of mainframes and glass houses dredged up by the mere mention of centralized administration of desktops, and you have an environment that appears inhospitable, at best, to a new model of network-based computing.

But take heart, all you advocates of thin. Thin client's time will come. Maybe not this year or next, and maybe not in exactly the form you would prefer, but the compelling benefits of thin-client computing can be realized.