Scott McNealy was on a roll. The chief executive of workstation maker Sun Microsystems was giving a speech singing the praises of Sun's Java programming language. He said Java would enable the development of a new class of "network computer," or NC, that would inexpensively give untold millions access to the Internet. Breaking from his text, Mr. McNealy practically shouted: "These computers will cost less than $500!"

Ordinarily, that claim would have faded just as fast as he invented it. Mr. McNealy makes wild predictions hourly. Outrageousness is much of his charm.

But, in this case, the claim fit neatly with the needs of Larry Ellison, chief executive of Oracle, the giant maker of database software. Mr. Ellison is always looking for ways to claim that Microsoft is falling on its face and that Oracle will take over the world. While Mr. Ellison has amassed several billion dollars in personal net worth, Microsoft Chairman Bill Gates is worth several times as much. Mr. Ellison can't abide that.

The idea of the $500 network computer appealed to Mr. Ellison because it would be too inexpensive to contain a hard disk and, thus, couldn't hold the normal complement of Microsoft personal computer software. Software for browsing the Internet, sending e-mail, etc. would be downloaded from a central computer over a phone line or office computer network each time someone used the device. Because Oracle's databases are used on so many central computers, it would win big from network computers. Meanwhile, Mr. Ellison argued, network computers would render personal computers--and Microsoft--obsolete.

Mr. Ellison knows better. He is, after all, a smart fellow. But he has pushed the network computer idea so hard that he has generated an amazing amount of debate that has left executives--even technical ones--disoriented.

Because silly debates arise all too frequently in the technical world, here are two thoughts on how to sort through this type of discussion:

*First, realize that debates can be childish even if they involve dozens of acronyms. Technical specifications can be pulled out of thin air. This happened not just with the network computer but also with the 500-channel cable system

that an executive hypothesized was possible some years back and that a whole industry then tried to create. Seemingly technical battles can actually be personal ones, because of the huge egos that have been built along with the fortunes in the computer world. High levels of vitriol are a pretty good sign that that's what's occurring.

*Second, when confusion sets in, history is a good guide despite the occasional claim that the computer world moves too fast for the past to mean anything about the future. When I did a book on IBM's problems in 1993, the company laid out impressive technical arguments to show that its workstation chips would cut deeply into Intel's sales of personal computer processors and that OS/2 would vanquish Windows. IBM browbeat me with similar cases for its joint ventures with Apple and for its on-line service, Prodigy. I didn't have the technical expertise to argue with IBM, but I still made dire predictions because I had seen IBM fail in too many similar situations in the past. I was right. History was a terrific guide.

With network computers, the history is again clear. The idea of dumbed-down devices connected to a smart network of big computers has arisen repeatedly because managing such networks would be easier for I/S departments. But the idea has failed time and again over the past dozen years whether the devices were called 3270 PCs (in the mid-1980s), diskless PCs (late '80s), X terminals (early '90s), or diskless PCs again (mid-'90s).

Mr. Ellison's history should also provide pause. In the late 1980s, he flogged "massively parallel" supercomputers, which were to run big Oracle databases but which never lived up to their promise. In the early 1990s, he pushed "video servers." They were to hold huge databases of movies and television shows, dispatching them to customers on demand as the core of the 500-channel cable systems that never happened.

So now it's on to network computers. If history is any guide, he'll give up on network computers soon and jump on another big idea.