Sometime soon, somebody out there is going to try to sell you on the virtues of the Microwave Bank. This startlingly innovative breakthrough endows your microwave oven with online banking and Internet access.

Cool, right?

Well, no. If NCR, which developed the Microwave Bank as a lab prototype, ever tries to sell this bastardized hybrid appliance to actual consumers, it will be met with responses like, "Why in hell would I want that thing?"

I thought of this peculiar technological dead end as I read the news of Microsoft's $5 billion investment in AT&T. As part of the latest baroque corporate merger play, in which AT&T gobbled up the MediaOne cable franchise, Microsoft pumped a significant fraction of its vast cash reserves into AT&T.

With the previously acquired TCI and MediaOne, AT&T will now own the largest single piece of the U.S. cable market -- and it hopes to use that entree into the home to deliver a new array of telephone, TV and Internet services. In exchange for its cash, Microsoft won a guarantee that AT&T would use Microsoft's Windows CE software as the basis for at least 2.5 million of the "set-top boxes" AT&T expects to deploy in consumers' homes to deliver those new services.

So what does the Microsoft bank's latest investment have to do with the Microwave Bank? Simple: Microsoft's investment is only the latest installment in the decade-long quest to marry computers and television in a grand "convergence" scheme. Bill Gates, who touted such a "digital highway" to the home in his "The Road Ahead," is betting his company on such a "converged" future. But the effort to crossbreed TV and the Net is likely to prove as ludicrous and pointless as the notion that you might want to surf the Web from your microwave oven.

Some new technologies take off effortlessly, as if self-propelled. Users grab them, find dozens of new uses for them and weave them into daily life and work. The Web itself is the perfect example: In its earliest days, no one marketed the Web because no one needed to -- people took a look at Mosaic, the Web's ur-browser, and realized they were seeing something wonderful; then they discovered that it took just a little technological know-how to build sites of their own, and there was no stopping the thing.

A lot of technologies, though, are born amid hoopla but fail to catch on with the public. The industry thinks it has found some holy grail, but people take one look and think, "Why bother with that?"

So far, despite all the hype and the billions being spent, TV-computer convergence -- with its set-top boxes and its visions of interactive nirvana -- remains a "why bother?" technology.

If $5 billion in Microsoft dollars is buying 2.5 million Windows CE set-top boxes, that means that Gates and company are paying about $2,000 for each proud home that ends up using CE, the stripped-down consumer-appliance version of Windows. That kind of spending is in the same league as the amount that Time Warner found it had to spend to wire homes to its prototype "full service" interactive network in Orlando five years ago -- and Microsoft's money is only buying the opportunity to serve a customer base. This isn't real market share; it's an expensive way to put dibs on a share of a market that doesn't yet exist.

It's also a colossal illustration of Microsoft's hypocrisy. For years now -- most vocally in its antitrust defense -- Microsoft has touted the technology marketplace's Darwinian efficiency, in which only the best innovations survive and prosper. Windows won its colossal share of the personal computing market not because of monopolistic practices or ruthless business tactics but because it was the best choice, the company argued. Now, though, it seems that Microsoft must offer what might be called, charitably, a subsidy (or uncharitably, a bribe) to get other companies to adopt Windows CE. That doesn't bode well for CE's future, to put it mildly.

But the larger issue here isn't which technology the "convergence platform" of the future will be built upon (CE or Java or anything else). The real question is, what do these companies expect us to do with it?

The logic of "convergence" assumes that the households of the United States want to do something more with the lines into their home than what we do already: watch TV via cable, make telephone calls and use the Internet. No doubt in the longest run the prophets of convergence are right: Once all these media are digital, there's little reason for us to use different wires (or "pipes," as telecommunications types are fond of calling them) to receive the data.

In the long run, though, as John Maynard Keynes said, we're all dead. In the meantime, no one has figured out, in the short run of the next five to 10 years, why one would want to merge TV and the Net -- or to what uses one might put the combined media. "Convergence" remains a technological innovation desperately seeking a practical (or diverting) use -- and so far, the best ideas the industry has concocted are pretty lame.

Do you really want to pay good money each month to put a Web window around a TV show, so that you can look up sports statistics or order the scarf that a character in "Felicity" is wearing? "Movies on demand," long considered the "killer app" of interactive television dreams, is a scheme for which there is no hard evidence of actual consumer demand. Even if there were, it's hardly "interactive" -- it doesn't go much beyond what's available in hotel rooms today.

The fact is that most "convergence" applications chiefly serve the interests of marketers and advertisers and content owners, who dream of "metering" their audience's access to music and movies and TV shows. So corporations spend fortunes developing them in labs -- and are then shocked and horrified to discover that people aren't that excited by the result.

The real problem here is that industry executives imagine that, by sending a "fat pipe" into the home, they are going to tap into a pent-up demand for digital broadcast-style entertainment -- basically, a gussied-up version of today's TV that layers on a few interactive bells and whistles to facilitate e-commerce. Meanwhile, the people who actually want "fat pipes" and are signing up for fast cable-modem service and DSL lines are classic early adopters -- heavy Internet users who want an "always on" Net connection so they can surf the Web faster, download MP3 files and streaming video, send and receive e-mail without dialing up a connection, and even run a Web server from the home.

These users couldn't care less about plugging their TV into the Net -- they're the demographic group whose members have already begun turning their backs on TV because they spend so much time online.

This is the colossal mismatch of industry offering and customer desire into which Microsoft is sinking $5 billion. How much would you like to bet that, if those 2.5 million Windows CE set-top boxes ever get manufactured, the vast majority of them will end up as landfill?