Forecasting the high-tech future is a lot like predicting the local weather two weeks from Tuesday: It's easy to cobble up prognostications that sound reasonable, but real-world specifics inexorably conspire against accuracy.

Take the long-standing predictions of a high-bandwidth future. It doesn't take a degree in quantum physics to anticipate that many of us will be getting bigger data pipes one day, but it's hard to guess how soon that will happen. Unexpected factors keep getting in the way.

A friend of mine recently tried to get cable Internet access. He had the lowest-tech problem imaginable: A wire had to snake invisibly from the living room to his home office. The first human installer was too obese to fit in the crawl space; a skinnier replacement was too dim to figure out how to hook things up. As long as people have to install it, you can bet broadband connectivity won't happen overnight.

Likewise, wireless data schemes will become an increasingly big deal as time goes on, but here too we can expect bottlenecks. I've tried several of the latest cell phones with data capabilities, and surfing the Web on a tiny screen at 14.4kbps isn't exactly a karmic high. Besides, if cell phones are so great, how come people always ask, "Are you calling on a cell phone?"

Net appliances may not apply So here's one of the few predictions I feel confident about making: Thanks to well-oiled hype machines, products still in the pipeline will continue to look far better than those that already exist. What's sexier- a future device that promises to let you surf the Web with ease, or a bloated computer with all its known problems? The forthcoming machine wins - until it gets here and you discover its own quirks and limitations.

The classic example is the infamous Internet appliance. For many moons, pundits have been pushing Net appliances that are supposed to be simpler, more reliable, and cheaper than today's PCs. Sounds great, but nobody has flocked to buy the few Web or email devices on the market. WebTV isn't bad, but though it's been around for three years, fewer than a million people actually use it.

How come? Consider this scenario: My wife wants to give a WebTV unit to one of her aunts. First we have to find out if her aunt's phone jack is anywhere near her TV; we're worried about Auntie tripping over the phone wire. Assuming we find the jack and trust Auntie to install the thing, we run into another issue: Who pays the monthly fees for this gift that keeps on giving?

Engineers tend to save problems involving usability and other human factors for last. I recently suggested to the publicity director of a major consumer electronics firm that she tape an episode of my TV show while she was out of town. Her response: "Don't go there."

If we can't figure out how to make a VCR usable for intelligent human beings, what chance do Internet appliances have? My two cents: Don't sell PCs short. They may be unreliable and hard to master, but nothing beats them for versatility or price/perfomance ratio.

The next shall be last The "everything goes digital" trend will only accelerate. But the real question is what it means for our lives. If its net effect is that we get to buy our cars and T-shirts online instead of from showrooms or catalogues, the growth of digital technology won't matter much.

And if digital technology does turn out to matter - as I believe it will - there will inevitably be trade-offs. Are we willing to trade cheaper phone bills for poorer-sounding calls? Will automation decrease the ranks of the employed and the employable? Will teleconferencing save our highways and flyways from gridlock, or make us crave more time in the presence of friends and colleagues? The answers are far from clear.

So are many of the questions. After all, the gift of prophecy can fail even a media-anointed wonder man like Steve Jobs. Remember, he christened his second computer company NeXT.

It wasn't.

Stephen Manes is the co-host of Digital Duo, a series appearing on US television stations nationwide. For program information, see www.digitalduo.com.)