NEW YORK (CNN/Money) - With the Nasdaq within spitting distance of its September lows, some investors are wondering if it's the time to sell everything and head for the hills.

Others, of course, ponder whether they should plunge in and start buying. Are there actual tech bargains to be found in the current dreary landscape?

The answer to that depends on the answer to a couple other questions: Are we merely in a cyclical trough, poised for another grand tech adventure? Or was the 1990s boom a one-time phenomenon not likely to be repeated, driven by the confluence of the Internet craze and last-minute Y2K preparations? Will the next tech boom look anything like the last one?

Signs point to no. Technology spending in the late 1990s was driven by what analyst Hugh Bishop of the Aberdeen Group calls "almost indiscriminate purchasing" that pushed up sales and margins.

How indiscriminate? Take a look at business software. Morgan Stanley's Chuck Phillips notes that even today the software industry is plagued with an excess of "shelfware" -- software purchased but not yet installed.

The corporate technology executives in Phillips' latest survey confessed that they'd bought 30 percent more database-software licenses than they needed. Chastened tech companies may keep a tight lid on software spending for a long time to come no matter how robust the recovery ultimately proves to be.

So what's a reasonable growth rate to expect over the next three to five years? The Aberdeen Group forecasts that U.S. IT spending will recover some 4 percent this year from its 2001 decline. Over the next three years, Aberdeen expects spending to increase perhaps 5 to 6 percent annually. Still, all such estimates are merely rough guesses.

The lack of certainty about tech's future makes it nearly impossible to figure out how to value tech stocks accurately.

Still, one intriguing attempt to do just that comes in a recent report from Soundview analyst Arnie Berman, who argues that tech stocks are cheaper than most analysts and investors seem to think.

"Technology stocks that were once valued on a price-to-possibilities basis are now valued as if history no longer matters," Berman writes. What he's getting at is this: Technology is a cyclical industry, with many peaks and valleys.

Looking at valuations now -- in the middle of a deep valley -- can steer you wrong, since the E portion of the P/E ratio is depressed, making stocks look very expensive. But Berman thinks the industry will be back to peak earnings by 2005, and on that basis tech look cheap, trading at about 18 times prior peak earnings.

Within tech, he sees semiconductors as relatively overvalued and hardware stocks as relatively cheap.

MORE TECH INVESTOR       AOL: You've got ail As the markets go, so goes VC Blockbuster takes on Netflix

While I'd hesitate to judge any stock particularly "cheap" based on what seems to me pretty rosy assumptions about earnings three years down the road, what he's doing is a useful exercise for tech investors -- in part because it forces you to make your assumptions explicit.

Investors need to get past their obsession with quarterly results and to try to figure out what longer term tech growth rates will look like in this new era. That's tough to do, but it's a heck of a lot better than buying blind and hoping for the best.

Sign up to receive the Tech Investor column by e-mail.

Plus, see more tech commentary and get the latest tech news.