About a decade ago, I was on a panel about the future of print media in the Internet age. One of my fellow panelists argued that digital technology would kill off dead-tree publications. In a world of hyperlinked information, he maintained, ink on paper simply couldn't compete.

I replied that media compete but never kill each other off. Movies were more efficient to distribute than plays, but theater didn't die. Television offers a richer experience than radio, but the older technology still commands a large audience. The only communications medium I could think of that disappeared was the telegraph.

My fellow panelist, an executive in charge of online operations at a major media company, frowned. Several years later, after he lost that job, he went on to write an excellent book that has no hyperlinks, video, or other multimedia features. Many people bought and enjoyed it.

I thought of this as I read a recent New York Times story about old technology. It starts with an anecdote about how mainframe computers, those crate-size dinosaurs, still have their uses. So, the article points out, do trains, which benefit from the congestion and fuel costs faced by cars and trucks.

None of these legacy technologies is exactly thriving. Theater doesn't qualify as mass media, and radio no longer fills the center of most living rooms but part of some commutes. The domestic theatrical audience for movies has been shrinking, at least relative to the American population, since the introduction of television. Judging from the reading habits of the young, newspapers are heading for the same fate. (Indeed, based on what people are reading online, I'd suggest that actually following the news is a dying habit, but that's the subject of another column.) All of those businesses are shrinking, at least in relative terms, and have been for some time. But there's plenty of money to be made in a shrinking business, especially as the cost of creating product declines.

If it sounds like I'm suggesting that everything you know about technology and progress is wrong, it's because much of it comes from pundits, who tend to focus on business (as the Times points out) and don't deal in nuance. It's easy to get attention by predicting, as the Times reports the computer industry observer Stewart Alsop did in 1991, that mainframes would disappear. It's hardly revolutionary to proclaim that mainframes will be displaced as the dominant business computers by PCs, but they'll still serve some important functions. It would have been poor investment advice, since nearly all the companies that made mainframe computers during their heyday have left the industry or gone out of business altogether. More importantly, it would have made a lousy headline. One day, perhaps someone will write a book about it.