The Problem with Prediction

Every now and then I may try to alert readers about some technology that I think will be disruptive (e.g. RFID to Complicate Database Administration), but I don’t usually try to predict what software or companies will succeed or perish.

Now as my regular readers know, I undertook a cleaning project earlier this year – organizing my office closet. While cleaning out my closet I took some time to blog about some of the things I found – well, mostly about research and predictions for the computer industry. For the curious, here are the links to those entries: Predicting the Future?, But When They're Right..., and Cleaning Out My Closet.

Well, that closet is as clean as it is going to get this year, but I periodically attempt to reorganize my many bookshelves, too. While doing so, I happened across an old book written by John C. Dvorak called Dvorak Predicts. Dvorak is an industry pundit and popular columnist for PC Magazine . I readily admit to flipping right to his column first when I receive each issue of PC Magazine. He is entertaining. But his powers of prediction are poor. And why wouldn’t they be? No one knows what tomorrow will bring...

Now I don’t do what I am about to do to discredit Dvorak specifically, but to alert everyone that even the most talented writers and pundits have a hard time predicting the future. Here, direct from his book, Dvorak Predicts (published in 1994 by McGraw-Hill, ISBN 0-07-881981-4) are some real stinker predictions:

“We can expect IBM to someday quit the mainframe business just as it quit the scientific computer business.” (page 20)

Although this prediction may come true some day, 12 years later the mainframe is still viable and still a core offering from IBM. Why didn’t he predict that IBM would quit the PC business (which it did in 2004)? – Craig

“Voice recognition will be the killer application of the 1990s.” (page 21)

Didn’t happen. Oh, IBM and Dragon had some voice recognition applications that sold in so-so amounts, but killer app? Nope. – Craig

“Microsoft will open stores.” (page 64)

Never happened; good thing for Microsoft, too since other companies that opened stores didn’t fare all that well (e.g. Gateway). – Craig

Furthermore, Dvorak predicted that Unicode would lead to the death of ASCII by 1995. Well, in the long term I’m sure that prediction will come true, but the “by 1995” part didn't! Did he ever really believe that? I mean, even though Unicode is important today, ASCII is still very much alive and kicking.

And a few things that Dvorak does not even mention include the World Wide Web, Java, and spam. One would think that a prescient prognosticator would foresee these three facts of everyday life in the world of IT. But no...

Of course, Dvorak did get some things absolutely correct. For example: “Piracy will increase despite efforts to stop it.” He wrote that prediction about software piracy but it is absolutely applicable today with regard to media, especially music and movies. And he said that “Gerstner will be good for IBM,” which he undoubtedly was. And he predicted a rosy future for recordable CD and optical media (then again, back then, who wasn’t predicting that?)

A lot of the predictions (and there are many more than are mentioned here) strike me, here in the future, as of the “who cares” variety. By that I mean, I would expect a useful book of predictions to predict about things that matter in the future, and not about things that are just dust bunnies from the past (e.g. Apple Newton, OS/2). And he never foresees the Palm or Windows PocketPCs – just calls the basic idea behind the Newton “stupid” (page 83)

So, what good is a book of predictions if the majority of them don’t come true. Exactly! No dang good at all – other than (maybe) as an amusing read. The book is out of print, but if you find a copy you might shelling out the 50 or so cents it takes to buy it these days. And here in the future, 12 years later, you might get a chuckle out of some of it... or at least enjoy reading it as an historical piece on an interesting time in the history of computers – that being, the timeframe right before the Web exploded.

What does any of this have to do with database management? Well, not much, at least directly. But the next time someone tells you that some new database technology is going to render everything you know to be obsolete, slow down and think about the problem with prediction.