Software inefficiency can always outpace Moore’s Law. — Jaron Lanier, quoted in Scott Rosenberg’s column today. This is one reason why I don’t tend to buy software, and,when I do buy it, it never runs faster than the thing it replaces. I really, seriously, have never had a program that looks up names and addresses as fast and fusslessly as a DOS TSR I used back in 1990, when collating the Best of the Independent book on a Compaq laptop with a gigantic 20 MB hard disk.
But the OCR program I use occasionally, TextBridge Pro 98 (the fruit of an ancient blag when writing for the Mail) has been getting slower and crankier with every improvement on the computer that runs it. So yesterday I tried a Russian program, Abbyy Finereader. Given the choice I would always buy a Russian program over an American one, because it’s more likely to have been written to run well than to sell well. I didn’t really want it to do anything more than output its results to the clipboard, something Textbridge refuses to do. Finereader will do that; but it also recognises at about three times the speed and twice the accuracy. So I’ll buy it.
I wonder if more software could be sold if it did the old job quicker, better, and with less fuss, rather than thinking up new jobs. I don’t suppose, though, that there’ll ever be a large enough sample of such programs to test the theory.