Geek News: Flawed computer technology frustrates

I’m in a cynical mood, so I’ve decided to write a little about computer technology that doesn’t work — and there’s a lot of it. I remember while working in the oil business long ago we had a special sensor that, in theory, could measure the density of the mud that was coming out of the well. (You pump mud down the pipe and it comes up around the pipe to bring up the rock cuttings.) This was a very expensive bit of equipment that was complicated to install and calibrate, but that would have been OK — if it had worked. In fact, it didn’t work — never did I ever see it working — and I don’t think I ever met a colleague who would admit to seeing it work, either.

I’d be interested in knowing what proportion of computer equipment doesn’t ever work. There are several categories of failures in the software business, and I realize that most software development falls into one of these categories. One category would be “outright failure,” development projects that are never completed — they run out of money long before anything useful comes out of the work. This amounts to literally billions of dollars a year. Another category would be consumer software that simply isn’t much good and doesn’t last long on the market. But I’m talking here about entire categories of software that sell for years but that never actually work.

This evening, my youngest son complained that his teacher had told him not to use Microsoft Word’s grammar checker anymore because she disagreed with the results. She’s quite right. In fact, grammar checkers simply don’t work — they’re a complete joke. I know enough about grammar to know that most of what Word tells me to do is either unnecessary or just plain wrong. A while ago I heard about a little experiment carried out with students. The researcher found that a grammar checker actually hurt the students’ writing — and the worse writers were hurt the most. In other words, if you understand grammar, you can use a checker to find a few things but you have to know enough to know when the grammar checker is wrong.

SPONSORED CONTENT

Business Cares: April 2024

In Colorado, 1 in 3 women, 1 in 3 men and 1 in 2 transgender individuals will experience an attempted or completed sexual assault in their lifetime. During April, we recognize Sexual Assault Awareness Month with the hopes of increasing conversations about this very important issue.

I first used a grammar checker (I wish I could remember its name) in 1988, I think it was, maybe 1989. In those days, grammar checkers were add-on products, not built into Word processors. They didn’t work then, of course, and they still don’t work, but what amuses me about these products is that people have been building them, using them, and now and even buying them, for 15 years, and they’ve never worked.

How many other categories of products, outside the computer field, could be sold for 15 years despite the fact that they simply don’t work? Sure, there are varying levels of quality. You can buy a car that’s unreliable, but most of the time it will get you to work. It’s only in the computer field that you can build and sell a product for years, despite the fact that it doesn’t work.

Another example is voice recognition. Again, a product that’s been around in one form or another for a long time. In some applications it does (almost) work. A lot of telephone-based, voice-recognition systems are in use now, and I’ll give them the benefit of the doubt. I have been incredibly frustrated by these stupid things on more than one occasion, but still, they work well enough to be put into use by thousands of companies and to reduce customer-service costs. But how about PC-based voice-recognition systems?

Of the tens or hundreds of thousands of copies of PC voice-recognition software sold over the past decade, how many are still in use? Very few, I’ll bet. Scientific American did an unusual survey of these products. They searched magazines for reviews of voice-recognition software, looking for writers who had praised one or more of these products. Then they tracked down these writers with a simple question: Are you using voice-recognition software?

Not surprisingly, none were using such a product. My experience with voice-recognition software is not huge, but I did play with the new voice-recognition system built into Office 2003, and will say that it’s interesting, sort of fun ? but not particularly usable. I spent most of my time fighting with it, trying to pronounce carefully and clearly, and fixing text that it messed up. Now, if I lost the use of my hands I might move to a voice-recognition system, but anyone with one or more functioning hands — or feet, perhaps — probably won’t find this system particularly useful.

It still amazes me that people buy these useless products, use them for a little while, then forget about them. But that’s good. Because one day, grammar checkers will work well, as will voice-recognition software. Someone’s got to pay for the decades of development to get these products to a workable stage. I’m just glad it’s not me.

Peter Kent is an author and consultant on technology issues. He can be reached at PK@TopFloor.com.