It's Not Just Carmakers Who Cheat

The news that Volkswagen "cheated" on emissions tests by using software was not a surprise to me. When I was the Editor of PC Magazine, we engaged in a constant battle of wits and ethics with the technology companies that sent us products for testing.

In the 1990s, PC Magazine was a dominant editorial force in the computer industry, with a circulation well above one million and outsized influence on the buying decisions of its readers. The magazine had built its reputation with transparent, scientific testing methods; our 400-to-500-page biweekly issues were chock full of tables and charts detailing the results of every available product tested and explaining our decisions about the best. You could even download many of the tests we used and run them on your own PC to see how it compared. The winners earned a coveted label: "PC Magazine Editors' Choice" that could have a huge impact on sales.

In those days, there were many more vendors in every category of product; it was not unusual that our call for PCs with a particular set of specs would result in submissions from 50 different companies - many like Northgate and Swan - long forgotten in the massive consolidation that would follow. Once, when we tested modems (remember those?) we had to review 100 different models. It took a full-time lab staff of 30 (including three people just managing the logistics of shipping and receiving) and an army of freelancers to manage the flood.
Facing such intense competition, some companies were more desperate than others to rise to the top of the charts. In one large review, our techs noticed that some PCs performed remarkably better than others on our tests. A little investigation showed that the central processors in these units had been speeded up above the rate set by the manufacturer. Like a car souped-up beyond its design limits, these chips would overheat and fail much more quickly - but last long enough, the manufacturers hoped - to get through the testing phase.

Another big differentiator in those days was the graphics card, the device that converts the bits and bytes to text and images on your computer screen. We had a rigorous test for these cards that put them through their capabilities and timed them. Again, our testers noticed that some graphics cards were scoring much higher than could be explained by superior engineering. We discovered that some vendors had embedded our entire test aboard their devices. Instead of taking the test off the hard drive where we installed it, the card took it off its own chip, which was much faster.

The problem was that these stellar results only showed up on the tests. They were useless to the everyday user of these products. In fact, once we removed or neutralized the cheats, these products often fared more poorly than their competitors.

The cheaters were in the minority, but they were a reminders that technologically-driven companies are not necessarily superior in morality, much as they like to proclaim. For Volkswagen, it was apparently a lot easier to deceive than to solve a complex engineering problem.