When I say Tom doesn't know hardware, I'm not saying he can't set it up. I'm sure the guy is A+ certified and can fix windoze problems, overclock his geforce, flash his bios or whatever the hell it is he does.
However, I don't think he knows hardware well enough to point out what makes a better product and why. Instead he relies on benchmarking tests, which are often misleading. Just about all the technobabble on his site is almost always taken from another source. Just because he can click some buttons and graph the statistics doesn't really mean he knows what he's talking about. Never in his benchmarks does he try to stress the computer, which, if you want to know which product performs better, you pretty much have to do that. For example, if you are comparing two different brands of saws, you can say brand A is better if you notice it cut faster right out of the box. However, you give that brand A to a contruction worker, and when the blade breaks after a day or two, he'll be wondering why the hell you didn't get him brand B like he asked for. It's the same with Tom's Hardware. If you take a look at his statistics, it's always some ungoldy high framerate in some popular 3d program. But what good is that for the consumer? Does it really matter whether or not you get 150fps as oppsed to 130fps? I hope not. It usually only matters when the framerate drops down below 25fps, and suddenly you find yourself lowering the detail levels, or combing your sofa for extra cash for better hardware. Also, Tom never mentions the detail and rendering options set for benchmarking, which tend to make a huge difference in not only determining performance, but also videol quality. Usually when comparing different chipsets of 3d cards, this is a big factor, since not only do competive cards have different features, but some may perform a lot better than other using these features.