I have 2 points, Nox.
1) Price. First of all, your comparison is misrepresentative. Not wrong -- just deceptive. Of course Dell offers a very good deal on the P4. As the only pure-Intel vendor left, they get great pricing and first-choice on new chips. Others get P4s also, but at a higher cost.
In the DIY market, which is all that most Anandtechers care about, the price delta is :
Athlon 1200 (263) + Asus MB (131) + 128mb DDR 2100 (144)
P4 1400 (549) + Intel/Asus/MSI MB (219) + 128mb PC800 RDRAM (140)
All prices are the LOWEST given on pricewatch at the time of this post. (yes, the P4 price is an OEM version without the included rimms)
You could buy the entire Athlon 1200 DDR system for less than the price of the P4 1400 alone! Our total DIY price delta is $370. I don't know about how much you like your money, but $370 is very significant! I could buy an Athlon and take a nice little ski vacation to Whistler tacked on.
True, if you you always want to buy Dell, then you get a better price. But you tell me, do you want to be shackled to one vendor for every purchase you make?
2) Performance. The P4 does look good on a few benchmarks. Sisoft Sandra. Optimized version of Flask (that is UNAVAILABLE for love or money). Quake 3. Care to name some more? In every other gaming benchmark that I've seen, an Athlon 1200 DDR over the P4 1400 (& 1500). Sometimes by a slim margin, but other times by up to almost 20%!
But you don't just play games. How about business apps? Nevermind, you definitely don't want to go there. DVD recording/video editing? Maybe the area that the P4 will really shine in when it gets optimized software, but not right now. How about 3d creation & rendering? Buahaha. That one isn't even competitive -- an Athlon 600 SDR could probably equal your P4 1400/1500. (See Tom's 3dsmax benchmarks).
So what's left? Well, that Tom article you mention used only 3 benchmarks -- Bapco, Quake3, and WebMark 2001 to show the performance of the P4 vs Athlon@1400. Just one problem with WebMark -- what exactly is it measuring? Other web sites that have attempted to use Webmark have stopped because of the unreliable and context-free results that it generates (I believe it was Hardocp or thresh's that said this).
Please benchmark using things that people will actually use. Someone else mentioned visiting
Ace's Hardware for their review, and if you haven't, I
strongly recommend you do. They benchmark the P4 with a variety of real-life modern gaming & business benchmarks that expose performance outside of Quake3 and Sandra. They have a very fair review, exposing some good points of the P4 in various areas, mainly video stream related, but overall they paint a very different picture of the P4 than the one you attempt to.
I hope I've made my point. The P4 will not die -- even if the chip sucked, it's promoted by Intel and would be a winner -- but the fact is, it doesn't suck. When it reaches 2ghz clockspeeds and gets some optimizations, it might be a nice purchase.
But right now, why pay more, even a modest amount more, for so much less?
edit : recalculated the benchmark deltas. Max lead is ~25% lead for the Athlon.