Dude, do a search, there a countless threads on this. There's also a tweakers.net article that actually answers the question fairly instead of ananadtech's bumbling. Even anandtech's "pure" hard drive performance was set to spread the IOs around, they didn't even mention the stripe size afaict going back and looking at the test page, and they used a crappy controller. If they were trying to replicate the conditions for the type of barely-know-what-they-are-doing users we all laugh at, then they succeeded admirably.
It basically boils down to if you use large files regularly you will be helped a lot. Often when you are stuck waiting on your computer it is precisely that. Additionally if you get a real hardware raid controller and set your stripe size large then smaller requests get sent to only a single drive, splitting activity among them and again helping performance significantly when you are actually stressing the drives - which is again precisely when you are waiting on them. People complain about failure rate, but you should be backing up import files anyway even without raid 0, and drives of this quality don't exactly die often - twice hardly ever is still hardly ever. The dark days of babying every drive and carefully plotting where you put all your data and programs to split activity across them are gone.
Anandtech's coverage of sata's new queuing options was a similar huge disappointment. I would have wanted to see performance when running two p2p apps, winamp, and office or a game all at the same time. Instead they pretended the feature had nothing to do with ameliorating performance drop under heavy load. Same with raid, you would only use it if you were stressing the storage subsystem and they completely ignored the power users. So basically if you barely use your computer and are terrible at setting up hardware, then yeah what they said is true.