They post an article up which is a review by some other site. The other site has over 20 pages of performance conjecture, which is good for the reviewer site’s ad revenue (Vonage and newegg.com ads).
Now for the interesting part. The article is a review of RAIDs and performance. Now, in all of the 20+ pages of graphs, there is not one single RAID failure test to see if the RAID even does a good job with a bad block or failure. To me, that is the reason I use RAID.