From the Washington Post article [washingtonpost.com]
The testing is done better than for the earlier BBC article, but it's far from exhaustive and completely US-centric. However, it attempts to give a fair impression of an end-user search pattern.
The article does mention at the end that MS are aware that their results aren't up to scratch yet, but it makes you wonder why Microsoft are making so much noise about it when first impressions count for so much.
There is also a small irony in the fact that the article page has Google search boxes top and bottom!