Very interesting interview, especially since Nielsen is not expressing much optimism about the improvements to the search in the near future. One thing that I wish was part of the discussion, though, is a potential shift of the search results paradigm. I don't mean the layout of results, nor better relevancy, but rather a different way of answering the user's query. At SenseBot, we provide a summary of the top relevant search results - kind of a digest on the query topic. This is in contrast to the current model where you still have to go to each individual source for the content. You can try it at www.sensebot.net.

"Wikipedia links are often relevant to the typical user."

So are made-for-advertising page links. Should we want to see more MFAs topping out the search results?

Wikipedia is still a very inaccurate resource that is often used by unscrupulous people to press their own agendas, agendas which in many cases pass "editorial review" because the promulgators "participate in the community".

When people search for information, accuracy and authority should be given more weight than popularity. If the algorithm cannot compute accuracy and authority then the worst possible thing it can do is compensate by looking at popularity.

I agree with the statement that personalizing search does not work due to the changing needs of people thru out the day.
On my site http://omadsense.com my reports show that on average people visit certian section more around the 10 am to 1pm afternoon time [ so the bulk of traffic shows me that these people have others places to visit during other times.