Should Wikipedia's Search be improved?

Greplin, a company making waves for “tackling the other half of search”, launched a competition to redesign wikipedia’s search experience: “everything from the behaviors on the search page, to the layout of content, to the details of typography, color, animations. Today, many people use Google to search Wikipedia, but imagine a whole new experience that makes using Google to search Wikipedia feel outdated.”
It is a novel and positive initiative that might yield some interesting results, but I am wondering if it is possible or needed. Let me explain:
How do you search wikipedia? If you are like the rest of us, you mainly go through Google, then click on the wikipedia link if it exists. When do I only use wikipedia’s search? It is when I am sure there will be content available (something broad like France or Gears) and even then I usually use the Google Chrome’s autocomplete search function.

I feel this is an important point, because it shows the difference between Google and Wikipedia. When searching for information related to the keyword that I am not knowledgeable about, I use wikipedia. When travelling to another country (say Singapore), you want information on that country, you don’t want something specific (not until you are more knowledgeable about the subject). Here Wikipedia crowns king. However this beats the purpose of searching (or trying to improve a search experience): when I know what to find and how to find it. (I don’t know, but I know where to find information so that I can know more).
More often that not, I use Google to search for answers (or when I don’t know where to go look for information). (I don’t know, and I don’t know where and how to find information that I want). Of course you can’t extrapolate one user’s habits to the whole population. Let’s assume I am searching for information (not answers) on something that I am unsure of whether it will be in Wikipedia (random small suburb in a backwater town). Should I use Wikipedia or Google? Google of course. If it is on Wikipedia, it should be in the top few links. If not, then I can find relevant information on other websites. If I went the Wikipedia avenue, I would’ve been presented with nothing and then had to go Google anyway.
In other words, Google will always be more useful when searching answers and information. It boils down to the fact that not all information is for the purpose of being encyclopaedic.
I do think there are perhaps areas where Wikipedia’s search can be made more useful:
> This is a big suggestion, but maybe Wikipedia shouldn’t have 2 frontpages. I like the globe, but if it wants to improve search, the bar has to be made more prominent.

> The second ‘frontpage’ is language specific ( Once you are on there, there is no (obvious) link back to the ‘globe’ page and you are stuck with the search in the right top corner.

> Increase the speed. Wikipedia’s search is slower than Google. If I do choose to use Wikipedia and I mistook my perception that there will be information on it, I don’t want to feel as if I wasted time that I could’ve spent by just using Google instead.
What if Wikipedia’s search wasn’t primarily about finding information, but an experience in itself?
A fancy, shiny minorty report whizzing and whistles experience (for example), could make Wikipedia’s search fun to use, but maybe that would just turn into qwiki?
In the end to me, the nature of the information on Wikipedia doesn’t lend itself to being about ‘search’. There isn’t much to find: either there is an article about, or there is not. The middle way is vague, and then I have to resort to Google to find what I am looking for.
What do you think? Can Wikipedia’s search experience be improved? Can it be improved enough so that when searching for information, users would go to Wikipedia first, instead of Google? If I am wrong, I am looking forward to the results of the competition. I am open to being pleasantly surprised!