In all seriousness, in some ways google was a significant step back in search technology. Many cool but dubious features were stripped out so that they could achieve the vaunted 0.01 second search over 4 billion web-pages for 1000 different people at once. One major thing that has been brewing in search research for over 10 years is being able to refine the semantic meaning of your search terms. For example, if you search for “mining”, it might popup saying “did you mean data mining or strip mining?”.
Also, search personalisation is a big thing. Google currently keeps cookies on your computer but, AFAIK, doesn’t do much with them. A personalised search would track what you clicked on and how long you stayed on each site and would push the webpages it thought you would like up to the top.
Another thing is that many people don’t realise that it takes quite a bit of training to search effectively and many of us search sub-optimally. Having a search engine that could take in natural language queries and return relevant results would be a major improvement. The askjeeves search engine attempted this with rather dismal results but its another area being heavily researched. eg. Searching for “what is the most expensive element”, “how much is the most expensive element” and “Who makes the most expensive heating elements” should produce drastically different results but current search technology doesn’t have a good method of doing this.
Another relatively simple thing is to just make search engines that trawl deeper into the web. There are still many valuable mines of data that are not exposed to search engines because the results are hidden behind interfaces that must be manipulated.
Image and other media search is a huge and very, very tough field. AFAIK, nobody has done any successful work on classifying images based on the actual pixels inside the image. If I wanted say, “picture of a bear riding a bicycle”, current search technology is not going to be adequate. Currently, image search depends to a very large extent on extracting meta-information from humans that have pre-classified the data in some way. However, there are many HUGE image archives in which many of the images have never seen human eyes. these are, in effect, completely invisible to current search.
In terms of more pragmatic issues, google has in part become a victim of it’s own success. People optimise their pages for google which means that any search about a product inevitably brings up a page offering to sell it to you rather than the company who made it. Blogs are overwhelmingly represented in the first page due to the peculiar link incest that goes on between them. spyware and adware specifically target google which cause people to blame google for their poor search performance.
All in all, search engines really are still truely in infancy, the basics won’t change between now and eternity but different bells and whistles will make future searching a lot more pleasant.