I have to say, Google is fantastic at indexing content, at aggregating content, and rating content, and bringing together many weak relations between data to construct astonishingly good correlations, and at driving new things forward. The one thing they have never understood though and probably never will is how search is meant to natural algorithms in preference to whatever structure the content comes with.
So that is why they are so terrible with search. They do not work towards making links between data accessible to everyone else because they are too busy locking in that indexing power to themselves. Indeed, even as their own websites are not semantic, meanwhile, the rest of the world drowns in an excess of information to the point that webmasters use Google to search their own sites because they never organised it well enough for their use. They have not gone all that far at promoting really rich metadata themselves, because they do not see the need for it in their pragmatism. It works, but is destroying the idea of a searchable internet.
There are some hopes though. Drupal and Wordpress and others are raising the proportion of websites with highly regular and predictable structures and generally improving the information architecture of more and more sites. Google might not use that data much yet, but in ten years the academic expertise now moving into the commercial and public sectors in information organisation might have filtered down enough to the bottom to make a difference (with web standardistas of one camp or another leading the way). It would take a colossal shift in Google though to build any of that into their mindset though and move from Wolfram Alpha style patterns to SPARQL. Note though it would be silly to drop what they have, which arguable will always be useful on the text-level anyway, so search is not going anywhere fast.
Hence, I am not optimistic about the chances of a dramatic move towards clear information in the immediate future because Google is unwilling to jump towards it. Alternatively, like the very trendy Diaspora, the latest challenger to Facebook (another piece of interesting news to follow), they might simply not have the expertise to do it all. It seems strange to say it of Google, but they do not hire everyone.
Ultimately though, they are just too drawn to search to see the potential of queries, or, if they can, too boring to be caught up in the vision.