I think a key distinction is whether we are talking about the science or business of search. The science of search is AI-Hard -- good luck with that.
I think it is reasonable to claim that the business of search is 90% solved. To me this means that Google has reasonable solutions in place for what they see as 90% of the search market (by revenue/profitability/etc) in the medium term. They still have plenty of things to improve upon, but what is the likelihood of another Google (in terms of growth, dominance) coming out of search?
Google has it locked up as far as profitability goes but the user experience has a long ways to go. Until I can spawn off a search to find the cheapest VPS host, returned with a spec sheet of every offering alongside speed tests of five sites hosted on each, we're not even in the ballpark.
Search should be able to solve problems all the way through, not just offer possible resources that may or may not help in your hunt for a solution. Which is, as you say, AI-hard/a mindbogglingly batshit crazy undertaking.
Which, in turn, means that someone will eventually succeed and make a lot of money doing so.
I agree.
An analogy could be the motor vehicle business. 90% solved by the 50s or 60 or 80s or whatever you place it at. But obviously cars still get better with time.
The short answer is that search isn't even 10% solved.
It's mind-boggling to me that people are so content to accept search the way it is right now. All anyone has right now is the first step.
And I'm not just talking algorithmically, I'm talking about user experience and interaction as well! Right now google's linklist format basically does nothing to help you refine and dissect information from your initial search. It doesn't extract meaningful data and bios out from links. It doesn't help you do common word sense disambiguation. It doesn't help you when your search terms can be taken in multiple ways.
And there is more yet to be done even in the keyword space search. For example, if I type "Who is running for president?" (or the keywordese: president candidates 2008) this may mean something different for a frenchman than an englishman than an american. Google doesn't really do a lot to help you here
I Mr. Arrington here confuses the process or algorithm of search with the data on which search operates. Apart from talking about natural language, everything else speaks to the data, which is of course an interesting problem (decoding data in streams computers can't currently interpret), but it's not a search problem.
I guess it would depend on how you define solved. I'd say it's close enough that I don't find myself trying search engines other than Google these days.
I think it is reasonable to claim that the business of search is 90% solved. To me this means that Google has reasonable solutions in place for what they see as 90% of the search market (by revenue/profitability/etc) in the medium term. They still have plenty of things to improve upon, but what is the likelihood of another Google (in terms of growth, dominance) coming out of search?