The thing that made Google unique is that they ranked pages according to the links between them (i.e. PageRank), as opposed to more traditional document ranking methods that are keyword based.
As a result, Google found showed you content that people liked (and linked to). From the end-user's perspective, Google search was noticeably better and more relevant.
As far as I know, it was the first massive application of data mining graphs, which now drives social networking, content recommendations and more.
The algorithm is quite fascinating (from what we can infer). Using things like a markov chain to model results in Google shows a big jump from human-compiled results in early search engines - it's a new application of theoretical knowledge that has greatly enhanced society, and frankly shows a degree of automation that could be a basis machine intelligence.
This is a very good point. Google is more accurate, and faster, but in the end it does the same thing. It helps the user find information on the web. It's not a great new idea, it's an improvement on an existing idea.
Sometimes a quantitative difference becomes a qualitative difference. For example, motorcycles are just faster horses (in some ways worse, like for off-roading), but you could hardly argue they are the same thing. I would argue that google is substantially more useful than altavista, or dogpile, ever were.
And ultimately Altavista and Dogpile are irrelevant, because neither existed just a few years before Google. Web search is revolutionary; search engines before Google were just halting steps on the way. That is, the contention isn't that Google revolutionized web search, but that web search revolutionized the world.
More stuff indexed, for one thing, paper books and videos included. And hopefully, more intelligent interpretation of queries, though that's more debatable.
As a result, Google found showed you content that people liked (and linked to). From the end-user's perspective, Google search was noticeably better and more relevant.
As far as I know, it was the first massive application of data mining graphs, which now drives social networking, content recommendations and more.