[yariv]

Search is far from archaic. It's here to stay. Even if Xobni or Facebook figure out how to make useful suggestions by predicting my intentions (something I doubt would work very well, no matter how much they know about me), it would have a near insignificant impact on the utility I get from Google. People will always want to search for information and unless someone figures out how to build a better search engine than Google, Google will remain king.

With regards to new interfaces (voice/visual/whatever) that people sometimes say will be the "next big thing" in search -- these are just features that Google can easily implement if they prove to be so effective (which they probably won't -- I'd much rather type in my query than say it). Replicating the sophistication, performance and scale of Google is much harder.

[xirium]

Cuill ( http://www.cuill.com/ ) and Powerset ( http://www.powerset.com/ ) are both contenders to make a better search engine. Cuill assumes that you can make a comparable search system more efficiently. Powerset assumes that more pertinent results have higher value. Scalability isn't an issue for Cuill. Cuill's president is Anna Paterson, who is partly responsible for Google's scalability.

Anyhow, Cuill and Powerset both have scope to make systems which are more efficient and could therefore be supported by less intrusive advertising. The switch from AltaVista to Google was influenced by the intrusiveness of advertising. Perhaps this consideration would inflence a subsequent transistion.