Professionally: I'm looking for opportunities to reduce "the load" on consumers. It's great that the "scratch your own itch" has significantly reduced production costs, but consumption is important too.
Consider web search (which I'm not working on). I think that search engines are technological marvels, but is the search box really the best possible interface? I'm pretty sure that it isn't.
Consider Word/Outlook, two heavily used applications that "provoke" searches. (That is, while folks do things in word/outlook, they often find themselves needing to do searches.) The steps "select, copy, paste into a search box" are at least two steps too many, even if the search box is in the application (as it is in Excel). (Yes, I'm aware of "lazy search box" - it's two steps.)
Moreover, when those steps are completed, they only get the searcher to a results page - said person still has to make use of the search result, which typically involves select/paste and the like.
At a bare minimum, applications should find reasonable search terms and provide shortcuts for using them. Note that this alone can significantly improve search results because the document has information that can be used to improve said results. (For example, the appropriate meaning of "dolphin" is usually unambiguous in the context of a given document.)
The next step is to make it easier for folks to use search results. Google Spreadsheets allows one to "link to" data extracted from a web page, but is that sort of "do it each time" programmatic interface the best that we can do?
Instead, we're getting things like "Enterprise search" that give us the same clunky interface to internal information sources.
Yes, I know that the "Semantic Web" (and Xanadu before that) will to address these issues "any day now", so folks could more easily incorporate outside information, but search engines are way ahead in terms of finding and extracting meaning.