I've noticed that these days, Google's answers to search questions are generally "why don't you try Wikipedia?"
These days, the Web is somewhere between three and five titans plus a complex of secondary sites (specialist wikis, newspaper sites, news aggregators, major blogs), with a background noise that chiefly consists of spammers mirroring 90% of Wikipedia to improve their search-engine rankings.
So, in a way, this search engine is inevitable -- but I wouldn't have imagined it as early as thirty minutes ago, so it's probably not inevitable, just an idea so good that you forget that it ever didn't exist.
"Duck It!" seems to work _really_ well, too; I'm switching. My only concern is that "Duck It!" doesn't include a way of telling you whether a given paragraph is true or false...
Good point on the try or false, which is why we don't promote that stuff into the Zero-click Info boxes. I have some ideas on how to address that problem, however. It should be one of the next major feature improvements.
You're absolutely right about major sites ruling content. However, I'd add to that there are many vertical search engines that do a great job searching more deep content in their verticals.
Originally one of the things I wanted to do was be a conduit to those engines, but throwing the user there proved too jarring a UI experience. So all that code was commented out. I'm now looking for good ways to revive it.
From playing around with DuckIt, I agree, there's more information out there than I'd imagined -- it's just strangled by a thick forest of spam, which you have devised a superb method of clearing.
My comment on the accuracy of zero-click info was inspired by an "acid-test" test search that turned up some zero-click information that was excellent, but some that was howlingly wrong. "Iberians" produces _splendid_ zero-click info, including a thumbnail of the Lady of Elche; "Y-haplogroup R1b" and "mtDNA haplogroup H" also worked very well; but "Iberian race" turned up zero-click info that's been refuted completely by genetic genealogy and was pseudoscience even before then.
Of course, there's an element of "garbage in, garbage out" here. I would hardly condemn a merely _mortal_ search engine for being less than up-to-date on the very young (<30 years) and distinctly arcane fields of forensic anthropology and genetic genealogy -- but Zero-Click Info mode makes it sound like this is _your_ information, not _the Internet's_ information.
Perhaps you could add a disclaimer to the effect that your elite team of ninja ducks has scoured the dark corners of the earth and returned with the best information they could find, but there's still a risk that their findings are a complete load of HealthBase?
(I would also _love_ to see a picture of your elite team of ninja ducks returning from their investigations. But maybe that's just a sign that I'm easily amused -- the kind of person who _still_ finds "All Your Base" to be funny...)
Considering the wiki nature of wikipedia a useful alternative for true/false could be the stability of a given paragraph, i.e. if it has received an high rate of revisions lately.
Assigning a "true" or "false" value seems a more complex operation that involves comparing different site's contents/assigning weight to these sites and so on, it's surely harder being able to get something objectively meaningful.
These days, the Web is somewhere between three and five titans plus a complex of secondary sites (specialist wikis, newspaper sites, news aggregators, major blogs), with a background noise that chiefly consists of spammers mirroring 90% of Wikipedia to improve their search-engine rankings.
So, in a way, this search engine is inevitable -- but I wouldn't have imagined it as early as thirty minutes ago, so it's probably not inevitable, just an idea so good that you forget that it ever didn't exist.
"Duck It!" seems to work _really_ well, too; I'm switching. My only concern is that "Duck It!" doesn't include a way of telling you whether a given paragraph is true or false...