Re:Forget Cuil, check this out: one guy and a bunch of pcs
by michaelfairley on Wednesday July 22, 2009 @08:37PM
(#718827)
This study has an inherent flaw: the queries used were selected from the most common queries. Answering common queries is (relatively) easy, but Google claims (and other independent studies have shown) that Google is much much better than most of its competition for returning good results for uncommon queries (the "long tail" if you will).
Edit: I just (briefly) tested this myself. Searching for for my username on Gigablast returned 3 results: 2 of my HN comments and a wordlist. Google on the other hand found 3k results, with the top results being my linkedin and facebook profiles (which I would deem the "best results").
Re:Forget Cuil, check this out: one guy and a bunch of pcs
by die_sekte on Wednesday July 22, 2009 @10:35PM
(#718945)
Can't agree more with that. I searched for some Cocoa part which was a tie between Google and Gigablast; a torrent which Gigablast could find, but at its origin; and the AD&D Sorcerer which Gigablast couldn't find anything about.
Gigablast might have better results for common queries, but those matter hardly to the typical HN user.
Re:Forget Cuil, check this out: one guy and a bunch of pcs
by breck on Wednesday July 22, 2009 @08:00PM
(#718792)
Not to knock them, from the little I know I'm very impressed, but this study showing Gigablast to be better than Google was conducted by Gigablast.
I just thought I'd mention it to save other people the 5 minutes I spent researching it because it just seemed to good to be true.
It's a dated study of course, and I can't find a reference for it (which doesn't help) but it impressed me quite a bit.
edit: found a link: http://www.searchenginejournal.com/gigablast-better-than-goo...
I misremembered the percentage though, not 66 but 55.