Sign in or Join FriendFeed
FriendFeed is the easiest way to share online. Learn more »
Louis Gray
Bing vs Google vs Yahoo: the Blind search engine test - http://www.istartedsomething.com/2009060...
Weird. According to the twelve tests I just ran it seems I should be using Yahoo for best results and that's actually the one I'm least inclined to use. - Mark H
Here's the test I did between Bing and Google. Bing is absolutely horrid: http://friendfeed.com/jamespo... - James Poling
So far, I've picked Google a little more than the others, but not by much. - Louis Gray
Current results: "Google: 46%, Bing: 34%, Yahoo: 20% | 6905 votes" but I worry a little bit about self-selection bias--a MSFT employee (Michael Kordahi) did this, so some of those votes are perhaps a little more likely to be MSFT employees. Barry Schwartz did a similar comparison a while ago, but the comparison had more granularity because people could rate search results from 1 (bleah) to 5 (really relevant). Here's more info: http://www.rustybrick.com/rustyse... - Matt Cutts
"more granularity because people could rate search results from 1 (bleah) to 5 (really relevant)." Not sure that makes a big difference, with enough votes you still get 1st, 2nd and 3rd positions, including granular percentages (and you may accumulate votes faster this way as less thinking is required). Of course if you're looking for data on how often a search engine REALLY screws up, this method may not work well. - Philipp Lenssen
"a MSFT employee (Michael Kordahi) did this" Didn't know this, added an update at Blogoscoped. "so some of those votes are perhaps a little more likely to be MSFT employees" Why? (Not that I'm doubting it, but it's a blind test so you can't just vote for Bing results, unless perhaps you hack the thing...) - Philipp Lenssen
Philipp, the nice thing is that with Rusty's poll, he provided a downloadable spreadsheet afterwards. For each row, I think he had the query, the result, and the relevance rating (1..5). That's much easier to dig into. Philipp, MSFT employees have been using Kumo/Bing for the last few months, so my guess is that they're more likely to prefer Kumo/Bing results. The fact I first saw it on istartedsomething.com (MSFT-focused blog) and it's by a MSFT employee makes me think that MSFT folks are more likely to see this poll before Google folks. Another example, Dare tweeted "According to blind search engine test at http://blindsearch.fejus.com people prefer Bing almost as much as Google and twice as much as Yahoo," and MSFT people are probably a little more likely to follow Dare. - Matt Cutts
Matt: "MSFT employees have been using Kumo/Bing for the last few months, so my guess is that they're more likely to prefer Kumo/Bing results." But why, considering it's a blind test? (I have some guesses as to why but I'd like to hear your opinion.) And by that reasoning would you also say that everone who's been using Google for the last few months and beyond is, for that reason, more likely to prefer Google results, even in a blind test which does not show the logo? - Philipp Lenssen
Yahoo went from 20% to 35% in just a couple hours and continues to rise. Seems like a large coincidence, but who knows? :) That's the problem with this sort of self-selecting poll; you're not sure about the demographics of who is voting. - Matt Cutts
Yup, Yahoo went from 20% to 40% in ~2-3 hours. I believe this poll has been astroturfed. :) Did someone invite 4chan into the party or something? - Matt Cutts
The app doesn't use any captchas so maybe it's abusable (not that captchas would stop 4chan, as we saw with the Time poll!). - Philipp Lenssen
Yahoo jumped from 40% to 44% in the past few minutes... - Simon
Yahoo went from 35% to 42% in 20 minutes. I have a hard time believing that's organic. Ooh, now it's at 46%! Whee! :) - Matt Cutts
Funny thing is, I could tell which one was google simply by the results. It was hard to find a search that I couldn't tell which one was google. I'd like to see the experiment replay with bogus headers half the time and blind the other half. It was suggested in a test (I think maybe Louis wrote about it?) that if google's name was placed on search results, they would prefer those, not matter whose search results they were. - guruvan (Rob Nelson)