4 ms·

We would appreciate any specific examples at https://duckduckgo.com/feedback.html https://duckduckgo.com/feedback.html -- it's a great way to improve! But high

We would appreciate any specific examples at https://duckduckgo.com/feedback.html https://duckduckgo.com/feedback.html -- it's a great way to improve! But higher level, !g roughly accounted for 1% of queries yesterday, which also includes all the cases where we ran out of results and people clicked on our Try google link.

heyitsnick 15y ago
Here's a real example that just occurred. A Search for "jquery select visited anchor" on DDG returns the jquery selectors docs first (okay match) and jquery.com match second (useless). A search on google.com, the first 2 results are SO answers that answer exactly my question. My point isn't specifically you rank SO lower than it should imo (although i do think this is the case) - its in general, for me, you return roots of homepages or brands at hte top when the query is clearly looking for something very specific.
- epi0Bauqu 15y ago
  I'm not seeing that result set at all. Do you have a region set? I'd appreciate emailing me a screen shot if you don't mind at yegg@duckduckgo.com Btw, we index SO ourselves, e.g. https://duckduckgo.com/?q=nginx+apache https://duckduckgo.com/?q=nginx+apache We just have a high bar to show it, but are working on that piece so it comes up more. It's also full punctuation.
  - heyitsnick 15y ago
    Email sent, but for other readers' reference, shots taken in anon chrome: ddg: http://i.minus.com/iKakLPppwCv14.jpg http://i.minus.com/iKakLPppwCv14.jpg google: http://i.minus.com/iZvY24AxfnHRR.jpg http://i.minus.com/iZvY24AxfnHRR.jpg
dredmorbius 15y ago
You're probably going to do a lot better automating data collection based on DDG/!g searches. I really don't know how you'd score for quality, but at least take this as an example of a search whose results were not satisfactory. Might also be the sort of thing to look at with a specific focus group. I know that the trend is to use a naive audience, but you might want to specifically recruit self-selected power-users. I'd also design such a study to blind the result sets. I know Microsoft raised a bit of a stir claiming that their search results in blind testing were better than Google's a fair bit of the time. Problem at least personally with those results are that the two things I care about in search are relevance, and credibility in protecting user interests. Microsoft have a very poor track record in that latter. I don't Bing. Period.
Arelius 15y ago
I have an idea, how about you add a feature, opt-in where you track the searches I do, and then do again with an added !g, at which point, store those in a list where I can go back and add feedback to each one where I got improved google searches, Or the ability to leave up a DDG feedback window when I do those searches. If you make it easy for me to give you feedback, I'll give you limitless amounts of it.

heyitsnick 15y ago

Here's a real example that just occurred. A Search for "jquery select visited anchor" on DDG returns the jquery selectors docs first (okay match) and jquery.com match second (useless). A search on google.com, the first 2 results are SO answers that answer exactly my question. My point isn't specifically you rank SO lower than it should imo (although i do think this is the case) - its in general, for me, you return roots of homepages or brands at hte top when the query is clearly looking for something very specific.

epi0Bauqu 15y ago
I'm not seeing that result set at all. Do you have a region set? I'd appreciate emailing me a screen shot if you don't mind at yegg@duckduckgo.com Btw, we index SO ourselves, e.g. https://duckduckgo.com/?q=nginx+apache https://duckduckgo.com/?q=nginx+apache We just have a high bar to show it, but are working on that piece so it comes up more. It's also full punctuation.
- heyitsnick 15y ago
  Email sent, but for other readers' reference, shots taken in anon chrome: ddg: http://i.minus.com/iKakLPppwCv14.jpg http://i.minus.com/iKakLPppwCv14.jpg google: http://i.minus.com/iZvY24AxfnHRR.jpg http://i.minus.com/iZvY24AxfnHRR.jpg

dredmorbius 15y ago

You're probably going to do a lot better automating data collection based on DDG/!g searches. I really don't know how you'd score for quality, but at least take this as an example of a search whose results were not satisfactory. Might also be the sort of thing to look at with a specific focus group. I know that the trend is to use a naive audience, but you might want to specifically recruit self-selected power-users. I'd also design such a study to blind the result sets. I know Microsoft raised a bit of a stir claiming that their search results in blind testing were better than Google's a fair bit of the time. Problem at least personally with those results are that the two things I care about in search are relevance, and credibility in protecting user interests. Microsoft have a very poor track record in that latter. I don't Bing. Period.

Arelius 15y ago

I have an idea, how about you add a feature, opt-in where you track the searches I do, and then do again with an added !g, at which point, store those in a list where I can go back and add feedback to each one where I got improved google searches, Or the ability to leave up a DDG feedback window when I do those searches. If you make it easy for me to give you feedback, I'll give you limitless amounts of it.