5 ms·

Ultimately the reason you should use it is because it gets you the information you want faster and with less mental effort. That's the primary goal. I'd encoura

Ultimately the reason you should use it is because it gets you the information you want faster and with less mental effort. That's the primary goal. I'd encourage you (or anyone!) to try it for a week and then let me know how it goes with this goal in mind. We've tried to achieve that goal via UI and a number of features that may or may not appeal to you individually. We also try to concentrate on features that the major search engines won't do because of their business position. A lot of those features are listed here: http://duckduckgo.com/about.html http://duckduckgo.com/about.html What's not listed there that I think appeals to this crowd in particular is the background relevancy stuff I mentioned in my other comment and answer to your other question: http://news.ycombinator.com/item?id=979479 http://news.ycombinator.com/item?id=979479 I think what also appeals to people like me (and by extension, I assume a lot of HN types) is keyboard shortcuts, aggressive filtering of parked/spam domains, and disambiguation. We do have our own index and use it to improve relevancy as well as to add our own Zero-click Info and links to search result pages. Search any major topic and you should see a red box on top with a quick snippet and related topics. We also use it to build our own category and disambiguation pages, e.g. http://duckduckgo.com/c/Social_bookmarking http://duckduckgo.com/c/Social_bookmarking Our most recent major feature launch was Duck It!, a search view accessible from our homepage that features more topic summaries, e.g. http://duckduckgo.com/?q=tinychat&v=d http://duckduckgo.com/?q=tinychat&v=d. HN feedback link here: http://news.ycombinator.com/item?id=919505 http://news.ycombinator.com/item?id=919505 Of course, we'd welcome suggestions for additional features/improvements!

jules 17y ago
Suggestion: Improve specific searches. Broad searches work very well (try "partial evaluation") but various specific searches (try "read file .net" and compare with google; their output isn't stunning but a lot better) don't.
- epi0Bauqu 17y ago
  Thx. Our specific searches certainly vary in quality, which is something we're actively working on. Actually, I think they are more of a curve where they are really good for broad and long (very specific) but are average on the middle range. We do semantic stuff on longer queries that make the results better. But in any case, you should be generally getting different results for specific searches. Sometimes these differences will be spot on; other times, not so much. I'll take a look at 'read file .net' now--please send other examples our way: help@duckduckgo.com.
joeyh 17y ago
Hey, while you're here, one thing about Duck Duck Go that concerns me is that it can be killed on a per-IP basis by locking on "safe" search. If I accidentially do this at home, or if I find that I have to use tor too often while in coffee shops to ensure I see all search results, (or if I use tor and some idiot has forced it on for those IPs), that could be a problem. Also, I wonder how hard it would be to spoof packets to kill IPs one doesn't even have access too? Do you send a nonce back to the user and check it before completing the banning process?
- epi0Bauqu 17y ago
  The short answer is when we start running into collisions we'll change the parameters of this service, or perhaps sooner. That is, I'm fully aware of these limitations/issues and am actively monitoring them. I wanted to try the extreme case (at least for a time) to see if a lot of schools and businesses that use one static IP would find the process super simple and just use it. After all, they could set all their computers to safe search in one click. This experiment is ongoing, but it will probably evolve into a program where we have more of a back and forth opt-in process. I hope this addresses your concern!

jules 17y ago

Suggestion: Improve specific searches. Broad searches work very well (try "partial evaluation") but various specific searches (try "read file .net" and compare with google; their output isn't stunning but a lot better) don't.

epi0Bauqu 17y ago
Thx. Our specific searches certainly vary in quality, which is something we're actively working on. Actually, I think they are more of a curve where they are really good for broad and long (very specific) but are average on the middle range. We do semantic stuff on longer queries that make the results better. But in any case, you should be generally getting different results for specific searches. Sometimes these differences will be spot on; other times, not so much. I'll take a look at 'read file .net' now--please send other examples our way: help@duckduckgo.com.

joeyh 17y ago

Hey, while you're here, one thing about Duck Duck Go that concerns me is that it can be killed on a per-IP basis by locking on "safe" search. If I accidentially do this at home, or if I find that I have to use tor too often while in coffee shops to ensure I see all search results, (or if I use tor and some idiot has forced it on for those IPs), that could be a problem. Also, I wonder how hard it would be to spoof packets to kill IPs one doesn't even have access too? Do you send a nonce back to the user and check it before completing the banning process?

epi0Bauqu 17y ago
The short answer is when we start running into collisions we'll change the parameters of this service, or perhaps sooner. That is, I'm fully aware of these limitations/issues and am actively monitoring them. I wanted to try the extreme case (at least for a time) to see if a lot of schools and businesses that use one static IP would find the process super simple and just use it. After all, they could set all their computers to safe search in one click. This experiment is ongoing, but it will probably evolve into a program where we have more of a back and forth opt-in process. I hope this addresses your concern!