Detailed tests of search engines: Google, Startpage, Bing, DuckDuckGo, metaGer, Ecosia, Swisscows, Searx, Qwant, Yandex, and Mojeek - LibreTechTips
libretechtips.gitlab.io
  • @[email protected]
    2•4 years ago

    It’s important to state which Searx instance is used in the testing, because every instance is different. Every instance operator chooses who to source from, and some of them even source from their own YaCy crawler.

      • @[email protected]
        1•4 years ago

        I should also mention a couple tests that would be quite useful in the search engine comparison:

        • count of CloudFlare links. CloudFlare results are useless pollution to Tor users and to everyone else CloudFlare links are privacy abusing. DDG is insanely overrated for privacy. One of the problems with DDG is a high number of privacy-abusing CloudFlare links getting high ranking results. Whereas Mojeek seems to have relatively few CloudFlare results. This is a purely anecdotal observation, however.

        • there are rumors that DDG results are consistent on a per IP address basis, but differing from one IP to another. This ultimately suggests that DDG analytics have manifested into a filter bubble – contrary to the users’ expectations.

        So it would be useful to test for presence of a filter bubble, and also to measure CloudFlare exposure. If you agree, then consider the importance of rank: a link is twice as likely to be clicked than the link that immediately follows it. So a measure of CloudFlare exposure should weigh the top results accordingly.