DuckDuckGo conducted a study in June 2018 that took a closer look at how Google processes results search for logged in, deleted and anonymous users.
Although the study of DuckDuckGo may seem quite biased, given that even a search engine is a direct competitor to Google, their findings are interesting for anyone who wants to better examine Google's internal features.
According to the findings of the study, Google's personalized results are displayed depending on the personal information previously collected about the specific users (e.g. search, tour and record purchases), automatically adding each of them to a so-called “filter bubble”.
Google uses this filter bubble to tailor particular search results that appear to a user who is searching for a particular term or phrase, so most of the study participants see results unique to them.
In addition, although Google claims that it worked to remove the "filter" function that showed the results that users would like, DuckDuckGo discovered, with the help of the 87 volunteers in the study, that users saw results uniquely in each, even when disconnected or in private browsing mode.
Users are trapped in Google's "bubble filter" even after logging out and browsing anonymously.
Google also showed some search results to some of the participants in the study, which other users did not see as if they were looking for them, although they were incognito or disconnected. This in itself means that if Google had eliminated the filter bubble bias, they would all have the same results, but this is not the case, and instead, by linking a user to the internet, the Google search engine uses the information it already knows about providing personalized results.
According to DuckDuckGo, using the private browsing feature or if you are disconnected from Google's account, do not expect the anonymity imagined by most people. In fact, you can not use Google search and avoid its bubble filter.
This is because fingerprinting techniques, as well as IP addresses, can be used to monitoring people even when they think they are visiting the web anonymously. At this point it might be worth doing a test yourself. Sign in normally to a Google account, do a search with a random English term and see the first 5 search results. Log out, go to your browser incognito and search with the same term. And finally open Tor or an Opera that has a proxy and do the same search again. The first 5 results for these 3 searches are compared.
All DuckDuckGo data collected during duration of the study, as well as the instructions and search results obtained, together with the code used to analyze the information; are free and available on its website and are ready to be examined by anyone interested in verifying their findings.
