The search engine Qwant currently offers the best mix of good search results, high data protection and usability without further IT knowledge.
Why not just Google?
Search engines are among the software solutions that most of us use on a daily basis. Using a “good” search engine should therefore have a relatively high priority.
For me, the global market leader Google is not one of the “good” solutions for several reasons. The more companies like Google or its parent company Alphabet exist, whose business model is based on data and data analysis and which at the same time tramples the privacy of users, the more we citizens lose control of our decisions and ultimately of our society. The result is wonderfully described by Mark-Uwe Kling in his book Qualityland, which presents the negative effects of a strongly data-driven society in a funny and understandable form (unfortunately, available in german only). Similar works are George Orwell’s 1984 or The Circle.
Parts of these dystopias are already a reality today. Algorithms increasingly successfully influence our opinions and decisions, e.g. by advertising products or political parties. More and more algorithms are also being used in life-critical areas, such as for supporting medical decisions by doctors. Controlled automation is generally good, but the algorithms of deep learning that are frequently used today and will so in the future belong to the class of non-deterministic algorithms, i.e. whose results are not always clear and difficult to control. Companies such as Facebook use a variety of deep learning algorithms and are increasingly losing control of their decisions. In the following 6-minute video, this fact is presented to Mark Zuckerberg in a beautiful way:
So what can we do? Do not use products by Facebook & Co or use them as little as possible. This reduces – albeit very slightly – the demand for their products and thus (hopefully at some point) their legitimacy.
So let us take responsibility for our future by using products that respect our data.
What must a good search engine offer?
As always when searching for specific products, initial requirements must be created. In my opinion, the following list contains the most important requirements for a good search engine:
- Sufficiently good results in at least 80% of the search queries
- Low request response time, less than 2-3 seconds
- No user tracking, i.e. no collection of personal data by the search engine
- The company is headquartered within the EU, especially not in the USA, where all companies are subject to the Patriot Act, which – to put it simply – gives secret services access to user data.
- All servers are located within the EU – data do not leave the EU and are subject to the GDPR
- Usability for everyone, even without further IT knowledge
I deliberately limit the set of possible solutions to easy-to-use search engines that do not require self-hosting. Otherwise, I’d rule out too many non-IT users. Thus, I also exclude the use of search engines via the anonymization network Tor.
Best solution: Qwant
The five-year-old search engine Qwant meets the requirements stated above to a very high degree:
- The manufacturer is based in Paris.
- The results are usually good enough and answered quickly.
- Qwant can be used as a service by everyone, similar to Google, without further IT knowledge.
- “Qwant does not collect personal information and this will never change.” (Source, Further details of the data protection declaration)
The manufacturer is obivously convinced that advertising that is displayed in connection with the search term is interesting enough for the user.
One of the requirements Qwant does not fully meet: Parts of the search results are supplemented by the Microsoft search engine, i.e. “Bing”. Qwant indexes (i.e. searches) the Internet itself, but this does not seem to work 100% well enough yet. Unfortunately, it is not known how high this additional share is. In the long run, Qwant wants to get away from Bing of course.
Other opinions on Qwant
The prominent blogger and IT security specialist Mike Kuketz also considers Qwant to be one of the best solutions for normal users.
Critics are skeptical about Axel Springer SE’s 20 % stake in Qwant. However, Springer’s strategy is not subject to any functional restrictions and is understandable. Similar to other holdings such as VG Media, Springer is trying to reduce Google’s market share and is placing its hopes on Qwant.
Critics also mention that for many years the French legislature has been oriented more towards surveillance than data protection in a European comparison. One example is data retention. In my view, in May 2018 the GDPR introduced a high and uniform standard for data protection throughout Europe, which to me is sufficient.
Tracking in Qwant? Interesting conversation with developers of Qwant
Evaluating the source code of Qwant’s web page, I found a function called “logHandler”. At first sight, this looked like a tracker to me. I’m obviously not the only one with this find either. Thus I asked Qwant directly about this function and two other suspicious findings. See my questions/statements and their answers:
Question 1: “All ads are links to Bing service, instead of direct links -> allows tracking of clicks”
Question 2: “In the lite version, there are redirects on results instead of direct links”
Answer: Qwant Lite redirects are only used to know which results are clicked for a given search, which enables for agregate stats. We don’t collect personal data in the process. Agregate stats are necessary to rank results.
Question 3: “In the non-lite version, there is a Javscript function listening on clicks on results called e.g. “logHandler”, instead of direct links”
Answer: Same answer.
To me, this is fine, I understand their motivation. Qwant is not alone improving their internal algorithms using feedback from clicks. Later in this article, I’ll mention another search engine which might become more interesting than Qwant one day. They also track clicks to improve the search engine algorithms to eventually provide better results.
In summary, Qwant is the search engine that meets the requirements above best – despite its current partial dependence on Bing.
How do I install Qwant?
Firefox: Install this add-on – that’s it. I have not found a way to install the search engine without this extension.
Optionally, I recommend setting a blank page when opening a new window or tab. By that unnecessary information is not displayed and the browser reacts somewhat faster. To do this, enter the following characters in the address bar in Firefox and press “Enter”: about:preferences#home . Further down on the page, you see the the option to configure behavior when opening a new window or tab.
Chrome: As in Firefox, the Qwant add-on must be installed – that’s it. If Qwant is set as search engine without this add-on, this setting will be removed after a restart of Chrome.
What about other search engines?
Startpage, Metager, UnBubble, Searx.me, Ecosia, Etools etc.: These so-called “meta search engines” aggregate and display the search results of one or more of the major search engines. Personal data are allegedly not transferred as a result. The problem is that their use increases the demand for the major search engines and yet personal data is still passed on.
Ixquick: Does not exist as an independent search engine anymore. It was merged with Startpage.com.
Findx: The first search engine that met all requirements and – even more – was open source! This is heaven on earth. Unfortunately, Findx shut down in Dec 2018. See the shutdown notice of the founder.
It will be exciting to follow the progess of Qwant and Findx. IMHO, Qwant should get independent from Bing, and Findx needs to provide better search results. If that’s achieved, Findx will replace Qwant.
Until then, let’s use Qwant for its good results-privacy-mix. Qwant aims to achieve a market share of 5 to 10 percent in Europe. In France, the figure is already 5 percent. Let us use Qwant to achieve this goal more quickly.
p.s. I do not receive financial support from any search engine or any of the companies or persons mentioned in this article.Edit this post on GitHub.