When setting up a browser on a shared workstation in a company or library, the question of the default search engine becomes a real concern. Google remains the reflex choice, but each query feeds an advertising profile linked to the IP address, cookies, and browsing history.
On a workstation used by multiple people, this profiling mixes profiles and exposes data that do not concern the next user. It is this type of daily situation that drives the search for an ethical search engine capable of providing relevant results without storing personal data.
You may also like : Major accident on the Chambéry VRU: disruptions and alternative routes to prioritize
Advertising profiling and search results: what changes in practice
With a traditional engine like Google, the displayed results depend on your history, location, and dozens of behavioral signals. Two people typing the same query receive different pages. This filtering, often referred to as a “filter bubble,” guides navigation without the user being aware of it.
A privacy-respecting search engine removes this layer of personalization. The results are identical for everyone, which facilitates collaborative work: one shares a link, and the other person sees the same results page. For those looking to explore this approach, a Francophone project details this functionality at https://www.seeks.fr/, with accessible technical documentation.
Recommended read : The subtleties of installing a toll badge: common mistakes and best practices
The European regulatory framework reinforces this issue. The Digital Markets Act (DMA) and the Digital Services Act (DSA) now impose transparency obligations on major engines classified as “gatekeepers” regarding profiling and result personalization. Alternative engines that do not engage in targeted advertising are automatically compliant with these new rules without having to modify their architecture.
Collaborative and open-source engines: SearXNG, YaCy, and others
DuckDuckGo or Qwant are often mentioned when it comes to privacy, but these engines remain centralized services. Another family exists: collaborative engines, built on open code, where each user can contribute to the infrastructure.
SearXNG: a self-hostable meta-engine
SearXNG aggregates results from multiple sources (Google, Bing, Brave Search, etc.) without transmitting the user’s identity to these sources. Its code is open source, and any organization can install its own instance on a server. Hosting your own meta-engine ensures that search logs remain under internal control.
In practice, SearXNG is installed on a VPS or local server, the source engines are chosen, and queries pass through this intermediary instance. Feedback varies on speed depending on the number of activated sources, but the relevance of web results remains solid for daily use.
YaCy: the peer-to-peer engine
YaCy works differently. Each node in the network indexes a portion of the web and shares its index with other nodes. There is no central server, no company behind it, no advertising model. The index is built collectively.
This model suits specific uses: indexing an intranet, searching within a closed document corpus, exploring the web without relying on a third party. For general web search, the size of the index remains smaller than that of commercial engines, which limits coverage.
Economic model without targeted advertising: how these engines are funded
The question of funding comes up systematically. If a search engine does not sell data and does not display targeted advertising, how does it pay for its servers?
Several models coexist:
- The paid subscription, tested by engines like Kagi, which offers ad-free access for a monthly fee. This model aligns the interests of the service with those of the user rather than with those of the advertiser.
- Non-targeted contextual advertising, used by Qwant and DuckDuckGo. The ad displayed depends on the keyword typed, not the user’s profile. A search for “cargo bike” shows an ad for a bike seller, without exploiting browsing history.
- Donations and associative funding, which support projects like SearXNG or YaCy. The code is maintained by volunteers and contributors, sometimes with public grants or funding from foundations.
None of these models generate the margins of Google. The trade-off is a complete absence of resale of personal data and a transparent relationship between the service and its users.
Setting up an ethical search engine as the default: points to check
Switching to an alternative engine is not just about changing a URL in the browser settings. A few criteria deserve to be checked before making a choice.
- The logging policy: does the engine retain queries, even anonymized ones? Self-hosted SearXNG allows for complete disabling of logs. DuckDuckGo and Qwant claim not to store identifiable history.
- The country of server hosting: an engine hosted in the European Union is subject to GDPR, which offers a more protective legal framework than hosting in the United States.
- Dependence on a third-party index: Qwant partially uses Bing’s index, as does DuckDuckGo. This dependence does not compromise privacy if queries are anonymized before being transmitted, but it means that the quality of results partially depends on an actor that does engage in profiling.
- Compatibility with browser extensions: some engines offer an extension that forces the default engine and blocks third-party trackers simultaneously.
In a professional IT environment, deployment is done via group policies (GPO) or browser configuration files. The search URL is defined, the setting is locked, and all workstations switch without manual intervention on each machine.
The choice of an ethical and collaborative search engine relies on a trade-off between index coverage, funding model, and level of control over data. The tools exist, they are functional, and the European regulatory framework is now pushing in their direction.