The second speaker in this session at the SEASON 2025 conference is Aurelie Herbelot, presenting the PeARS search engine as an experiment in building a different type of Web search. Conventional search has significant issues with truth, biases, information verification, market monopolisation, and environmental impacts; one way of addressing these problems could be to decentralise search engine platforms, in the same way that Mastodon has sought to decentralise and federate social media.
So, there are now several PeARS instances, which are each dedicated to a specific topic of interest and curate information on their specialty topic. Linked together, this then eventually reconstructs a conventional search engine. The logic here is that human knowledge itself is also a decentralised system, so PeARS essentially seeks to model that knowledge structure.
This comes with several benefits: it creates greater freedom from economic structures; is more inclusive of specific cultures, communities, and groups; encourages the development of information literacy amongst such participating communities; offers greater control and transparency about the ranking mechanism and other aspects of the search engine process; and ultimately also reduces the environmental impact of search engines. Indeed, these benefits are usually related to and interwoven with each other: PeARS’s approach to implementing inclusive, multilingual search has also resulted in improved computational efficiency, for instance.
PeARS also provides new opportunities for providing information resources; it is possible to add distinct new pointers to information that conventional search engines would not normally cover. The project is open for participation and further federation.
