Seeks – The New P2P Based Search Engine
Seeks builds on the paradox that while we have the largest, most connected network of all time, websearch is still performed by individuals, isolated from the others users (Fig.1). Since we are millions performing hundreds of millions of queries everyday, it is legitimate to believe that many of those queries are the same, or at least similar.
Seeks proposes that users share their queries to the main websearch engines. Seeks is a decentralized system that regroups users in real time based on the similarity of their queries.
Once regrouped, users can do a whole lot of things, such as rating URLs, reranking results, commenting websites, sharing queries (Fig.3), … In the Seeks architecture, these functionalities are offered through a set of plugins.
But most importantly, Seeks allows to avoid the mainstream system of crawlers, rankers and indexers of existing search engines. This is done very naturally: since users are regrouped in real time on the basis of the similarity of their queries, it is possible to propose content and URLs to an existing group, based on their queries (Fig.4). We are not talking about advertizing here, but mostly of blogs, and fresh news that you may want to see reach the users very rapidly, faster than by being crawled, ranked and indexed by Google or others.
Seeks implements a decentralized Peer-to-Peer architecture: install Seeks on your machine, server or laptop, and automatically start sharing results. While users share queries over the Peer-to-Peer network, Seeks protects your privacy by sending encrypted query fragments to peers. This scheme makes it difficult for other peers to devise your initial query.
More info, documents, source code, downloads, development, blog, FAQs, wiki and more can be found at http://www.seeks-project.info/ or https://www.seeks-project.info/ if you prefer a more secure connection.
Update: One more thing you may wish to note.
Seeks is not a distributed search engine. Seeks is a distributed hashtable of queries, users and contents (mostly URLs).