Tether tests Wikipedia on 100 nodes
Tether says its peer-to-peer search team has run a test that placed the full Wikipedia archive across 100 distributed nodes. The search interface stayed usable, and results came back through the network. CEO Paolo Ardoino shared the results on Sept. 8 and called the performance impressive. He also said the team is building what it calls the Unstoppable Search Engine. The idea is to let the system grow across thousands of nodes without depending on one central search provider.
How the system is meant to work
The project builds on Hypersearch, which Ardoino first described in April. It uses distributed hash table technology, plus HyperDHT from the Holepunch stack. That setup lets peers find each other and exchange data directly. Instead of keeping the search index and query process with one operator, the work is spread among participants. Fault-tolerant structures are supposed to help the data organize itself. If some nodes leave or fail, the search service should remain available. Tether also wants censorship resistance. That is a common goal in decentralized systems, though it is hard to prove at scale.
Tether has not said when the public will get access. The latest demo is a controlled test using a large existing document set, not the open web. Important questions remain. How would the engine crawl and index the wider internet? How would ranking work across thousands of nodes? What happens with spam, manipulated results, or conflicting data? Those problems are not solved by distribution alone. I think the Wikipedia test is useful, but it is still a test.
Part of a wider P2P push
Search is not Tether’s first peer-to-peer project. Tether and Bitfinex backed Holepunch and Keet, an encrypted communication app. Tether later used similar ideas in Bitcoin mining tools, including Mining OS and an open-source Mining Development Kit. Its QVAC work also aims to run AI models on local devices instead of centralized cloud systems. The company is hiring engineers for its P2P search team. Job requirements include distributed databases, CRDTs, ranking, inverted indexes, lexical and semantic search, NAT traversal, and network protocol work. The listing names HyperDHT, Hyperswarm, Hypercore, Hyperbee, Hyperdrive, and UDX.
For now, Hypersearch remains under development. Ardoino’s demo shows Wikipedia search running across 100 nodes. Tether is designing the architecture to reach thousands. Whether that becomes a real public search engine will depend on the next set of tests, and on how well it handles the messy open web.
