Hyphanet (Freenet): The Original Anonymous P2P Data Store

The origin story

Ian Clarke developed the original Freenet concept as an undergraduate student project at the University of Edinburgh in 1999. He turned it into a thesis, then a 2001 paper co-authored with Oskar Sandberg, Brandon Wiley, and Theodore Hong — "Freenet: A Distributed Anonymous Information Storage and Retrieval System." CiteSeer ranked it among the most-cited computer science papers of 2002. The project shaped the academic and engineering vocabulary of censorship-resistant networking that I2P, Tor, and IPFS later built on.

In mid-2023, the project was renamed to Hyphanet to disambiguate from Freenet Networks LLC, an unrelated company that had been using the Freenet name. The technology and team are continuous from the 1999 origin.

What it actually is

Hyphanet is a peer-to-peer platform for anonymous, censorship-resistant publication and retrieval of data. Unlike Tor (which is a transport overlay for accessing existing internet services) or I2P (which is also primarily a transport), Hyphanet is a data store. Content lives inside the Hyphanet network as encrypted fragments distributed across participating nodes.

This has significant implications. Once content is uploaded, it persists as long as nodes choose to keep cached copies. The publisher can disappear, the originating computer can be seized, and the content is still retrievable. Popular content replicates more widely; unpopular content fades to a few nodes that hold it.

How content addressing works

Files are split into fragments and stored at locations derived from cryptographic keys. Two main key types:

• Content Hash Keys (CHK) — SHA-256 hash of the encrypted content. Immutable: the same content always produces the same CHK.
• Signed Subspace Keys (SSK) — uses public-key cryptography for verifiable pseudonymous publishing. The same publisher can update their content while readers verify it's from the same identity.

Routing is content-addressed and decentralized — nodes specialize in keys near their assigned location over time, an emergent property the original papers called "spontaneous symmetry breaking."

Opennet and Darknet modes

Hyphanet supports two connection topologies, simultaneously if desired:

• Opennet: nodes auto-discover each other via central seed servers. Easy to use, more vulnerable to surveillance and Sybil attacks because any attacker can join.
• Darknet (also called friend-to-friend or F2F): nodes only connect to peers whose cryptographic identities you've manually exchanged. Much stronger anonymity — an external observer can't even tell you're using Hyphanet because the network has no central entry points to monitor.

Darknet mode is the recommended configuration for serious adversarial threat models. It requires social bootstrapping (you need at least a few real-life trusted contacts also running Hyphanet) but provides anonymity properties no other public network matches.

What's actually on Hyphanet

The plugin ecosystem includes:

• Sone — pseudonymous social networking (Facebook-style microblogging).
• Infocalypse — decentralized version control built on Hyphanet, akin to Git over Freenet.
• FMS (Freenet Message System) — anonymous messaging with trust-based spam filtering.
• jSite — static-site publishing tool.
• Frost — file distribution.

The user base is much smaller than Tor's millions; the use cases are correspondingly narrower. Journalists handling leaked documents, activists publishing under hostile regimes, and academic researchers studying censorship-resistant systems are the typical population.

The legal history

Hyphanet has been the subject of multiple law-enforcement investigations attempting to deanonymize users sharing illegal content. The 2011 "Hops-To-Live" episode involved investigators misunderstanding how Hyphanet's probabilistic hop counter actually worked. Subsequent academic papers by Brian Levine in 2017 and 2020 made similar protocol-modeling errors per the Hyphanet maintainers.

A 2020 US Court of Appeals decision noted that one law-enforcement deanonymization method "could not differentiate between an active requester and a passive node" — i.e., it couldn't reliably tell whether a Hyphanet user holding a fragment had originally requested the content or was just routing it. The legal cases proceeded on other evidence; the protocol-level deanonymization claims were undermined.

Recent vulnerabilities

Researchers have identified Routing Table Insertion attacks that, under specific network conditions, can let an attacker increase visibility into request paths. Mitigations are documented and the protocol design continues to evolve. As with every anonymity system, no design eliminates all attacks — the meaningful question is whether the residual risk fits your threat model.

How it compares

• vs Tor: Tor is for anonymous access to the regular internet. Hyphanet is for anonymous publication and retrieval of data that lives inside the network. Different problems, complementary tools.
• vs I2P: I2P also focuses on services inside the network rather than clearnet exits, similar to Hyphanet's data-store model in spirit. I2P is transport-layer; Hyphanet is application-layer.
• vs IPFS: IPFS is a content-addressed distributed file system focused on performance and not privacy. Hyphanet sacrifices performance for anonymity.

The verdict

Hyphanet is one of the most influential anonymity research projects ever built. The 2001 paper laid the conceptual groundwork for an entire generation of censorship-resistant systems. The actual network in 2026 is small and slow but operationally still works — and for the specific niche of publishing documents that need to survive their author and source computer, it's still arguably the cleanest design.

For most users wanting browsing privacy, Tor is the better daily tool. For users specifically wanting to publish or retrieve documents that should survive arbitrary takedown attempts, Hyphanet's darknet mode is unique.