Prepared for Naomi Brockwell / NBTV, for the upcoming series on mesh networks, decentralized VPNs, and decentralized social media.
The shared thesis
Your recent coverage of the suspected dragnet on VPN users makes the argument for us: any privacy tool with a central operator is a subpoena away from becoming a surveillance tool. The same logic runs through your writing on decentralized infrastructure, and through your Foresight work on identity graphs. Surveillance at scale is built from aggregated metadata: who talked to whom, when, from where. The content is increasingly encrypted; the graph is not.
Mesh networks, dVPNs, and decentralized social media are usually treated as three separate exits from three separate platforms. They fail in the same place: the transport layer. A nostr note is censorship-resistant, but your IP is visible to every relay you use. A dVPN removes the single operator, but the node that carries your traffic still sees the link between you and your destination. Decentralizing the app while leaving the network observable moves the chokepoint down a layer; it doesn’t remove it. The fix belongs in the transport layer, underneath every app.
What Logos is
Logos is a modular stack for building local-first, decentralized applications: a blockchain with private proof-of-stake, a messaging layer, decentralized storage, and a shared networking layer underneath all of them. Think Linux distribution, not app: an opinionated default assembly of components you can also take apart and recombine. Overview: Introduction to Logos.
Logos Mix: privacy in the layer below the apps
A mixnet hides who is talking to whom, even from an adversary watching the entire network: messages are encrypted in layers, routed through independent nodes, shuffled, delayed, and mixed with cover traffic until traffic analysis yields nothing. Nym has proven this model as a standalone VPN product; Logos builds the same class of protection into the networking layer of an application stack, so apps inherit it rather than users buying it.
The designs split on spam protection. Gateway-based rate limiting, the Loopix/Nym approach, concentrates that decision at entry nodes. Mix uses Rate-Limiting Nullifiers (RLN): a zero-knowledge proof attached to each message that any node can verify the sender is within its rate limit, without learning who the sender is. The rate limit becomes a property of the message instead of a checkpoint in the path. RLN itself is production technology, already deployed for spam protection in other systems; what’s being tested is its use inside a mixnet. That integration was validated in simulation against testnet v0.2 and lands with v0.3.
Mix is the piece most relevant to your series. It’s a mixnet (specs here) built directly into the p2p networking layer (libp2p), beneath the modules, so chat messages, storage requests, and other module traffic get the same anonymity properties without any app opting in. (Blockchain traffic takes a different path: Blend, a specific network for validator privacy in Logos Blockchain; background in Anonymous Block Proposers.)
How it works: messages are wrapped in layered Sphinx encryption and routed through multiple independent mix nodes. Each node can see only the previous and next hop, adds a random delay, and forwards. Cover traffic makes real messages indistinguishable from noise, so an observer watching the whole network can’t correlate senders with receivers by timing. Anonymity-preserving DHT queries over mix are already live; next on the roadmap are hidden services (Tor-style rendezvous, the closest thing in the stack to VPN-like function).
Mapped to your three topics:
dVPNs. Most dVPNs decentralize the operator but keep single-hop routing, so the exit node still sees you and your destination together. A mixnet is the stronger version of the same instinct: no single node ever holds both ends, and timing analysis is defeated by design rather than by policy. Key-identity overlays (Tailscale-style meshes, including Nostr-keyed ones) solve a different problem: linking your own devices without an account. They authenticate endpoints; they don’t hide who connects to whom from whoever carries the packets.
Decentralized social media. Nostr and Mastodon solve censorship of content, not observation of the graph. Logos Messaging rides the mix layer, so who-talks-to-whom is protected at the same level as what-is-said. That’s the missing half of decentralized social media. And Nostr’s centralization problem isn’t only the IP layer. Popular clients ship default relay lists, traffic clusters on a few high-uptime relays, and NIP-65 still resolves to the same handpicked sets, so discovery and storage follow a power law. Logos has no relay role to concentrate on: discovery runs over a DHT queried through the mixnet, and messages propagate through nodes that are interchangeable by design, so no node accumulates a privileged view of the graph. Every open network fights concentration somewhere; ours shows up in node count and operator incentives, which is what the incentivization track exists to solve.
Mesh. Logos is not a mesh network. There’s no BLE or off-grid radio in the stack. What overlaps is the goal: reducing dependence on infrastructure that can be observed or shut off. If the mesh episode is about “what happens when the network itself is hostile,” Mix is the answer for the internet-connected case.
Where the project actually is
Mix is working code on a live testnet, not a product. Testnet v0.2 (released mid-2026) added cover traffic, hardened DoS and exit-node abuse protection, chat over mix, and anonymity-preserving DHT queries. Hidden services land in v0.3. Incentivized participation is designed but not live. Your audience can try the demo today; they can’t route all their traffic through it yet. This is infrastructure being built in the open, and the series could catch it at the stage where the design decisions are still visible.
What we can offer
Technical interviews with the researchers designing Mix and the networking layer, a walkthrough of why mixnets beat onion routing for this threat model, live demos on testnet, and background on the broader stack. We publish research and drafts in the open; happy to share anything ahead of an episode.
Contact: Corey Petty, corey@status.im
Roadmap files consulted (strip before sharing externally): context/roadmap/content/anoncomms/roadmap/mix.md, context/roadmap/content/anoncomms/furps/mix.md, context/roadmap/content/testnets/v01.md, context/roadmap/content/testnets/v02-release.md, context/docs/docs/get-started/introduction-to-logos.md.