Skip Navigation
Applied Paranoia @hackertalks.com
jet @hackertalks.com

SimpleX founder Evgeny Poberezkin Interview - Techlore

SimpleX Chat is the only messaging network where users have no identifiers—no phone numbers, no usernames, no user IDs at all. Henry interviewed founder Evgeny Poberezkin about how unidirectional message pipes create a network where servers don't even know users exist, why this isn't federation, how it compares to Signal and Session, and why the company is based in the UK despite encryption battles.

It's interesting that he see's his project as just a distributed message queue and not really as the simplex chat application

The discussion about how federated censorship could be compounded from the users perspective... its a good motivation for a unopinionated message queue, users get the authority not the servers.. fungible network operators.

Philosophy and motivations

  • The internet should work like the early web: people own their space, audience, and rules.
  • Platforms replaced the web by taking ownership; the goal is to rebuild the web’s utility without surrendering control.
  • Privacy is not about being unseen; IP addresses are observable; the goal is minimizing who must be trusted.
  • Avoid building a new centralized service; publish open software so anyone can run servers; operators are replaceable.
  • Build for the 99%: strong defaults and UX so security is not limited to experts and custom ROM users.

What SimpleX is building

  • A messenger without user IDs; connections start via one-time links or QR codes.
  • Messaging uses unidirectional message queues; each direction can use different relays.
  • Each contact can use different relay servers; rotation limits correlation and reduces single-operator power.
  • Relays route encrypted blocks and cannot enumerate users or social graphs.

From messenger to “next web”

  • SimpleX extends into primitives: messaging, groups, channels, bots, and “sites”.
  • Communities become user-owned spaces like websites: owners control content, moderation, and membership.
  • Scaling model: many rooms and roles; a 100,000-member community should not be one chat.
  • Target experience: Discord-like communities with far more owner and user control.

Decentralization and moderation realities

  • Federated networks form clusters where admins own accounts and can coordinate policy and censorship.
  • If a few percent of nodes are captured, randomized routing can still be forced into an attacker’s path.
  • Better model: many independent operators with low individual visibility; users choose and can switch.

Metadata and transport privacy

  • IP metadata is theoretically observable; Tor/VPN/mixnets change who can see it, not whether it exists.
  • Padding sends fixed 16KB blocks so relays can’t infer content size or activity type.
  • A relay sees counts, not contacts; 100 messages could be 1 or 100 recipients.
  • Roadmap includes supporting alternative transports like I2P and mixnet-style routing.

Security engineering posture

  • Deniability matters for casual conversation; OTR introduced practical repudiation and forward secrecy.
  • Two security audits completed; recurring audits planned.
  • Spam and abuse controls avoid identifiers; optional user addresses can be deleted or rotated.

Business and distribution constraints

  • App stores gate distribution; sideloading and F-Droid matter for reach.
  • Funding reality: privacy tech competes with products backed by 100x–500x more investment.
  • B2B2C model: communities pay so members can be free; 80/20 traffic economics inform pricing.

References

Comments

1