Strong privacy for ordinary people — from advertisers, data brokers, and
casual surveillance — built so that abuse is preventable and provable. The opposite
of a no-rules darknet. Pedophile-free, trafficking-free: safety by architecture, not by hope.
● The demo's addressing + blocking run in your browser — no server, no plaintext store
The honest tradeoff (read this first)
You cannot have perfect anonymity and reliable
abuse-prevention at once — they pull in opposite directions. Anyone selling both is
selling you something. LightWeb makes a deliberate, stated choice: accountable
pseudonymity over absolute anonymity. Strong privacy from the public; built so the
worst things are stoppable, not perfectly deniable.
See it: a relay blocks known-bad without decrypting
Published content gets an address that's a
deterministic fingerprint of the content (convergent encryption). A relay matches that
address against a known-bad denylist (an NCMEC-style hash set) — and refuses it
without ever holding a key or seeing the words. Try a normal post, then the
simulated known-bad item.
Safety by architecture
Vouched, revocable membership (web-of-trust) — no anonymous throwaway flood.
Non-repudiable authorship — every object is signed; publishing isn't anonymous, and it's revocable.
Block known-bad without surveillance — relays match content addresses, never decrypt legitimate traffic.
Tamper-evident moderation — report → confirm → denylist → revoke, on a hash-linked trail.
Private 1:1 messages stay ordinary random-key end-to-end — not relay-scanned; abuse there is handled on-device by Qapha.
The honest limits (loud)
Exact-address blocking catches known bad content only — one
changed pixel changes the address and evades it. Real prevention needs perceptual
hashing (PhotoDNA-style) + human review + NCMEC reporting — named as
roadmap, not claimed. This reduces and exposes abuse; it does not end it. Strong,
lawful privacy for ordinary people is the goal; shielding predators is explicitly not.
the proof ↗