[Rendezvous] Tracking issue on how to mitigate spam attacks

[thomaseizinger]

This issue tracks the to-be-defined way of how the rendezvous protocol can mitigate spam attacks against rendezvous points.

Originally, PoW was considered a viable option but this approach is now being considered inadequate (see https://github.com/libp2p/specs/pull/334 and https://github.com/libp2p/specs/issues/337).

Alternative ideas mentioned so far are:

• Verifiable / deterministic randomness in how the rendezvous node was chosen among a fleet of nodes. The exact behaviour is yet to be defined.

[calummoore]

Could we not use drand for this?

[thongpv87]

How about using verifiable credential. The app is also an issuer, and any machine need to publish/subscribe into the channel need read/write credential from the issuer.