have we considered reversing the connections for Broflake? I think we've seen a lot to suggest that inbound connections are less heavily scrutinized than outbound ones. I'm really not sure how well substantiated that is though.
Concretely, this means making sure that the uncensored peer is the one who initiates the DTLS handshake.
IIRC, with WebRTC, the DTLS client/server roles are determined during signaling. I can't remember exactly how this works. Does it correspond to the controlling agent and controlled agent that are designated during ICE?
Notably, connecting from uncensored to censored is an essential aspect of this grant commitment. If we can make that happen with Broflake, we could probably meet the grant commitment with this.
On Slack, @hwh33 said:
Concretely, this means making sure that the uncensored peer is the one who initiates the DTLS handshake.
IIRC, with WebRTC, the DTLS client/server roles are determined during signaling. I can't remember exactly how this works. Does it correspond to the controlling agent and controlled agent that are designated during ICE?