--- Log opened Fri Jul 19 00:00:02 2013
11:13 < jgarzik> petertodd, RE identity + IRC replacement via P2P flood-fill network...  do you think a PoW element should be included, a la BitMessage?  Or just rely on identity cost and shared opinion
11:13 < jgarzik> ?
11:14 < petertodd> I think identity cost is enough because the domain over where the message is sent is fixed - there's no re-use potential.
11:15 < petertodd> rb
11:15 < petertodd> brb
11:47 < petertodd> back
13:08 < petertodd> jgarzik: I suspect dealing with the graph of trust is going to be tricky... smells like a computationally intensive graph problem.
13:09 < jgarzik> indeed
13:09 < petertodd> One subtlety is you have to apply the same anti-spam rules to messages stating who you trust.
13:10 < petertodd> The other one is how do you find peers who have similar ideas of what to filter.
13:11 < petertodd> For v1.0 maybe the right approach is to not do it as a graph, but as a simple accounting of the sum sacrifice ignoring someone.
13:13 < jgarzik> certainly easier
13:14 < jgarzik> though disappointing there must be some sort of state
13:14 < petertodd> Yes, more minimal state, but that's still state.
13:14 < petertodd> At least it's state without user-controllable parameters - like bitcoin peers can sync to each other and come to consensus.
13:15 < jgarzik> also I wouldn't want everyone in the world on the same P2P network.  My proxy would join user-specified networks, each with their own DNS seeds or methods of address gathering/bootstrapping/sharing.  i.e. join "freenode" network with specified network magic and DNS seeds
13:15 < jgarzik> enables darknets and scaling
13:16 < petertodd> For bitcoin P2P flood fill jdillon suggested that you split things up into different domains by a simple UUID.
13:16 < petertodd> Nodes can even advertise a bloom filter of what UUIDs they participate in.
13:30 < jgarzik> Perhaps, but ultimately I think people should be able to avoid transiting data for networks they care nothing about
13:31 < jgarzik> Proxy can talk to multiple P2P networks just as easily
13:32 < petertodd> Point is with those UUIDs that's exactly what happens, yet to an observer the behavior of all those networks is identical.
13:33 < petertodd> Also allows for a meta-UUID(s) to make peer discovery for a given UUID easier.
16:38 < sipa> every time i (re)join here, it seems the number of people has grown :)
16:39 < petertodd> we'll have to make -gods eventually
16:39 < sipa> well, there's always #bitcoin-satoshi above...
16:40 < petertodd> heh
16:43 < gmaxwell> this is the best bitcoin channel.
16:43 < gmaxwell> well, other than the one where you have to solve the cryptographic puzzle embedded in the blockchain to join...
--- Log closed Sat Jul 20 00:00:05 2013