This relied on being able to send lots of messages, and having the user visit a link contained in them. The first issue can be fixed by upping the proof of work required to send a message, although this will not stop a determined attacker who has lots of cycles to throw at the problem. As for the second issue, users should not be visiting links from addresses they do not trust. As with most anonymity systems, it is only as good as you treat it.
> The first issue can be fixed by upping the proof of work required to send a message, although this will not stop a determined attacker who has lots of cycles to throw at the problem.
Could you implement something like IRC's "flood prevention" in a proof-of-work based consensus algorithm -- so sending messages closer together costs prohibitively more?
The network could require, say, the work in a transaction to be proportional to ∑(1/message dt) for the messages signed by the transaction.
> Could you implement something like IRC's "flood prevention" in a proof-of-work based consensus algorithm -- so sending messages closer together costs prohibitively more?
This seems easily circumvented by creating lots of identities. Though maybe creating an identity could be costly?
