Post
Topic
Board Development & Technical Discussion
Re: CoinJoin: Bitcoin privacy for the real world
by
AnonyMint
on 12/03/2014, 18:34:15 UTC
Two orthogonal issues.
First, an adversary could make a 1 Satoshi input and DOS on the (3) step. You ban that address but adversary has billions more at neglible cost.
I suppose you could set a minimum input amount to avoid this. But still no problem for the adversary, he passes his BTC through a mixer can comes to hit you again and again.
I am sorry to bring you bad news Gregory but with a non-atomic operation you can always be DOS-attacked.  Zerocoin may be the solution?

Transaction fees and confirmation times should slow down the attacker.

As for slowing down, the adversary can have many parallel addresses in play so I don't think so.

Transaction fees might work if they are significant enough. I haven't studied how much the tx fees are in Bitcoin much. I think I read that certain txs can be 0 for some cases?

If the adversary is mixing through CoinJoin transactions (hehe, uses what he also DOS-attacks against itself), then the blockchain tx fee is going to be shared between all parties of the CoinJoin transaction, so could it be insignificant?

Edit: I've just realized the adversary can eliminate the transaction fees too, by spending those banned amounts as he normally would (e.g. day trading), thus he doesn't incur any extra cost.

Edit#2: unless all decentralized CoinJoins share their ban lists (which is quite impractical to achieve as it is the antithesis of decentralization), adversary can just round-robin through them.

So I've won the argument. Checkmate.