Bitcoin uses a distributed timestamping server. The paper literally states that "new transactions are broadcast to all nodes". This effectively means that they need a global broadcast to the majority of nodes for every bit coin transfer.<p>This doesn't scale in terms of number of nodes, and it doesn't scale in number of transactions. They can not compensate for an increase in the number of transactions by adding more nodes because instead of distributing the work load over the nodes, they are in fact replicating it.<p>The rationale seems to be that as long as you have more honest nodes than malicious ones the system will not be hijacked. Of course, the chance that a Russian botnet has more CPU power than you do is quite real. They go on to say that if an attacker has more computational power than everyone else they would do better to use it mining coins than to break the system, and they seem to leave it at that. Counting on the rational behavior of potential attackers does not seem like a good strategy to me.<p>Maybe I'm missing something here, but I'm pretty sure this thing will simply not scale. It reminds me of Gnutella, which implemented search through scoped flooding and somehow promised both scalability and coverage. I have the same feeling here. Scalability _and_ security in a peer-to-peer network? Surely you jest.<p>When something sounds too good to be true it probably is.<p>edit: link to the paper: <a href="http://www.bitcoin.org/bitcoin.pdf" rel="nofollow">http://www.bitcoin.org/bitcoin.pdf</a>