What government attack vectors against Bitcoin do you deem most likely to work?

From Wikipedia:

In order to prevent double-spending, the network implements some kind of a distributed time server, using the idea of chained proofs of work. Therefore, the whole history of transactions has to be stored inside the database, and in order to reduce the size of this storage, a Merkle tree is used.

So I would transact the heck out of it and make the database huge. IIRC at the moment every user needs a full copy of the database of every transaction, so if the .gov can make a multi-terabyte database a requirement, that would knock it on the head quite hard.

Also, the last time I glanced at the source code it looked quite ropey, and that makes me think it will have lots of exploitable parts lurking for the right skilled people to find and attack.