The "brainstorming" is simple: stop adding any transactions to blocks once you have 51% of the hash power, and ignore any other validly mined blocks. If you can't even move BTC, it is effectively toast.
This is a very interesting attack and similar to scenarios I had in mind as well. May I brainstorm on a defense?
We open up a second block chain with a fresh genesis block. We build the genesis block with lots of coinbase transactions which restore the account values as they were in the old block chain. Now we have a copy of the block chain. If the good guys have 49% and the attacker has 51% of hash power on ONE chain, the attacker wins. With two chains and hash power switching back and forth between two chains, there probably once in a while will be the chance to be head on at least in one of the chains for the good guys. However, I am not sure whether a disrupting attacker still is in the same situation as with a single block chain.
Now assume we have a large number of parallel chains, competing with each other (if we use chain identification codes, this could even be done as part of the same software and the same data structure). I have some hope that there are scenarios where an attacker cannot stop all chains from progressing.
Let me compare this to PirateBay and email. Stopping filesharing in the PirateBay model is easy, in the email model it cannot be stopped. So: Could a world with a large number of parallel Bitcoin chains be more resilient than a world with a single Bitcoin chain?
Well if they don't call you out for spreading FUD and try to marginalize your points, it would make it harder for newbies to throw cash into the bitcoin black hole.
Never saw it that way. It could be a valid point.