The answer to your question of attacking one of the chains seems to be valid, still.
I don't think so...
When a fork is proposed, like XT or Classic (with 75% trigger and a few weeks of grace period between the trigger and the switch to new rules), I think that:
* Before the fork, holders should pray that it resolves neatly and quickly as a non-event, and keep quiet or voice their preferences quietly, so that it does not upset the price;
* If the proposal gains some support, but neither reaches 75% nor drops back to zero, and it looks like the impasse may continue for a while, they may want to speak out for one outcome, to help break the impasse -- but that may make things worse if they themselves cant agree on which side to support. They may want to sell while the price is still OK, justin case; but that may cause the price to crash. Or they may choose to bet on the price recovering later, and keep holding.
* If the proposal gets little suport and seems to be a sure fail, the holders shoul shout it down.
* If the proposal gets 51% and keeps increasing, the holders should cheer it along.
* If the proposal gets the required support and triggers, the holders should upgrade their clients accept it, and do what they can to convince the remaining miners and players to accept it too.
* If the change has triggered, but at the end of the grace period there is still a non-negligible fraction of the miners that refuse to accep it, then the holders should try to convince the exchanges and other services to boycott the minority chain and refuse its coins, and convince the miners to sabotage the minority chain.
When a minority can veto, we have what amounts to a game of "chicken". In the game of chicken, the side perceived to be the craziest and most hell-bent on winning at any cost (including their own destruction) is the side that wins. This right now is the side of the smallblockers, unfortunately. That's why I think this will take a while to play out, longer in fact than it will take to fill the blocks and hike fees. Possibly long enough to get a network congestion failure.