The only advantage to shifting miners from using a centralized pool to P2P pools is that it may spread out the hashing power across multiple independent groups - balkanization of the hash compute power. This doesn't solve the problem, it just disrupts it slightly.
A P2P pool is still a collective effort working together as a single block mining entity. There may not be a central server involved in a P2P pool, but the pool is still collecting transactions and creating a block to add to the blockchain.
If a P2P pool is large enough to produce 51% of the global network hash output, the pool would have the same potential to disrupt blockchain transactions as any other pool - if someone with malicious intent were able to gain control of the block creation process.
So, while a P2P pool might change where the block is created (central vs super node vs any node) and therefore the point of attack, shifting everyone from GHash to one P2P pool won't change the 51% threat.
Are the trade offs? P2P mining pools? – None – 2014-01-10T08:34:01.963