In Bitcoin land people pay transaction fees as the incentive for miners to include transactions; however, ProtoShares has shown that these fees are not always enough and that network latency advantage might be more important than the fees.
To solve this problem in future chains, I believe we can use the coinage consumed metric to dynamically reduce the difficulty of finding a block by as much as 50%. The more coinage days that are destroyed by the block, the easier it gets to mine.
The net effect is that mining is now proof-of-past-stake combined with proof-of-future-stake due to the vesting period. Spending a lot of old coins will therefore result in faster transactions because it will dramatically reduce the difficulty of finding the next block.
With this in play, the situation with y pool would be far less likely to occur because their miners would insist they include transactions.