However, as others have mentioned 'OpenTransactions' already seems to be approaching 'p2p contracts' and that may be objectively better if one is particularly attached to the current blockchain. IMO, CBM is a bit more elegant (and it allows the miners to get directly involved in predictions and doesn't preclude the additional options/contracts markets from developing around it either). There is something to be said for leaving a working system alone though and building onto it.
One problem with using the existing block chain is that it ties up BTC in escrow. Suppose we want to wager 1 BTC on bitcoin difficulty 1 year in the future. To do this, we would need to hold 1 BTC in escrow for the next year. If on the other hand we loaned out the BTC at a hypothetical risk-free interest rate of 1%, we could earn around 0.01 BTC in interest during this year. Thus, there is something like a 1% annual tax on speculative transactions that use regular bitcoin.
But, different brokers will have different margin requirements. Surely those won't always be 1:1.