It looks like Bitcoin has been on the radar of Microsoft's research group for awhile now. This was from back in June 2012:
On Bitcoin and Red Balloons
http://research.microsoft.com/apps/pubs/?id=156072
On Bitcoin and Red Balloons
http://research.microsoft.com/apps/pubs/?id=156072
http://research.microsoft.com/pubs/156072/bitcoin.pdf
The article does raise a very serious point (oversight) in the bitcoin protocol -- suppose I put a high transaction fee on my transaction because I want it verified quickly....and the first node that I send to (and let's say the only node I can connect to) see that high fee, why would that node send the transaction to other nodes rather than spend a few weeks trying to include it in a block that it solves itself (and hence collects the fee)?
Receiving nodes have no reason to forward transactions on, and in fact, have an incentive to not forward paying transactions. That to me sounds like a big deal...you don't want each client to have to purposefully connect to thousands of nodes to broadcast transactions to get a decent confirmation time...that's not very p2p
Strange, I haven't heard of this problem playing out in practice.
addendum -- nothing compared to the block reward, that's for sure...
To be clear, this disincentive to not broadcast applies to mining nodes, right? Personally, I just run the Bitcoin qt client and have no reason to stop transactions from being broadcast.