Would it benefit both smaller traditional pools, and P2Pool, if the traditional pools mined for P2Pool?
P2Pool as a 10 second share time doesn't it? If this means a traditional pool would have to expose 10 second longpolls to its clients I don't think it'd work very well.
A new Bitcoin-P2Pool could be made for pools with a much larger share interval (2 minutes?). In fact, it would be possible to create a P2Pool-P2Pool that mines for shares on the new Bitcoin-P2Pool.
Down the road that might be a good idea. The reason is that as p2pool grows to maintain 10 sec share time will require more and more hashing power.
At 100 GH - 10 sec = 1 TH. So a 1 GH miner will get a share every 1000GH /1 GH/s = 1000 s = ~17 minutes
If the pool grew to 300 GH, it now takes the same 1 GH/s miner almost an hour to find one share.
At 1 TH it would take 1 GH/s miner almost 3 hours to find a share. Remember these are average. Share time of 300% won't be that rare and occasionally a miner will get even wider swings.
By building a higher level network (possibly w/ longer share interval) it could be used to connect MULTIPLE p2pools, small traditional pools, and even large hashing farms.
The p2pool client would connect to the "higher" network query it and find the available p2pools. Say p2pool A, B, C and have the new miner join the smallest. When a p2pool gets too large it could split. If a p2pool gets too small it could join another one.
That way you have a dynamic network of p2pools w/ upper bound of say 150GH/s per pool and a larger network consisting of the multiple pools, major hashing farms, and traditional pools.
