ZenGround0
commented
6 years ago For reference there is existing work exploring something similar to the issue described in the second paragraph in the context of old fashioned PoW mining pools: http://webee.technion.ac.il/people/ittay/publications/btcPoolsSP15.pdf. The takeaway is that it is sometimes actually rational to not report mined blocks. Not sure if the analysis extends to the construction above or storage mining generally.
(I'm just writing down a thought I had today, not really expecting it to go anywhere)
In the current imagining of filecoin, every on-chain miner entity has an 'owner' to which block rewards are paid out. We could construct a profit-sharing pool to act as the owner for small miners, where the rewards for every miner winning a block in that pool are split amongst the participants. The nice thing about this is that it gives you the benefits of traditional mining pools, without the centralization of block production. Entry and Exit to and from the pool could be time gated, meaning that it takes N blocks to join, and N blocks to leave, to avoid peers leaving just before they might mine a block to gain the entire reward for themselves (and other potential attacks)
The one issue i havent worked through is that while miners in the pool must continue generating their PoSts, they don't necessarily have to ever mine a block, meaning they could just leech money from the pool (though, less money than if they actually mined) without actually contributing anything. This strategy is not rational, but it could be used to attack the pool in some way.