Post
Topic
Board Development & Technical Discussion
Re: Elastic block cap with rollover penalties
by
Meni Rosenfeld
on 11/06/2015, 07:44:29 UTC
Because if he starts creating 5943-tx blocks, the penalty pool will be smaller, and so will the amount he collects per block. See scenario 1 here - if the big miner behaves as the small miners do, it's the same as if they're all small miners. He will get just 4.50 BTC per block, instead of 4.68. He can't have the cake and eat it too - not pay a penalty, but expect to reclaim it...
That's not the point. What prevents the competition from pulling from the pool without contributing to it? I understand that it's in the interest of a massive node or cartel of nodes to mine large blocks, but only if they aren't undermined by their competitors, which can pull from the pool without paying any penalty themselves.

The nodes can't have their cake and eat it too... But they can eat it. Ultimately, there would be no pool.
That's exactly what the small miners are doing. In my scenario 2, the small miners do not contribute and are not expected to. They just maximize their own (tx fees - penalty) at around 6000 txs, and enjoy the collection from the penalties the big miner is paying. They eat the cake and let the big miner bake it. The big miner creates supersized blocks knowing full well that the small miners will not follow in their lead. It's in the interest of a massive node to mine large blocks regardless of what the competition are doing. The optimization criterion is different for small and big miners. The 90% miner knows that he'll be able to reclaim some of his own penalty (not the penalty of the other miners), so he effectively gets (tx fees - 0.1*penalty) per block, which is optimized at a higher size. If there are several big miners, each of them creates supersized blocks because it is in their selfish interest. If miner 2 pulls out, miner 1 will continue creating supersized blocks. But in fact, miner 2 will not pull out, because it would just reduce his own profits.

As I mentioned, the above is as long as we ignore difficulty changes. In practice, what will happen is that the supersized blocks will attract more miners, which will increase the difficulty and reduce the profits. A big miner which takes this into account will probably want to avoid it, and will not create supersized blocks. If everyone (including the big miners) just maximize (tx fees - penalty) without regard to reclaiming, then all miners get the same reward per block. This, again, proves my original point, that the method does not create an additional incentive to centralize.


Also, numbers speak louder than words. In scenario 2 here, I've claimed that the optimal strategy for the small miners is 5943 txs/block, and the optimal strategy for the big miner is 7251 txs/perblock. This is a Nash equilibrium - each player has no better option, assuming everyone else sticks to their strategies. Are you saying this is not so? Can you suggest a different strategy for the big miner, which will result in getting more than 4.68521 BTC / block (without the small miners changing theirs)? Can you suggest a different strategy for a small miner (just one, not all of them as a collective) which will result in getting more than 6.8552 BTC / block? Again, we're talking about time scales long enough to reclaim penalties, but short enough for difficulty targeting and new miners to not be a factor. Or perhaps, you'd like to investigate a scenario with a different distribution of miners?

If there are several big miners, they should all create supersized blocks, and then every miner will enjoy the supersized blocks of the others.
Yes, they should, and this would be a cartel; where each is cooperating with each-other for greater profits, and it would break up in the same fashion as other cartels; Any member of the cartel profits more by not contributing to the penalty pool.
No. They are not cooperating. Each miner is simply doing what is best for him, expecting others to be similarly selfish. This is shown in my calculations, and if you'd like, I can repeat the calculations for a different distribution of miners.

Again - a big miner (regardless of whether the other miners are big, small, whatever) creates supersized blocks (blocks which are bigger than what optimizes tx fees - penalty) simply because he expects to reclaim some of his own penalty. Not the penalty of others.