Published on: 2013-11-13T20:27:52+00:00
In a communication between Brian Goss and Peter Todd, the topic of propagation time within a mining pool is discussed. Brian questions whether the propagation time from the pool to the miner affects the analysis of Q for a mining pool composed of ASICs connected by 300 baud modems.Peter explains that even now, the propagation time from the pool to the miner is significant, especially for pools that do not pay for stale shares. He also mentions the growing importance of selfish mining and orphans, which could result in the centralization of hashing power's physical location. If there is a 100ms delay to a pool, miners may choose to locate their mining equipment closer to the pool to avoid financial losses. This could potentially lead to pools wanting to be located near each other.Peter suggests that if all Bitcoin mining were concentrated in one place, like Iceland, it would have negative consequences. Addressing Brian's concerns, he assures him that his questions are not ignorant.In an email exchange on the Bitcoin development mailing list from November 2013, Peter Todd responds to a question regarding propagation time within a mining pool. He explains that the likelihood of fork occurrences depends on the amount of hashing power a miner possesses. Peter outlines three potential outcomes: the block propagating without any other miner finding a block, another miner finding a block during propagation resulting in a tie that the original miner either wins or loses, or another miner finding a block before propagation is complete.To further illustrate his point, Peter introduces variables such as t_prop (the time it takes for a block to propagate from a miner to 100% of the hashing power), t_int (the average interval between blocks), and Q (the probability that a miner will find the next block). Using these variables, Peter calculates probabilities for each outcome while considering Q. He defines P_fork(Q) as (t_prop/t_int * (1-Q))(1-Q) = t_prop/t_int * (1-Q)^2.Additionally, Peter recalculates the break-even fee/KB using Q and discovers that larger pools have a significant advantage as they can charge lower fees for transactions and earn more money. He emphasizes that this issue is inherent to Bitcoin's design and regardless of block size or network speed, the current consensus protocol rewards larger mining pools with lower costs per KB for including transactions.Peter argues that an unlimited block size would exacerbate the problem by increasing fixed costs. However, maintaining the block size at 1MB indefinitely does not solve the underlying issue either. He warns that the perverse incentives to publish blocks only to a minority of hashing power would be detrimental to decentralization.
Updated on: 2023-08-01T06:31:31.994858+00:00