Hash rate is not the only part of the Red Queen game. A miner is incentivized to improve his bandwidth, and number of connections. It is in a miner's best interests to send his block to as many other mining nodes as possible, as quickly as possible to reduce orphan risk (upload speed). It is also in a miner's best interests to receive and validate blocks from other miner's as quickly as possible, to reduce orphan risk (download speed). Connectivity is part of the Red Queen game, and the network will improve through rational self interest.
Consider the case with checkpoints. So far, the argument has been focused around the potential dangers from various attack vectors. One attack described a network segmentation attack in which an attacker maintains a hidden lead, and releases blocks publicly as though they are merely resolving a natural orphan race. At the 10th block, they attempt to cause a "tie" in which a significant proportion of nodes differ on which 10th block is the "first seen". This could splinter the network, and could be repeated until the total hash power is scattered into a thousand pieces.
I have seen it argued that this is merely FUD, and that the network connectivity would have to be severely degraded for such an attack to be possible. Some have even argued that the network would have to be degraded for hours, which seemingly reduces the probability of success for such an attack.
There are two problems with these arguments. First, if the blocks are released in lock-step with the so called "honest" blocks, the network degradation window needn't be hours, but rather minutes.
Second, and more importantly, the unprecedented nature of the kinds of network outages required for the success of such an attack cannot be relied upon with any credibility in forecasting the future likelihood of such events. The checkpoint "protection" both relaxes incentives for miners to maintain the strongest possible connectivity, and simultaneously incentivizes network disruption.
This externalization (or socialization) of network risks is not strictly a matter of centralization. It is a matter of incentives, just as the rest of Bitcoin. Also, the disruption of a mining network may involve disruption of unrelated infrastructure. This could be a foot in the door for regulating bodies to intervene with favor from the general public. We must be vigilant to distinguish any resulting consequences of Nakamoto Concensus and it's reckless circumvention.
Edit: spaced out paragraphs to make it less of a wall of text.
No comments:
Post a Comment