the box on page 5 defines DefaultCompliant as "The default Bitcoin mining strategy, including all available transactions, mining on the end of the longest chain, choosing the older block in a tie, and publishing all blocks."
the core code does not choose the older block in a tie, the code chooses the block with the largest legitimate chainwork (under current TARGET epoch)
Second note: this paper investigates the era in the future when there are no mining rewards, only transaction fees. How far away is that ? How valuable is one satoshi now? Much of what is said is relative to that far-away case
(reading through the paper) This is an interesting (and thorough) thought experiment for what behavior might emerge when only transaction fees are the reward for mining. However the case the paper makes for a serious security problem gets weaker, as the argument depends on a growing number of assumptions as the paper goes on..
I fail to understand why a PETTY-COMPLIANT miner would ever realistically take a set of Tx that does not short-term maximize profit, given the competition for new blocks goes WAY up as the value increases. In other words, the undercutting and LAZY-FORK behavior would be crowded out right away, as it is insufficiently popular.
In the AGGRESSIVE-UNDERCUTTING discussion, it assumes a lot of "forks" or alternate chain tips, to chose from, is this true in practice? Are there really that many chain tips to chose from, to make this practice even a consideration?
I see in page 10 that there is a more refined and detailed descriptions on miner strategy. From a math perspective this is interesting, but the comments above stand.. would anyone even have a chance to move ONE BLOCK using these considerations, given the value and competition for every single block?