The Nb/Nc DAA doesn’t have any problem with differences in latency topology. Even if only 5% of miners have up to 20x higher latency, and if the DAA looks back only 7 blocks (just as fast as parent method), they throw off everyone’s ability to form a “consensus cohort” . It automatically sees they are referencing a lot of grandparents and great grandparents. This keeps the difficulty high and it stays close to the ideal Nb/Nc = 2.42 ratio (Parents = 1.44/block). At 5%, the grandparent method isn’t any better than the parent method. The difficulty doesn’t rise as high in this chart compared to the others because it’s only 5% of the hashrate .
