For anyone interested, there’s a dedicate thread for changes to the subnet affected by this proposal → Subnet Management - io67a (Application) - Developers - Internet Computer Developer Forum (dfinity.org)
@Lerak also provided his network latency analysis there. Great job @Lerak!
Note that the consensus protocol is designed to be able to cope with network delays that occasionally fall behind the 0-rank block notarization delay.
I am curious as to why this parameter is a fixed value rather than adaptive (adaptive to network conditions that ebb and flow - incrementing when network conditions are bad, and decrementing when better performance can be achieved). An implementation like this would have avoided the need to update the config (the notarization delay would have gradually reduced to an equilibrium point that optimises for throughput) - just thinking out loud.