>doesn't using your own UNL basically equate to a hard fork in the event that the nodes don't agree? Short answer: No, if one or a few dont agree they are just

>doesn't using your own UNL basically equate to a hard fork in the event that the nodes don't agree? Short answer: No, if one or a few dont agree they are just ignored. If the majority (>50%) does then it halts and if a quorum is reached (>80%) it could fork. long answer: Its complicated. Under normal conditions nothing happens but in edge cases (large node outages/physical disconnection/like if and undersea cable is cut) it could fragment. This can be avoided if the core of the nodes mostly has each other on the UNL. But since we cant and most importantly should not enforce that its encourages to create and publish an UNL if you think the existing UNLs are not good. Its also encourages to use an UNL with large overlap to other UNLs This leads to the assumption that almost everyone uses one of the published UNLs with minor changes (mostly additions) else if you had mayor changes you would publish an UNL to convince other to also drop/add specific nodes (after all if you have a reason you probably want to tell everyone) Since the UNLs overlap the real nodes UNLs should too and forking is not possible with large overlap. On a high overlap it could only halt if too many nodes are separated/offline which is the wanted behavior rather than making uncertain progress. There is a amendment coming (most likely) for negative UNL where nodes can signal that they lost contact to a node and temporary remove it form the UNL. This would reduce the risk of halting if large percentage of nodes are temporary lost. Instead of 20% it would tolerate more nodes to go offline before the network would halt. It basically adjust the 80% quorum requirement down with every node that goes offline but has a minimum of 60% of the original 100%. 60% is the minimum to prevent forking/fragmenting again. only 51% would be needed if the UNLs would had perfect 100% overlap, so its a bit higher to compensate for the fact that UNLs dont have 100% overlap. https://xrpl.org/negative-unl.html https://xrpl.org/negative-unl.html