On Sat, Jun 20, 2015 at 7:26 PM, David Vorick wrote: > I see it as unreasonable to expect all nodes to upgrade during a hardfork. > If you are intentionally waiting for that to happen, it's possible for an > extreme minority of nodes to hold the rest of the network hostage by simply > refusing to upgrade. However you want nodes to be able to protest until it > is clear that they have lost the battle without being at risk of getting > hardforked out of the network unexpectedly. > You can't observe the majority of nodes, only miners, and weighed by hashrate. If you need a mechanism for protest, that should happen before the hard fork change code is rolled out. I am assuming a completely uncontroversial change, in order to not confuse this discussion with the debate about what hard forks should be done. So I am not talking about protest, just about deploying a change. And yes, it is unreasonable to expect that every single node will upgrade. But there is a difference between ignoring old unmaintained nodes that do not influence anyone's behaviour, and ignoring the nodes that power miners producing actual blocks. In addition, having no blocks on the old chain is safer than producing a small number, as you want full nodes that have not noticed the fork to fail rather than see a slow but otherwise functional chain. -- Pieter