This correct except when it gets to a point where on-chain fees out-price the cost of running a node.
No one will be incentivized to run a node for a chain they can't use. That in turn weakens group rule validation.
I'm confident that with reasonable estimates for technological progress, a reasonable compromise can be reached.
I still don’t think the idea of a “compromise” makes sense. 
We've already compromised on a 4x increase with segwit and it didn't break the network.
It wasn’t a hard fork. I’m talking about a hard fork. And it wasn’t realistically a 4x, closer to a 2x when it comes to measuring in node costs (though that highly depends on what costs you measure).
Great Horse Manure Crisis of 1894
