[fruitworks]

I think it would require something like 200MB blocks, which is within the realm of possibility.

The only thing stopping you from adding payment channels on top of that is transaction malleability, but I am not really caught up on the topic.

[ifwinterco]

That still means the chain is growing by 28GB every single day, so 10TB a year.

That's arguably past the point where running a small node is viable, so I would argue you're well into the territory of losing some of the decentralisation properties you want in a cryptocurrency

[fruitworks]

10TB/yr is like $100-200/yr of hard drives and this price will continue to decrease. Compare this to the transaction fees for a small buisness accepting bitcoin and you will find that it is reasonable. Especially when you consider that reasonably efficient miners start at thousands of dollars.

small nodes don't matter. They can prune or shard the blockchain.

What matters is the economics of medium-sized nodes that are operated by small buisnesses. These are are the entities that have material reasons to run a full node (to accept transactions in an automated manner while preventing theft), and these are the entities that evaluate the rules of the cryptocurrency at the time of transaction.