I can be wrong, correct me if I do.
To use LN you must create payment channels. If I want to buy something on Amazon, I should open a payment channel with Amazon. But, if tomorrow I want to buy something on Ebay, I must also open a channel with them? And if later I want to buy a coffee at Starbucks, should I open another channel?
No such need, as long as you share a channel with a node (that has a channel shared with a node (that has... ...)))) with Starbucks. Same for Amazon. It's a mesh network. The general opinion is that there will be enough channels open to reach any relevant agent.
Ah you mean like Bitcoin? But then without the peer to peer? Sounds great.
Unless I'm mistaken
Right, not much need for on-chain stuff except opening/closing channels. Or possibly topping them up, I am still not clear on what will happen with "refill" operations as the protocol evolves. The hubs will be "central" only in the sense of "hyperconnected, with lots of money at stake in multiple channels". They won't be able to censor transactions or tamper with the protocol - or even figure out where the coins ultimately go. Hubs can be routed around if desired or necessary. Someone else will get the fees - if any. Not that shabby, is it?
Very tough question for me, how shabby? On the face of it sounds quite cool. But deep down I remain very suspicious of lightning network which stems from my distrust of banks. It feels like a banking layer being plonked down on top of bitcoin with the aim of bypassing hard money constraints that having transactions logged on a blockchain imposses. I have immense dislike of banking system which is what led me to bitcoin in the first place. "Be your own bank" At least having bitcoin as a settlement layer is a constraint of some form. It could be that the system if it does go that way will be a bit like an updated gold standard system whereas gold used to be the settlement layer now it will be bitcoin.