Dave Friberg good question.
LN is a good idea, if not pushed down the throat of people and used sensitively.
It is a very bad idea is applied to scale the number of transactions in the network.
LN allow two persons to lock some funds (say 100$) each on a Bitcoin address with a smart contract. Then they can transact as much as they want go and forth, until one of them get to 200$ and the other get to zero.
Then it stops.
At time T0 both have 100$ locked in. AT time T1 maybe Adam has 50$ and Bob 150$; at time T2 Adam has 0$ and Bob 200$.. In time T3 Adam has 200$ and Bob has 0$.
Now, suppose at T3 Bob has 0$. He cheats and publish the state of the channel when he had 200$. Now Bob has some time to see it and, in turn, publish the last state of the channel (when he has 200$).
If Bob can publish his transaction in time he get what is his; if not he lose the money.
Problem: blocks are full and transaction’s time to confirm are a wild guess.
Vays talks about fees of 100$ for Bitcoin.
If Bitcon get there, how much would cost me to make a transaction to open a channel (100$), make a transaction to close the channel (100$). How much should I lock in to justify this type of fee? 10K or more? Who has that type of funds to lock in just to trade.
Normal people have mainly unidirectional flow of money (from your employer / customers to you and from you to your shops / suppliers.) We rarely have given to and from the same sources.
How many LN channels do you think a normal person could open? I say one or none. If any user opens just a channel, we have just a big bank-like hub where all transactions move thru. And it can steal our funds, as its transactions will always be able to pay more than ours.
Even if we had blocks large enough and cheap transaction to open many hundreds of LN channels, big hubs are probably needed anyway. There is no working solution for the routing of payment thru LN now. And “We will get one” is not a good answer to the problem.