Lightning Network: Can it be fixed?

I see scaling as a huge issue we need to solve to enable global adoption.

I’ve been wondering why LN is being seen as a failure now - it had so much promise (tiny fees, instant(ish) settlement.

From what I see, the criticisms are that:

1) It’s not generally self-custodial 2) It frequently encounters errors 3) Requires recipient to create invoices to initiate a transaction. 4) Others?

Self Custody Whilst self custody is a great option, I don’t see it as workable for the majority of people who rely on banking wrap-around services to help support them. As fees on the main chain increase with any global adoption, it will be untenable for “normal people” to use it (assuming todays tech). We will end up with banking-like services on a layer 2 or some future technology inside the main chain. I think what matters most is the ability to verify funds inside some future banking-service to prevent fractional reserves / fraud being enacted.

LN errors I haven’t used it - I don’t spend my BTC. Can these errors be eradicated or does the architecture mean they can’t be overcome?

Recipient Action I think it might be possible to have a direct-debit type service where automated processes interact to send->receive->pay.

Other issues? I’m interested to hear about the reasons why LN won’t work.