LN doesn't really work either - it can't really guarantee that funds have been transferred until you close a channel (you commit the transaction on the Bitcoin chain). So, unless you trust the other end of the channel not to fraudulently close the channel, you can't actually consider the transaction settled. And of course, in the case of routes, there has to basically be end-to-end trust for this to work, or someone will be left holding the bag.

It also can't scale the way it's usually promised, in part due to Bitcoin's horribly low transaction rate (you can't open LN channels fast enough, given Bitcoin's transaction rate), in part because of limits on the number of transactions in a live channel.

And in practice, it seems it's highly centralized [0].

0: https://iopscience.iop.org/article/10.1088/1367-2630/aba062

No, that's not how LN works. If the other party suddenly closes the channel at an old state that is beneficial to them, the funds will get locked for 2 weeks and you will have enough time to publish a transaction with a more recent state, which will reward you with whole balance of the channel [0].

[0] https://blog.bitmex.com/lightning-network-justice/

As that article explains, that only works if the non-closing node notices the problem in time to contest this. And, according to your article, this period is 24h, not 2 weeks.

I'd also note that this seems to make DoS much worse of a problem. If I have a transaction with a node that has gone offline and can't be reached, and I initiate a forced closure, what's to stop the other node from taking all the money in the channel, especially if I am not constantly online and the channel state I can't know for sure if the other node may have made some transactions while I was away?