A fairly easy to read paper relative to the original Lightning (ie ln penalty) paper.

Important to note that it requires BIP 118 aka the SIGHASH_ANYPREVOUT (or some equivalent) soft fork to Bitcoin core. But it seems widely agreed that this is the direction Lightning will go in when this sighash is available.

There are a number of benefits to the protocol upgrade. Most notably easier to manage channel state - both participants have symmetric state and it’s as simple as storing the latest update transaction. If you get behind and broadcast stale state there isn’t a penalty - your channel partner just broadcasts their latest state with your stale state as an input.