To get some interest going. How L4S can help...

To the left is your regular network queue when you load it with data. The exact nature of it depends on how it is managed, but the current AQMs will let the queue grow to a certain size (larger than when using L4S), and they drop packets to control the sender's rate.
When LLD is enabled, you get an additional queue, where any L4S traffic goes through. L4S sees the bottleneck much sooner and controls the sender's rate more gradually. It only drops packets as a last resort. The endpoints must also support L4S.
There is a L4S discussion club here: https://mailarchive.ietf.org/arch/browse/l4s-discuss/ Livingood also comes by now and then.