Bill got me thinking about how lightning fees could be priced in order to allow micropayments.

At first I thought, tiny transactions like 1-10 sats could be free, but then fees would kick in for larger payments. Then I realized that would just result in 1000 transactions of 10 sats each to pay 10K sats. That sounds terrible.

So then lets flip it, maybe? What about cheap fees for big amounts and higher fees for tiny transactions? Well, big transactions are going to make liquidity management more difficult, so the node operators have an incentive to not do that. Plus it kinda throws microtransactions under the bus.

What currently makes the most sense to me is a flat fee. Maybe something like 30 sats. It means that sending a fraction of a penny (USD) wouldn't make a lot of sense because the fees would be the majority of the cost. This would encourage more meaningful sized zaps. For things like reading an article, paying them a few cents seems reasonable too. Plus it's easy to understand (less so after other nodes' routing fees are added in, but our node would be doing it's part).

Large transactions can still mess up liquidity, but I don't think we are going to solve that problem with fees. That seems like it'll only be improved by increasing lightning adoption, which can only happen if node runners understand enough about the technical workings to be confident that they can still get their on-chain funds back if something goes terribly wrong with their node (e.g. keys get lost).

I still haven't seen an explanation of how lightning works in practice to have this confidence. So I wouldn't put more liquidity into lightning than I am willing to lose.