I'm liking
JR's paper. Micropayment channels is the most constructive approach to permanently dealing with the block size limit, since defenders of the limit are mostly concerned about the data overhead on full nodes.
Maybe if the limit was higher. Payment channels still require at least one transaction per payment channel set. How often are you willing to reset payment channels? Once per week, once per month, once per year? The longer the period the larger the initial escrowed amount. How many payment channels if the average user going to have. Say the average user accepts escrowing 6 months worth of future payments and has 3 payment channels payees each. That still requires 6 on block transactions per user per year. 2 tps @ 6 txn per user = 10.5 million users.
Do you really think Bitcoin will be a global revolution with an upper limit @ 10 million users?
Payment channels, side chains, and cross chain atomic transactions are transaction multipliers. They move us from one economic transaction per block transaction to more than one economic transaction per block transaction. In essence they make the primary block space more efficient but you can't get any kind of scalability out of 1MB permanent limit.
I was talking about payment channels between nodes so they can negotiate peering agreements, not for replacing regular user transactions.