Just read that white paper another time, now I get a rough idea about how it might work in reality:
First two entities establish a common deposit by each sending certain amount of bitcoin into an address, similar to two banks each opening an account in counterpart bank and credit the counterpart same amount of money
Then all the transactions between these two entities will just change the ratio of each party's ownership of this common deposit. At the end of the clearing period, they settle the difference by a blockchain payment to make the ratio 50/50 again
This can help two large institutions, it is not very practical for single average user because the required deposit and one way payment nature. It seems single average user would still need to rely on large institutions, and existing large institutions might use lightning network or establish the clearing channel through other arrangements
I guess the most promising application of LN is where payments between entities are frequent and roughly predictable over the life of a channel. These might be BitPay<->Coinbase, but also it might be a man that routinely buys a cup of coffee somewhere in the morning, so setting up a channel between him and the cafe might make sense, especially because of near-instant 'confirmation'.
What you might be missing here is that all these channels can be chained together, forming a mesh-like network. In this case, you can have only one channel open, and send a payment to anyone in this network. This payment will be routed through intermediate hops right to the receiver.
The well-connected hops can be called
hubs, and will receive a fee for their service. In theory, anyone can become a hub, it's only limited by how much BTC you have, i.e. how many channels can you fund with them.
True, LN seems to be very similar to how banks and institutions work in a closed loop. However, in legacy financial system, one weak link on the chain might trigger a systematic failure like Lehman brother's case, because the whole system have very little real money in circulation. If LN chains are widely applied, it will also have such kind of risk, and without central bank bailout
Most of the retail transactions are very sporadic and unpredictable. If you routinely buys a cup of coffee somewhere in the morning, then it is very likely the shop will sell you some batch discount coupon that give you 10 coffee for the price of 9, and you pay the whole package at once, reducing the transaction fee. This is also observed in mobile fee charge: Previously telephone company charge you based on how much and how frequent you use the service, now they are using a bulk model to charge you regardless of usage, to dramatically reduce the amount of transactions
I guess there will be VISA-like mechanism if the clearing based settlement is widely used. Consumers will periodically (when they receive the salary) charge their web wallet in mobile that they can pay at any location that accepts bitcoin payment. And the real payment happens between the web wallet company and Bitpay. But unless we have many payment processors and credit issuer, this seems like a single point of failure
Use blockchain to do large deposit/withdraw, use web wallet to do casual spending, this could be the trend for the coming years
