Thread by @DoctorOrrery

A lot of people claiming the “lightning network” will save #bitcoin are referring to two completely different ideas and networks, one decentralised, the other centralised. Neither one works, but for very different reasons. (1/17) 🧵

First, the decentralised version. This is the original layer 2, first proposed in 2015. It’s meant to solve the fact that bitcoin’s blockchain is slow, expensive and hilariously limited. It has many problems but the two most obvious are pre-funding and spanning (2/17)

To use LN you must prefund a channel between you and the counterparty. Eventually this will settle on blockchain, but you want to wait as long as possible to avoid the blockchain fees, so maybe you prefund a month’s worth of transactions. (3/17)

If you want to do this for multiple businesses it becomes expensive quickly. Say the cafe, ($100), local eatery ($200), supermarket ($500). This is the opposite of a credit card, you shell out BEFORE spending, but there are still fees (4/17)

So you don’t want to open too many channels. That’s fine, you can rely on the network to pay people you don’t have a direct channel with. But that’s problem number 2, spanning. The system has to find the shortest, cheapest link between you and your recipient (5/17)

This is a solved problem in graph theory. One solution is Dijkstra’s algorithm. It says a network with N nodes and C connections can be solved in order(C + N log N). Order refers to the amount of computation and therefore time required. (6/17)

Trouble is, with a global network these numbers get very large, very quickly. Even at just 100k users, N log N is getting big. Your phone is unlikely to be able to find the shortest route in anything like real time. (The longer the route, the more expensive it is). (7/17)

Add to that the fact that many of these nodes are phone apps or computers that are regularly being turned on or off and the network topology changes MUCH faster than it can be spanned. The shortest route, often any route, just can’t be found! (8/17)

There is an easy solution: Centralise. Large hubs that do all the intermediary work, but then you’ve just invented a worse banking system. The other centralised option is the many companies that claim to use Lightning, like Strike. (9/17)

Strike is a payments system that claims to be built on the Lightning Network. However, it has no channels to the actual Lightning network. You interact with it through the app, where it dutifully shows you buying and transferring Bitcoin. (10/17)

Strike claims its backend is Lightning but it might just as well be an SQL database. Nothing happens on chain, you can’t interact with anyone else on the network. If you take Bitcoin OUT it stops being free, they charge you the on-chain fees. (11/17)

So Strike is a traditional money transfer service cosplaying as a cryptocurrency company to get VC money. But they still allow instantaneous near free transfer, so that’s an improvement, right? No, it’s an illusion. (12/17)

Money transfer rarely nets or is symmetrical. More USD goes to Mexican Peso through Western Union than the other way round. FX is highly volatile, so companies want to transfer back ASAP. This is where the cost comes in. (13/17)

Strikes CEO Jack Mallers belled the cat in a Jan 2021 Medium article. He identified the issue then said Bitcoin solved it, but nothing in his article said how. If more EUR is going to USD than vice versa they have trade it back or end up with a massive FX position. (14/17)

That trade needs to be done in the FX market, paying spread like everyone else. (As an aside, they also don’t do the kind of anti money laundering everyone else does, which is faster and cheaper for them now but likely to bite them eventually). (15/17)

How do they make it so cheap? Best case is they’re loss leading with VC money. Worst case would involve massive fraud. Their users’ money and bitcoin are in Strike’s custody and it wouldn’t be the first crypto rug pull. (16/17)

But the most important thing about Strike when it comes to crypto discussions is STRIKE DOESN’T USE BITCOIN. The payments, transfers, etc, never touch the blockchain. To the extent it works it’s BECAUSE it doesn’t use blockchain. (17/17)

In summary, the decentralised Lightning network can’t scale, and centralised Lightning networks don’t use any of the elements of crypto. Neither one is going to help make Bitcoin mainstream. End 🧵

A postscript to this, as pointed out by @BennettTomlin. As mentioned above, prefunding means you probably don’t want to store more than around a months payments. To open an close a channel requires an on-chain Bitcoin transaction.

@BennettTomlin However, Bitcoin can only handle 450k transactions a day. Which means, if it did nothing else, the bitcoin blockchain can support around 13.5M LN channel open/closes a month. If the channels are cycled monthly, that’s <7M channels, total.

If each person/business has a couple of channels, as the LN envisages, you can support a mid sized city IF no other bitcoin transactions are conducted, globally. The Lightning Network really, really, REALLY doesn’t scale!