Lightning Network on BTC is a dead end even if it works as claimed

It has been expertly argued that Lightning Network (LN) does not work, and has security vulnerabilities and legal compliance issues (see, for example: Matthew Zietzke: Fundamental errors of BTC and Lightning Network; and Why Lightning will never be currency, and why BSV matters, by Craig S. Wright).

LN’s shortcoming and vulnerability are also hard to fix since they arise from fundamental properties of the off-chain payment network, such as:

1. LN payment network has no persistent ledger nor reliable records for accounting, and therefore unsuitable for businesses or any legitimate and non-casual payments.

2. Payments are executed off-chain but in a trustless manner (an inherent contradiction).

3. Expiration times are long in order to maintain the network capacity, creating large security risks, such as room for an attacker to maneuver.

4. The LN network is inherently unable to recognize where payments originate and where they are going, resulting in a disconnection from the chain.

5. The relationship between LN network and the underlying blockchain (BTC) is fundamentally discordant and corrupt. Others (including the miners) are incentivized to spy on the LN network for front-running the participants in LN network or causing premature closing of channels to their own advantages. With this, whatever consensus LN participants think they have is just a façade because ultimately they have to decide either to accept the risks or drop off the chain completely.

6. At the same time, LN transactions involve multi-parties and multi-hops (required for every simple two-party payment by design in order to artificially create a network effect which is otherwise nonexistent), thus creating obscurity and anonymity. Good miners are deprived of the ability of knowing the truth in the settlement, and end up not caring at all. Not their customer, not their concern.

7. The way LN network is configured, especially its requiring anonymous multi-parties and multi-hops transactions, leads to violations of various financial laws and regulations such as Bank Secrecy Act (BSA) administered by the Financial Crimes Enforcement Network (FinCEN).

Overall, LN is structured with misaligned interests and incentives of parties, and incorrectly delegated responsibilities, all characteristic of the BTC culture which has a skewed and social engineering agenda-driven understanding of economics (anarchism-leaning or communism-leaning depending on the context).

But what if LN works as claimed?

For the sake of argument, however, let’s say LN works, then what?

It would make a limited type of payments scalable (see further below). But that’s it.

We will discuss the several different aspects of this conclusion in detail below.

LN is not a blockchain solution

LN is promoted as a Layer 2 (L2) blockchain solution. But L2 solution as a concept in itself requires careful examination, because there are two types of L2 solutions that are fundamentally different. The first type is a legitimate second layer built on the same network with the base layer (L1), but the second type is defined and structured in such a way that it requires its own network and is no longer part of the same network with L1.

Lightning Network is the latter. Its name even tells that. And it is no longer Bitcoin blockchain.

There can be application layers built upon the base blockchain layer in the blockchain network, just like application layers can be built on top of the Internet TCP/IP layer. In this sense, L2 applications are not only permissible but in fact necessary. Even with a superbly scalable L1 blockchain such as Bitcoin SV blockchain (BSV), you may still need L2 solutions for certain type of applications.

However, to be a legitimate L2 blockchain application, everything that is meant to enjoy the benefits of being a part of the global network should be within the blockchain network protocol, just like everything on Internet is within the TCP/IP protocol. The Internet has multiple layers, but there is only one Internet, not multiple layers of ‘internets’. Overlay networks can be built for special purposes, but to be part of the same network the overlay networks must use encapsulation and virtualization in a way that the encapsulated packets and the virtualized network follow the overall network protocol.

For example, payment channels designed by Satoshi for Bitcoin and implemented on BSV blockchain is a legitimate L2 Bitcoin solution. A Bitcoin payment channel is not a network. It is just a channel between two or several related parties (peers). Within a payment channel, multiple payment transactions happen sequentially, each at an iteration step, and every step happens according to a Bitcoin script verifiable according to the Bitcoin protocol, except the verification of each interim step does not settle on-chain until the end of a coded nLocktime or the transaction’s nSequence has reached the maximum value (0xFFFFFFFF). The transaction can be updated many times between the peers until either, the nLocktime is reached, or nSequence is finalized and one of the parties transmits the transaction to the blockchain to be settled. No separate network is formed, and at no time during the channel sequence does the transaction fall out of Bitcoin protocol.

In contrast, Lightning Network is literally a separate network in itself, and the separate network has a type of structure and properties that absolutely have nothing to do with the Bitcoin blockchain network. And it has to be such, because otherwise it would not work for is claimed purpose (remember, LN has a completely nonscalable BTC network at L1 to solve).

Payment channels work for extremely high frequency payments. For all other payments that do not get to a high-frequency level between relevant parties for the parties to justify using a channel, the blockchain must be able to handle them directly without the help of channels.

But BTC can never do that at all. It is a fact. So as a result, BTC had to come up with a solution to at least ostensibly cover all kinds of payments, leaving only the high-value and low-frequency transactions to be settled on-chain. But once they’ve made that choice, creating a separate unreliable network and promoting multi-hops and multi-intermediaries becomes inevitable.

This is because this new network created by LN needs to cover all kinds of payments, not just high-frequency small payments between certain pairs of parties. Channels alone would not work for the type of payments that need to be covered by LN. This in turn is because a channel is set between a particular pair of parties having a specific payment relationship, and each channel works independently without the need of a network other than the Bitcoin network itself; but the BTC Core needed a network to create routes, liquidity and balances, hence necessitating the design of Lightning Network, which is fundamentally a separate non-blockchain network and a contradiction to the Bitcoin blockchain.

For BTC, the above result is inevitable and also self-inflicted because it is determined by its ideological choice for small blocks.

In contrast, Bitcoin payment channels are direct agreements between the relevant parties to handle high-frequency payments between those same parties only. The relationship between the parties in each channel arises from an actual business relationship that exists in the real world and is not artificially created by a network that is beyond control of the actual parties. Bitcoin channels do not rely on a separate network, as the blockchain itself provides sufficient network effect.

LN thus creates a separate network outside of the blockchain network, no longer a real L2, which means a layer within the same network rather than a separate network.

Further, LN is not an overlay network, which can be useful. An overlay network uses encapsulation and virtualization in a way that the encapsulated packets and the virtualized network follow the overall network protocol.

Even a separate network like LN may have its own use. But if the special circumstances require such a separate network, it is then what it is, and one needs to be honest and should not pretend it is still a blockchain solution. When both user/consumer perceptions of security, reliability, privacy, and the applications of law and regulations (especially securities law and AML) all depend on the meaning of terms such as blockchain and decentralization, semantics matters. A narrative based on fundamental misconceptions can amount to deception.

Furthermore, a whole different kind of problems arise when L1 and L2 implemented for solving real world problems are considered in aggregate (versus each application being considered separately). In aggregate, both L1 and L2 must be scalable, each in its own realm. This is the case even if the so-called L2 application has already made the sacrifice of not enjoying the benefit of the blockchain for all its transactions except for occasional settlements, as in the case of LN. As said, such applications are on a different network and should not even be called L2. But even if accepting this misnomer, it would still be a dead end without a scalable L1, because a blockchain that cannot scale in aggregate with all applications on it is not going to be adopted by the market at the end, regardless of whether a particular application needs such an L1 scale or not.

See below for more details.

The Layer 1 scalability

Because LN’s payment network does not have any settlement capability to start with, everything that needs settlement will need to be settled on L1, and this will require L1 itself to be sufficiently scalable as well in real use.

The core issue is that, to have full blockchain capability (micro payments, smart contracts, tokenization and decentralized data), even with a powerful L2 solution, you still must have a scalable Layer 1 (L1) blockchain to be able to run large numbers of transactions on-chain including settlements and verifications.

To become a solution for global transactions, being scalable means L1 TPS well beyond hundreds of thousands, if not millions, in the foreseeable future, and L2 TPS a further thousands of times of the L1 TPS.

To understand this, let us have a very fundamental look from the entire Internet viewpoint.

A 2011 paper on Science estimated the communication capacity of Internet was 3 x 1⁰¹² kb/s. See: The World’s Technological Capacity to Store, Communicate, and Compute Information (science.org).

Assuming that the Internet capacity has since doubled every 2 years, it’d be 1⁰¹⁴ now, and 3 x 1⁰¹⁵ kb/s in 10 yrs. Even if conservatively assuming that it has doubled every 3 years, it’d be 1⁰¹³ now, and 1⁰¹⁴ kb/s in 10 yrs. With the explosive growth of Internet of Things (IoT) which wasn’t considered by the above referenced paper, these numbers are only vast underestimates of the actual Internet capacity.

If just 1/100 of the Internet capacity needs to be on a value transaction layer (i.e., a part of the Internet of Value — IoV), the blockchain transaction capacity needed to support the IoV would be as follows:

1⁰¹² txs/s (which is 1 trillion TPS) now, and

3 x 1⁰¹³ txs/s (which is 30 trillion TPS) in 10 years,

assuming an average transaction size of 1 kb/tx (as a reference, the size of a simple Bitcoin transaction is 0.25 kb/tx).

Therefore, to serve as the value transaction layer on the future Internet, the blockchain network would need to have transaction capacity of TPS in trillions already. Even if only one thousandth of the blockchain transactions are on L1, it would mean L1 capacity of TPS in billions, and L2 TPS in trillions.

The above estimates assume the IoV to be 1/100 of the Internet.

Even if the actual blockchain IoV is only 1/100,000 of the Internet capacity, it would still require L1 capacity of TPS in millions, and L2 capacity in billions.

In more practical terms, it is highly conservative to say that a fully capable blockchain supporting Web3 and LoT would need to have L1 TPS in hundreds of thousands and L2 TPS in hundreds of millions in the near future and continue to grow without limit.

This is the kind of the scale for the future Web3 and IoV supported by blockchain. People who think that maximum TPS at a few thousands is going to be sufficient simply have no touch of what the scale of the reality is. It is one thing to be off by a few multiples, but entirely another to be off by thousands or even a million times.

The notion that one could push nearly all transactions to L2, hoping that an L1 blockchain limited to a TPS of 5–7 transactions could serve as the settlement layer for the whole world, is not even serious enough to be entertained.

BTC as an L1 blockchain will never be able to do that. Capped at 5–7 TPS by design, BTC is short by many orders of magnitude, so it is a dead end no matter what L2 solutions there may be.

The whole story of LN is reactionary in nature. It is reactionary to the fact that BTC does not scale. It also projects from a prefixed position that BTC is the only choice and must be made to work, and that an alternative is not allowed to be considered. It is an ad-lib patching device, not a principled invention.

The ‘court’ metaphor

LN does come under a very nice cover, however. The advocates of LN say that LN uses BTC blockchain as a last resort arbiter, like a ‘court’, which users do not resort to every time when having a transaction. It seems to make sense.

As a generic L2 logic, this is fine. But the whole ‘court’ analogy with respect to Bitcoin blockchain misses the fundamental difference between real life off-chain court and a blockchain:

In the off-chain world, the court is part of the human system that runs on human processes such as laws, lawyers, witnesses and evidentiary rules. For that purpose, the entire physical world is a supplier of the information and attestation needed in court processes. That’s how the traditional world settles disputes. It has its own physical ‘ledger’ so to speak. Backward it may be, but it has its own out-of-court ‘ledger’ to support the in-court proceedings.

But on a blockchain, the chain itself is the ledger for settlement. When there is a problem in L2 (LN), say a fraud, the chain has no ability to obtain attestation from L2 outside of L1. As a result, the settlement is either accepted or rejected, but either way it happens on-chain automatically, and is not going to solve any problem that arises from L2.

In other words, as far as L2 providing inputs to the L1 for settlement is concerned, once the chain-of-truth is broken, it is garbage in and garbage out.

Examples of use cases that LN cannot do

By design, LN delegates bulk of the truth-in-settlement away from L1 blockchain in attempting to scale it, but breaks the function of blockchain settlement in doing so.

Therefore, any application that requires truth in settlement simply cannot run with LN on BTC.

Esports: For example, Cryptofights couldn’t be built on BTC even if LN had worked perfectly the way it claims to. The very selling feature of Cryptofights is its verifiable moves each recorded on-chain. But even at an early beta stage, the daily transactions of Cryptofights has already reached over 500 K, which alone would completely break BTC’s maximum on-chain transaction processing capability. And this is only one application, and one that is just starting and has a potential to grow thousands of times with just one type of games in the future.

What can LN with BTC do for such an application? Nothing.

An LN solution would be running games off-chain in L2, and having only the final result of each game recorded on the L1 chain. This is still too much for BTC given its low TPS and high transaction costs, but even if we assume it is doable, it is a useless value proposition in the realm of gaming and esports, because the real issue is never in the recording of a final score, but cheating (real or perceived) by some players in their moves.

Tokenization: Tokenization solutions like Tokenized could not be built on BTC with LN.

Tokenized has an L2 structure because it is designed to accommodate the flexible business logic in the real world including multiparty collaboration and regulatory compliance, but has all messaging in every token transaction recorded on BSV chain. It does this not simply because it can (thanks to BSV’s scalability), but because it needs to. And being a scalable settlement L1 layer capable of unbounded number of verifications and unlimited level of ledger management, BSV is able to provide exactly what Tokenized needs.

Because Tokenized aims to serve the global market of all kinds of commercial records and transactions, it will not be surprising if the TPS of Tokenized alone reaches thousands. Again, there is nothing that LN with BTC can do for an application like Tokenized.

The same goes with most other projects and applications on BSV.

Comparison with the Internet architecture

BSV blockchain and network will eventually work just like the multilayered Internet, on which the transport/Internet protocol (TCP/IP) layer(s) only handle transporting and routing of information packets, while a vast amount of information processing (e.g., JavaScripts) is handled at the application layer, either locally or on a cloud.

The important thing one can learn from the Internet architecture is that, although the vast majority of processing is kept to the application layer, the TCP/IP layer itself must still be scalable, or there would be no Internet.

When applied to Bitcoin blockchain, the lesson learned is: even with powerful L2 solutions, the L1 blockchain itself must be scalable, or there would be no next generation Internet that has a global value network built in.

Scalable L1 plus unlimited L2 solutions is exactly how BSV blockchain and Metanet are designed.

In contrast, beyond certain limited payment applications, Lightning Network on BTC has nothing to go for, even if it works as claimed.

Inherent limitations of payment using LN

And the type of payments that may be suitable for LN is very limited. As said above, LN payment channels have no persistent ledger nor reliable records for accounting, and therefore are unsuitable for businesses or any payments that need persisting records.

In addition, the multiparty and multi-hop money transmission design and anonymity may make the system noncompliant with financial laws.

Therefore, only very casual payments or payments that intentionally keep anonymity knowing it is in violation of laws may find LN useful.

Lightning Network (LN) relies upon several assumptions, each requiring special circumstances to be valid.

These LN assumptions are:

#1: That the LN network itself is secure from attacks (note that the security of Bitcoin network has no contribution to this part of the LN network security, as they are separate);

#2: That payees and payers don’t care much about the security of individual payments, nor the accounting records at such individual payment levels, as long as there is a chance for a final settlement on-chain after a certain period of time;

#3: That the number of transactions per pair of payee and payer should be significantly greater than one in a certain period of time in order to make the costs and the sacrifices made of the L2 network worthwhile;

#4: That the flow of money needs to be reciprocal, go back and forth between payees and payers to maintain overall account balances off-chain.

Every above criterion in the above assumptions is unrealistic, at least lacks generality for scaling. Among them, #1 is not satisfied in any condition because it’s a general problem with LN; #2 and #3 are not satisfied in most peer-to-peer transactions (but may be partially satisfied with parties having an established business relationship); and #4 is not satisfied by the real-life payment activities in most commercially important areas.

Retail, for example, clearly doesn’t satisfy the criterion #4. It is the opposite. Money almost always goes from buyer to seller, and doesn’t flow back to buyers. As a result, with LN, buyers need to either frequently transfer money to LN account, or risk keeping lots of them in an unsafe account.

Every one of the above four assumptions poses a great problem. But because consumers don’t understand security, it’s possible that LN promoters will be able to fool the users by hiding these problems for a while. Not forever though.

But even if ignoring the above limitations and assuming the success of LN as claimed, LN is a solution to payment only (and a limited one for that matter even). LN cannot make BTC competitive in broader areas such as smart contract, tokenization, decentralization of data, as LN simply does not contribute anything to BTC in this respects, and BTC itself does not even pretend nor plan to develop such capabilities.

A future prospect

This does not mean that LN won’t have any temporary success before it sees a dead end. It may very well have that kind of short-living success, as the current scale of payments supported by LN is still too small to create a real test environment for a global blockchain (yet “big” enough to fool uneducated users), while its competitor (BSV)’s real capability still remains unrecognized.

This may last for a while, because in the crypto world, the mass follow influencers, and the influencers are the kind of people who become influential not because they understand the truth, but because they know how to take advantage of the mass who tend to believe in things that they can easily understand.

But when the real test and the real competition eventually show up (and they will soon), it will be way beyond what LN and BTC are capable of coping with.