You can't scale technologies in layers.

🧵 👇

You *can* build applications and reusability frameworks in layers.

And no, the internet does not *scale* in layers either. I'll explain below.

There is an absurdly common misconception that technologies like Bitcoin can scale in layers. When responding to criticism of Bitcoin's lack of sufficient throughput (4-7 tps) and extremely high latency for full confirmation (30-60 minutes), proponents will often claim that "second layer" systems like the Lightning Network magically solve this problem.

This just simply isn't true.

*Lightning is not Bitcoin*, and is not attempting to scale bitcoin itself. It is an entirely separate application that leverages Bitcoin as a *feature* for settlement.

Unfortunately, as a separate application, it also has an entire transactional model of its own, completely separate to Bitcoin's. One that is far more complex and prone to bugs, faults, and vulnerabilities, is far less decentralized, and tends to be very custodial.

Lightning is indeed relatively scalable, but it makes tremendous sacrifices in complexity, security, and decentralization in order to achieve that scalability.

Unless running your own node, you're using LN custodially to some extent, whether a fully custodial wallet or at a minimum custodial channel management. This is completely antithetical to everything Bitcoin represents.

Graph theory tells us that the complexity of searching for paths between two vertices in a weighted directional graph means that Lightning simply cannot scale without dependency on the formation of massive centralized liquidity hubs, acting as middle-men collecting more fees from the users.

This is essentially a reincarnation of traditional fiat banking/credit/payment applications, only built on top of Bitcoin rather than fiat. Yes, Bitcoin provides a much more sound base layer, but the "transactional" Lightning layer is just as compromised as in current fiat monetary systems.

We can do better. We can keep the transactional layer on the base layer itself, with significantly improved decentralization, security, throughput, latency, and usability.

I also frequently see Bitcoin supporters pointing to the internet as an example of a system scaling via layers. This is very inaccurate. The internet does not scale in layers.

Each layer of the internet protocol suite is an application or reusability framework layer, and is as optimized as theoretically possible for its specific purpose. And all of them fully rely on the underlying layers for all relevant activity.

None of them are attempting to scale their underlying layers.

The HTTP protocol does not implement its own solution for data transfer. It relies fully on protocols like TCP, UDP, QUIC, etc for data transfer.

This is not true about Lightning. As I said before, Lightning is a separate application entirely, using its own protocol for transactions and other operations, circumventing Bitcoin's transactional security entirely and using it only as a feature for settlement.

Many Bitcoin advocates have seemingly been convinced that it's impossible for a base layer to provide adequate scalability (throughput and latency) without sacrificing decentralization or security.

This problem is often referred to as the "blockchain trilemma". What people tend to misunderstand about this trilemma is that it only applies to parameterizations of a particular system architecture.

If you take Bitcoin, and make its blocks bigger, or its block time shorter, you can increase scalability significantly. But in doing so, you're likely making sacrifices to decentralization and security. This much is true.

But by completely re-architecting the solution from the ground up, we can and have achieved significant improvement across all areas of concern, including scalability, decentralization, security, and usability.