There is no such thing as a trustless L2
Everything you need to know to cut through the L2 marketing grift.
As I’ve written before, Bitcoin will never be cheap to use. As that becomes more obvious more people are pinning their hopes on second-layer networks to keep their dream of cheap universal decentralization alive.
Bitcoin Layer 2s are important and valuable but they are also wildly over-marketed. Don’t buy the hype. There is no such thing as a trustless L2.
Inside this issue:
What is a Bitcoin Layer 2?
Why would you want an L2?
There are only three kinds of Bitcoin L2
How to evaluate an L2 (and why you don’t need to)
What is a Bitcoin Layer 2?
Loosely, a Bitcoin Layer 2 (L2) is a secondary network or protocol built on top of the Bitcoin network. To use an L2 you send some of your bitcoin to an address controlled by the L2, so you have to trust the L2 to keep your bitcoin safe and eventually return them to you. In exchange for that trust you can do things on the L2 that you couldn’t do using the base (or L1) Bitcoin network, like move coins instantly for free or use them as collateral in a smart contract. Using an L2 requires additional trust, but that additional trust allows for greater flexibility.
I say "loosely" because Layer 2 / L2 is a cultural definition, not a technical one. When people argue about the proper definition of L2s what they are really arguing about is what kinds of trust models they find acceptable. You could argue, for example, that Coinbase is functionally an L2: you deposit Bitcoin, use it to trade (or whatever) and then later withdraw.
Custodial systems are not typically what people imagine when they think of L2s but whether we think of them as 'valid' L2s or not they are empirically the largest and most widely used second layer systems built on Bitcoin. Bitcoin ETFs for example hold ~4% of all bitcoin. Another ~1% of all bitcoin are held by wBTC on Ethereum (~160k bitcoin) and BTCB on Binance Smart Chain (~60k bitcoin) — both fully centralized custodians that essentially offer bitcoin escrow services to the Ethereum and BSC economies respectively. By comparison, only ~0.02% (~4.5k bitcoin) are held on the Lightning Network.
A lot of Bitcoiners bristle at calling centralized custodians L2s because you have to trust the custodian. In my opinion they are right to be skeptical of custodians but wrong to turn that skepticism off for other, more complicated L2s. To be useful an L2 needs to be able to do something that the base layer of Bitcoin can’t already do and to do something that Bitcoin can’t do it needs to make new promises that Bitcoin doesn’t already make. New promises mean new categories of trust. People often call something a Bitcoin L2 as a shorthand for 'trustless' when what they really mean is that they trust it. There is no such thing as a trustless Layer 2!
The right mental model for depositing your bitcoin into anything (whether you consider it a Layer 2 or not) is to think of it as buying chips in a casino. You need chips to play the games, but the chips are worthless outside the casino. When you trade bitcoin to buy casino chips, you are trusting the casino not to lose your money, or steal it, or rig the games or abruptly shut down and disappear. Bitcoin is very valuable, so trusting anything with your bitcoin is a very big risk.
Why would you want an L2?
The simplest reason to use an L2 is to save on Bitcoin transaction fees, which are expensive by design. The Lightning Network, for example, lets users set up payment channels where they can settle transactions offline without needing to pay for an expensive update to the main chain. Similarly, Coinbase users can send each other bitcoin instantly for free because Coinbase doesn’t need to pay for a Bitcoin transaction to update its internal account database.
Another potential reason to use an L2 is privacy. Transactions on the base Bitcoin network aren’t just expensive and slow — they are also public forever. Transactions that have been moved off-chain to an L2 don’t leave behind a permanent footprint. Payments within a private Lightning Network channel for example are only visible to the owners of that channel — or you might want to use Blockstream’s Liquid Network because it supports confidential transactions.1
You can also use custodial services for privacy — such as depositing bitcoin into Coinbase from one address and withdrawing to another instead of making a payment directly. Coinbase itself would know those transactions are related — but blockchain analysis firms will have no way of connecting those two addresses. Whether that is a meaningful improvement in privacy will depend on your threat model. It’s not a good way to hide from a government interested in prosecuting you, but it is a pretty effective way to hide from most prying eyes.
The last basic reason to use an L2 to do things with your bitcoin that can’t be done directly on the L1 Bitcoin network. Bitcoin ETFs, for example, let you hold bitcoin in a traditional brokerage account, which is helpful if you want to use bitcoin as a tool for diversifying your portfolio or as part of the collateral for a mortgage. Or the Liquid Network lets you use bitcoin as collateral in smart contracts, which is useful for creating derivative hedge trades or borrowing against your bitcoin anonymously.
There are only three kinds of Bitcoin L2
Custodial L2s (“Trust me”)
Custodial services like Coinbase have a single centralized operator and you are completely vulnerable to them — they can take (or lose) all your money and there is pretty much nothing you can do about it except sue. Custodial L2s are cheap, fast and flexible — but they require trust in the custodian. They are the most convenient and cheap Bitcoin L2s, which means they are also the largest and most liquid.
More importantly, custodial L2s are important because many security models functionally are custodial even if they are advertised as being something else. If the developers of a project have any "emergency powers" like the ability to roll back state in a crisis or unilaterally update the withdrawal rules of the network — the system is custodial. Security guarantees that can be disabled in an emergency are like seatbelts that release if the car decelerates too quickly.
Federated L2s (“Trust us”)
Some Bitcoin L2s are operated by a federation — a committee of independent stakeholders that make decisions by M of N majority vote. In its simplest form a federated L2 is just a multi-signature address: members of the federation "vote" on the legitimacy of a withdrawal transaction by signing it. In more complicated systems (like Fedimint or Liquid) the federation can work together to run an entire parallel network with its own rules and features.
The advantage of a federation over a custodian is that no single member of a federation can betray you — a majority of network operators have to cooperate to change the network rules. That means to trust a federated network you only need a diffuse trust that the majority of network operators are honest, rather than the specific, focused trust you need in the custodian of a custodial network. Federated L2s can do all the same things that custodial L2s can do with potentially less trust.
In the ideal case a federation can be as almost straightforward and efficient as a custodial L2 while being almost as reliable as a fully decentralized system. The catch is that the federation is only meaningfully different from a custodial system if the federation operators are genuinely independent. A federation of many different companies with unique business interests in distant countries is a strong defensive bulwark. A federation of executives from different portfolio companies funded by the same venture capital firm is just a custodian with extra steps.
Decentralized L2s (“Trust this code”)
The rarest (and most commonly advertised) Bitcoin L2s are decentralized networks where there are no central operators, such as the Lightning Network. No central authority means there is no one who can betray or censor the network — in theory. In practice, decentralization is easy to claim, hard to achieve and even harder to evaluate. Decentralizing a system doesn’t make it trust free — it just replaces trust in the system operator with trust in the design and implementation of the system itself.
First that means being able to trust the code — both that it was designed and implemented correctly but also that you have access to the original, unaltered software. More importantly it means trust that the rules the code is designed to enforce are good ones. Designing decentralized trust structures is like making a wish on a cursed monkey’s paw: easy to do badly, REALLY HARD to do well.
To trust a decentralized system is to believe "even the smartest and most ruthless people in the world couldn’t take advantage of these rules." As a general rule, you probably shouldn’t be ready to believe that about very many things! Even if you happen to be one of the smartest and most ruthless people in the world (and you probably are, since you’re a Something Interesting reader) you should still be worried because all the other smart and ruthless attackers still greatly outnumber you and only one of them needs to find a loophole.
The only practical way to know if a decentralized network is secure is to let it grow in value and see whether that value stays secure. The best argument for trusting Bitcoin security isn’t how many experts have reviewed the code — it’s how much value Bitcoin has been able to successfully secure. Bitcoin is running a round-the-clock multibillion dollar bug bounty and no one has claimed the prize — and that’s the only real way to measure decentralized security. Every crypto-economic Bitcoin L2 is by definition both newer and smaller than Bitcoin, which means you should trust their security strictly less — regardless of how they work. They simply aren’t as battle-tested.
How to evaluate an L2 (and why you don’t need to)
Most people today don’t need a Bitcoin Layer 2 at all, unless they happen to already use a custodian for custody/security. To be clear, federated/decentralized Layer 2s are powerful and important — I worked on both at Blockstream! But they are also extremely overhyped, especially to retail users. You don’t need a second layer if your bitcoin is sitting still and unless you are very reckless or very wealthy sitting still is what your bitcoin should probably be doing.
If you are one of the rare people making payments with bitcoin or actively using it as financial collateral, then the efficiency and flexibility of an L2 might be worth the incremental risk — but you should certainly still proceed with caution. Being an early adopter of Bitcoin is both high risk and high reward — but being an early adopter of a Bitcoin L2 is just high risk. You should investigate those risks carefully:
Who are your counterparties? Who operates the network? Who writes the software?
How much value has this L2 secured and for how long?
Are there cheaper / safer / simpler ways to do what you are trying to do?
Can you do what you are trying to do with less important money?
Bitcoin is an incredible tool for saving and an extremely mediocre tool for spending. The people who insist that everyone should be using it for payments are virtue signaling, not solving a real problem. The right way to use Bitcoin is to put it somewhere safe and leave it alone until you really need it. Use your credit card to buy things — they have much better customer support.2
Disclaimer: I worked as a Director of Product at Blockstream on both the Liquid and Lightning Network teams. You can read more about my background in my disclosures.
If anyone gives you grief for it send them the Wikipedia entry for Gresham’s Law.