Two-Sided Coin Control

This article is featured in Bitcoin Magazine’s “The Withdrawal Issue”. Click here to subscribe now.

A PDF pamphlet of this article is available for download. 

Self custody is an essential requirement when using Bitcoin to fully benefit from all the properties that make Bitcoin valuable in the first place. To be able to truly transact without permission, benefiting from the censorship resistance of the network, you have to control your own keys. You can’t outsource that to someone else, you can’t trust the neutrality or honesty of a custodian, you must solely have direct control of corresponding private keys to your UTXOs. If you fail to do this, you will always be a second class user. Bitcoin as a system gives you almost total control over your own funds; control of custody, when it is spent and how it is spent, even the ability to completely destroy your coins through deleting your private keys.

When you outsource that direct control of the actual Bitcoin UTXOs on the network to a third party, you relinquish that control in its entirety. That’s not to say that there aren’t middle grounds to that, such as Lightning, Statechains, and other proposed second layer designs, but ignoring those for a moment, when you do not control your UTXOs directly, you do not have the ability to transact whenever and however you want. You do not have the ability to destroy and render your coins inaccessible if you want. You do not have something that is permissionless in your ownership and control.

So why do people choose not to withdraw their coins and leave them with a custodian? Some combination of apathy, lack of understanding, fear or doubt about their ability to correctly manage their own keys without losing money, or even concerns over being able to physically keep their keys safe. There are numerous reasons, and over time we will have different solutions to address the root cause. But one of the big causes for such a choice has yet to even really happen to any serious degree; the raw economics of blockspace utilization. If you only have a couple of dollars of bitcoin –or even less in the case of zapping satoshis around with things like custodial Lightning solutions– you cannot practically take control of those coins or spend them on chain cost effectively. Even when fees get that high however, it’s still cost effective for a user in such a situation to handle their Bitcoin until they have enough to be able to afford to withdraw to self-custody at a reasonable cost.

That is not going to be the case forever. No matter what happens, if Bitcoin actually succeeds and becomes widely adopted for real use among normal people, that cost of blockspace is going to trend up; a tide that continues rising in sync with the growth of users forever. It will even rise without user growth whenever economic activity and money velocity picks up among the existing userbase. It is an inevitable reality, it cannot be stopped by anything short of the stagnation or complete failure of Bitcoin itself.

So what is the solution here? That is pretty much the root of the tug of war between the old big block versus small block divide that has been going on since the beginning of Bitcoin. Taking custody of your own bitcoin by having them sent to key pairs you control is a foundational aspect to Bitcoin, but so is being able to actually validate that a Bitcoin UTXO controlled by a key you possess was really created on-chain. The relationship between the costs of these two things is, and will forever be, an eternal tug of war between the costs of one versus the other. If you make the verification cost of blockspace cheaper and increase its availability, more people will utilize it. If you make the use of it more efficient, more people will utilize it.

You can tweak those variables all day long, back and forth, you can make computational verification cheaper, you can make blockspace use more efficient, but either one will just enable more people to use it and inevitably (unless we are all wrong about Bitcoin) lead to an increase in demand for blockspace. And that is just looking at things in a basic vacuum of economics and how demand and availability regulate each other. That isn’t even considering the actual engineering trade-offs of the specific ways to accomplish either thing, and the downside risks each optimization creates.

And there are a lot of trade offs involved in all the specific ways that either of those goals can be accomplished. A lot. Even the Lightning protocol, with all the engineering brilliance behind it, giving an exponential increase in transactional throughput, has massive trade offs and limitations. It is the most scalable while simultaneously being the most trustless second layer protocol proposed so far in terms of throughput versus trustlessness. But even it has downsides and fundamental differences.

Lightning’s security model is reactive, meaning that the only way to ensure that you don’t lose money is to pay attention to the blockchain and react quick enough if someone tries to steal funds from you by submitting an old channel state to chain. While this is a perfectly workable solution to that problem, it is a great departure from the security model of just unilaterally holding a UTXO. All you have to do in that situation is verify once that a coin sent to you on chain was actually confirmed and then you are done. You do not have to continuously pay attention to anything after that in order to keep your money secure.

This fundamental difference between using bitcoin through Lightning rather than directly on chain will have a lot of consequences for users with less money or cost tolerance for blockspace. The higher the average fee rate trends up, the more people will be pushed into locking their coins on Lightning to be able to actually spend them more cost effectively. It doesn’t even begin to end there with them being forced into a reactive security model though. Lightning routes payments through Hash Time Lock Contracts to guarantee that the money is fully sent or fully refunded across an entire payment route. This is actually never done for small value payments that are not cost effective to enforce on the blockchain if necessary. Those 1-2 satoshi payments getting zapped around for fun are sent in an entirely trusted fashion without using HTLCs and just hoping no one along the path screws up or refuses to cooperate. As fees rise on the base layer, this will have to be done for larger and larger payments. It makes zero economic sense to spend $5 in fees to enforce a payment worth only $1. Imagine $10 fees, $20 fees, etc. As the fee market matures and the base level of fees rise, even the nature of payments across the Lightning Network will fundamentally change, moving from a trustless system enforceable on-chain to one ultimately depending on honest behavior.

The same dynamics will bleed into whether or not a user can even open and maintain a Lightning channel in the first place (or whether someone else will want to allocate liquidity to that channel so the user has receiving capacity). If it’s going to cost $10 to transact on-chain, then you are immediately on the hook for 20$ –assuming fee rates don’t get even worse– for opening and inevitably closing that channel. If you have to close non-cooperatively, even with no HTLCs in flight, it is $30 because that closure takes two transactions. How much money are people going to need to put in a channel to consider fees that high worth it? Things will start getting very exclusionary very fast when fees truly start growing for good when blockspace demand saturates.

So what does this mean? Lightning isn’t enough. It gives a lot more headroom in scaling self-custody, but it does not completely solve the problem and will itself wind up subjected to the exact same economic scaling issues that are present on the base layer of the blockchain. Not to mention introducing new security assumptions in the process along the way. It’s like building up a barrier of sandbags around your house in a flood; it will keep your house safe as long as the water level doesn’t rise above it. But if we are right about Bitcoin and its adoption continues unabated, the water level will keep rising well above the top of that barrier. Lightning by itself is not enough to raise the barrier much higher.

What concrete and deployed alternative can raise it higher? Statechains are a concrete example. They can accomplish a massive increase in the efficiency of blockspace use, but surprise surprise –it shouldn’t be a surprise–, they introduce even more trade-offs than Lightning. When you deal with a Lightning channel, you open it to a specific counterparty and that is the only person you can interact with. In order to change the person you are interacting with to access routes to other people, you actually have to close that channel out on-chain and open a new one with someone else. Statechains completely change the dynamic there.

With a statechain, you can transfer coins to any new person you have never interacted with before completely off-chain. But you can only transfer the entire UTXO and a third arbitrating party is involved. Downside number one; once you lock a coin into a statechain, the whole thing can be transferred off-chain, but only all at once. Secondly, the entire way it works is by essentially trusting a neutral third party to exclusively cooperate with the current owner. The actual way its enforced on-chain can be done a few different ways, but the long and short is that the original owner creates a statechain by locking coins up Lightning-style with a service operator, and gets a pre-signed withdrawal transaction that is timelocked just like in Lightning to unilaterally withdraw. The trick is when setting up the “multisig”, you use a scheme like Schnorr where there is only a single key that each party has a part of. There are cryptographic protocols that can be used to regenerate shared keys in a way that successive users and the service operator wind up with different key shares, equaling the same public key. When you transfer a statechain, the sender, receiver, and operator engage in an off-chain protocol and the operator deletes their old share for the prior owner so they are not even capable of signing something in cooperation with that user.

Lightning is essentially a unilateral agreement between two users in which either can enforce on-chain at any time, as long as they pay attention to the blockchain. But you cannot change the channel participants in that agreement without going on-chain and paying the necessary fees. Because of how the penalty security mechanism works (take all the money from someone who tried to cheat with an old state), you cannot create those agreements between more than two people either. It is (practically, not literally, because of the computational cost) impossible to figure out a way to assign blame and penalize only the correct party in agreements between more than two people.

Statechains are that same type of agreement, except open ended in whom can be involved, as long as anyone wanting to be is willing to trust the service operator, which it should be noted can be federated among a group, and can be enforced unilaterally as long as you watch the blockchain and the service operator(s) behave honestly.

What happened here in this progression, from Lightning to Statechain, is you have made it possible for more than two people to interact safely in an off-chain manner if they are willing to trust a neutral party to enforce an honest outcome. So a great deal of scalability was gained for the cost of introducing trust on top of the already existing requirement to stay online and watch the blockchain.

Why? Because that’s really the only way to accomplish that greater scalability without adding new functionality to the blockchain. Add trust into the picture. As things stand now we can probably achieve quite a lot of scalability to the blockchain without resorting to full on custody trusting a single entity not to steal your money, but each step we take towards greater scalability will introduce more trust.

There is no way around that; either new functionality needs to be added to the blockchain or we as a collective of different groups of users need to accept that is how this is going to go. More trust creeping in at the edges for lower value use cases and lower net worth users.

There has been quite a lot of concern and discussion around this entire dynamic this year. The higher the average fee trends for space in a block, the more people will be priced out of using Bitcoin, even when you take into account things like the Lightning Network. Inscriptions and Ordinals caused a massive divide in the more active minority of people in this space, and all of it at the root was centered around the dynamic of one use case potentially raising the fees for blockspace to the point that another use case was priced out of being viable on Bitcoin.

It has been a very illuminating year so far watching people call Taproot a mistake, rally around publicly decrying the incompetence of developers in not realizing what they did, and dig in further into a dogmatic attitude. “Never upgrade or change Bitcoin again because it is perfect and infallible.” These same people in a vast overlap tend to also be the same people championing Bitcoin as a tool for self-sovereignty. They seem to always be the same people preaching self custody as a magic remedy for everything, and when scaling problems get brought up? Oh, Lightning is THE solution to that. Then they point at Ordinals and inscriptions again and start screaming about how one use case will price out another one, and so that bad one has to be stopped.

It is missing the forest for the trees. Any use of bitcoin that is profitable and cost effective to deal with demand is going to happen. There is literally no way to stop that, and Bitcoiners convincing themselves they can are fooling themselves. All of the backlash against Ordinals and Inscriptions very quickly led to people intentionally doing even more costly things like STAMPS, which instead of using witness data that doesn’t have to be stored in the UTXO set, puts their data inside the actual UTXOs. Rather than acknowledging the reality that if people think it is profitable to pay for blockspace they will, many people are falling victim to a knee jerk reaction of trying to stop what they think is bad while completely ignoring the reality that there are other worse ways to accomplish the same thing anyway if it makes economic sense. An impulsive reaction to the rise of Ordinals and Inscriptions is dragging down the entire attention span of involved people in this space into a pit of wasted efforts to stop things causing fee pressure that they don’t agree with instead of considering how to adapt and scale things they do agree with to that fee pressure.

A good percentage of the people engaging like this are literally arguing with the wind. They’re trying to tell us to stop blowing because it is knocking things over instead of tying things down or weighting the foundation to weather it. If you successfully block or censor Inscriptions, people will just use STAMPS, or OP_RETURN, or techniques even more wasteful of network resources.

Ultimately no technical filter will be good enough to stop people from doing dumb or non-monetary things with the Bitcoin network. The only filter that will successfully stop anything from being done on Bitcoin is economics. And that filter is equally created and equally affects every use of Bitcoin. It’s time to stop trying to fight externalities driven by economic demand and try to counter them through improving efficiency.

If you think Bitcoin’s primary value and purpose is to transfer value, then rather than obsess over somehow stopping all other uses of Bitcoin, you should be focused on considering the trade offs of different mechanisms that can improve its efficiency in transferring value. You are either going to have to choose between progressively adding more trust to things in order to accomplish that, or adding new features to the Bitcoin protocol itself to build more efficient things without depending on trust.

Buraq, the infamous slayer of Lightning, has recently proposed TBDxxx, a new second layer protocol. It is essentially a big multiparty statechain/ecash system that is non-custodial, does not require trusting the service operator like a statechain, and can pack many users into a single on-chain UTXO. This requires ANYPREVOUT(APO) or CHECKTEMPLATEVERIFY(CTV) to work, so it needs a consensus change. Channel factories are a way to take a single UTXO and stack Lightning channels on top of each other, so one UTXO can represent dozens of users who all have a regular Lightning channel at the top. This also requires ANYPREVOUT.

Both of these proposals can scale the use of Bitcoin to transfer value much further than Lightning can now, but ultimately both of them are subject to the same economic fee pressure that Lightning and on-chain use are. To join one of these multiparty channel pools, or exit one, or enforce something non-cooperatively on chain you still have to pay fees. For something like a channel factory this will involve one person who needs to close or enforce something actually unfurling and closing (fully or partially) the entire channel factory with everyone in it, creating costs and on-chain implications for everyone. Even despite accomplishing a huge increase in scalability without trust, it still falls victim to the effects of the blockspace market maturing.

In order to mitigate (not solve) that, we will likely need even more OP codes. Things like OP_EVICT or TAPLEAFUPDATEVERIFY. OP_EVICT lets a group collectively kick a non-cooperative member out of a multiparty channel without closing or affecting anyone else in it using a single transaction with one input and two outputs. This doesn’t solve the issue, but it makes it a lot more efficient by allowing one person to be evicted with a much smaller on-chain footprint. TLUV accomplishes the same thing except instead of everyone else kicking someone out, it allows a single user to withdraw all their funds without disrupting anyone else or needing anyone else to cooperate.

To address more of the issues, we need to make more changes to Bitcoin. There’s no way around that. Taproot “opened the door” to Inscriptions in the sense that it relaxed limits enough for people to go nuts with it, but they were already possible before Taproot. You can look at Taproot as having provided efficiency gains for both monetary use cases as well as non-monetary use cases. It made multisig the same size as a regular single sig address, which helps make using a higher security set up for keys or second layer protocols cheaper, but it also made it cheaper to inscribe arbitrary data.

Two sides of the same coin. And that is how it is. Same as it ever was. Making use of the blockchain more efficient is not always going to improve solely the use case you want, but it is absolutely necessary to scale Bitcoin in a way that is self-sovereign and self-custodial. It’s time to either accept that and start considering the reality of finding the optimal efficiency gains for value transfer with the least efficiency gains for detrimental or non-value transfer uses, or it’s time to accept that the only way to scale value transfer is to introduce trust.

A good number of people in this space have already made their choice one way or another, but there is a large contingent of people in the middle who refuse to accept either. This loud group in the middle needs to wake up and smell the coffee, and accept the reality of the situation. This is how blockchains work. Pick one; either brace yourself to accept the injection of trust into things, or accept the reality that changes need to happen. You can tell yourself all day long that you don’t have to choose, but your actions in attacking the notion of any change to Bitcoin at all while simultaneously championing self-custodial Bitcoin as a solution for the world are implicitly making the choice to accept more trust being introduced into the system, whether you want to acknowledge that or not. 

This article is featured in Bitcoin Magazine’s “The Withdrawal Issue”. Click here to subscribe now.

A PDF pamphlet of this article is available for download.