The idea of Virtual Payment Channels was initially outlined by Lightning Network (LN) developer Rene Pickhardt in his blogpost. The main advantage of these specific channels might be in increasing the scalability and reliability of large payment providers. In addition to the purely technical advantage, channels of this type might be used by counterparties with certain trust assumptions for increasing the reliability of their routing while decreasing onchain costs. The author refers to the creation of Virtual Payment Channel as an equivalent to the creation of satoshis out of thin air and thus assumes that they generally enable fractional reserve in LN.
Rene Pickhardt’s post inspired Bitcoin Lightning Wallet developer Anton Kumaigorodsky on the creation of so-called Hosted Channels. These channels may be bootstrapped by the user of the BLW in a matter of seconds. Together with other design decisions, it allowed BLW users to get ready for receiving LN payment faster than with any other wallet available on the market. However, even after a couple of years of successful operation Hosted Channel concept remains largely unknown to the broader Lightning Network and Bitcoin community.
In this post, we consider the main properties of the Hosted Channels and evaluate their economic effect. Along with them Public Hosted Channels are going to be adopted in the first mobile-optimized LN library IMMORTAN.
Fractional Reserve or Bitcoin DeFi
Fractional reserve is a general definition for a banking system running on deposits to credits ratios different than 1-to-1.
In the modern world paper money and fractional reserve altogether create tremendous credit expansion. It is important to stress that fractional reserve forms the foundation of the contemporary financial system. For every single unit of account commercial banks are allowed to emit approximately ten times more units to finance somebody else’s operation via credit and my information about the actual ratio may be outdated. The obligations to pay debt may be used as an asset nearly everywhere depending on the externally defined ratings or grades of the debtor. The deposits are only the tip of the iceberg.
On the contrary, Hosted Channels allow leveraging trust in the payment network built exclusively on hard money. When established such channel has zero liquidity on receiving side which initiated channel creation. The receiver generates a new invoice and then satoshis get routed into his Hosted Channel. At the network level, nothing has changed, no extra satoshis created nor destroyed. However, Hosted Channel provider now liable to route payment from Hosted Channel to any network node on demand. This moment defines the actual situation when the host and the client establish trust relations equivalent to borrowing money.
Loans are the thing DeFi projects usually advertise as a milestone achieved using the Ethereum protocol. But this is just marketing. Credit by definition requires trust and in the decentralized world, it is usually platforms that aren’t protected against Sybill attack. Hosted Channel provider can build a reputation using at least its constant domain name and therefore replicating known concepts like firm or trademark in the digital world. Going further we could remember that in Lightning Network Hosted Channel provider has its public key with associated history in the network gossip, even partially recorded in the Bitcoin blockchain. The Lightning node has a natural incentive to build a good reputation which can be leveraged in the creation of economically scalable trust relations in Hosted Channels while operating on common standards with the rest of the network. Channels of this type are real and cheap “checking accounts”.
We’re talking about credit analysis. So, please: let’s stop calling it fiat money. Let’s start calling it what it is: credit money. Source
To some extent, after creating a Hosted Channel the client now rents liquidity in the provider’s channels. However, the provider still may use it for routing and it is not stuck in the dead-end of the LN graph. On the lower level, the system appears now much like early Middle age banks with double-entry bookkeeping which allowed to lower transaction costs in gold and silver. At the higher level, everything is plugged into Lightning Network built on open standards and providers compete with each other in doing better service available 24/7/365. The Host and the Client are in trust relations much like clients of banks from the free banking era. As Lawrence H. White wrote in his book Free Banking in Britain: Theory, Experience and Debate 1800–1845, Second Edition, the Scottish free banking system was robust and resilient.
“…of one of the most remarkable features of Scottish free banking: the unlimited liability of a bank’s shareholders. Despite their magnitude, Ayr Bank’s losses were borne entirely by its 241 shareholders. The claims of its creditors, including note-holders, were paid in full.” (p. 29)
We do not have enough historical data to judge LN operators but we could directly project equity capital requirements for Scottish banks into real capital demands for working on the Bitcoin Lightning Network.
Evaluating Economic Effect
The authors of the “Lightning Network Economics: Channels” (or refer to the video below) conducted a theoretical study of the Lightning node economics using applications of Queueing theory to the payment systems and comparing channel costs for two configurations: unidirectional (also called unbalanced) and bidirectional (balanced).
Let’s assume that for the total 19990 Hosted Channels we have 76580 transactions for a year approximately. These numbers were obtained in personal communication with BLW developer Anton Kumaigorodsky. It brings an average payment rate equal to 0.26. In the paper referred to earlier, the maximum payment rate considered by authors is 0.2, and the maximum interest rate reflecting opportunity cost is equal to 1% per annum. The unilateral channel is reliable when payment rates are close to the maximum of 0.2 if the opposite side doesn’t do any payments (so its rate is 0).
I tried to look at possible interest rates which could make paying with Hosted channel an unreliable action and simulated optimal unilateral channel costs for the normal channel according to Guasoni et al paper. At the r=1% chosen by authors it is possible to get the positive economic effect in 0 < lambda < 0.2. Higher rates demand higher payment frequency than average measured for Hosted channels. Sidenote: we could safely assume that currently many wallet providers heavily subsidize their clients if BLW statistic is representative.
The original model was built for “real” trustless LN channels, but Hosted channels do not create any blockchain footprint and thus have no associated onchain cost. It does mean that we can offset the economic effect horizontal line for value B=1 (dashed line) and accept such difference as a competitive advantage of a hosted channel over a regular lightning channel. It gives us lambda=0.2 as a point where it is equally valuable for the user to use onchain transactions or pay with liquidity from the hosted channel. Having them implemented BLW had a competitive advantage over other wallet providers and likely had the lowest possible infrastructure costs in a hindsight.
From a technical point of view, Hosted Channels provider has an effective way to reduce costs of maintaining custodial payment service native to the rest of the network. The end device is unique and has its own key secured by a PIN or biometrical data and therefore anti-fraud protection becomes a non-issue for the service. Sybils in such network also have no advantage over regular users since provider in basic configuration does not lend liquidity and users must deposit satoshis into hosted channels before start using them.
Further development of Hosted channels may include more risky technologies having fewer assurances like creating hosted channels with initial deposits which is equivalent to creating credit money in fractional reserve banking or making hosted channels with different denominations than the basic currency of LN. At this point, hosted channels provider indeed ceases to be a full reserve bank and puts its reputation at stake, and has to manage various risks. History knows examples of relatively robust banking systems running at reserve ratios less than 5% on gold money. It is unknown if some proof-of-reserves technology may be applied to such banks.
If you like this post, please do not hesitate to donate via LNURL.