LN-Penalty

LN-Penalty is a state protocol that penalizes a party who publishes a past state by allowing their funds to be seized by their counterparty. The protocol is most widely used as the payment channel protocol (PCP) at the heart of the original (and still only fully developed) version of LN. It is also used to manage state for repeated off-chain DLCs between two parties.

The concept for LN-Penalty was developed by Joseph Poon and Tadge Dryja in the original Lightning Network paper, with refinements focused on practical implementation proposed by Rusty Russell and others. At the time of its development, it offered several advantages over previously proposed PCPs:

• ● Bidirectional: the BIP65 PCP provided the first trustless payment channel protocol construction for Bitcoin, but it only allowed payments to safely flow in the direction of the receiver. By comparison, LN-Penalty allows payments to flow in both directions.

• ● Non-expiring: the Spillman PCP allows bidirectional payments but the channel must be closed by a deadline (and each change of direction reduces the time to the deadline). By comparison, LN-Penalty allows payment channels to stay open indefinitely.

• ● Unlimited state updates: because each BIP65 state update must change channel balance by at least one satoshi and each change of direction in a Spillman channel reduces the time to until expiry, both only support a limited number of state updates. By comparison, LN-Penalty theoretically supports an unlimited number of state updates. In practice, the deployed version of LN-Penalty is currently limited to about 1.5 billion state updates as part of simplifying support for watchtowers.

At the time LN-Penalty was proposed, it required relative locktimes (added through soft fork activation of BIPs 68 and 112) and a fix for txid malleability (added through soft fork activation of BIPs 141 and 143).

Several limitations of LN-Penalty have been identified:

• ● Toxic waste: if any of a user’s old states is broadcast, that user will lose all of their funds. That can happen if the user’s backups are compromised or if their software resets to an earlier state (e.g. due to a loss of newer state stored in volatile memory). This runs contrary to the general design in software to make backing up and restoring as easy as possible. PCPs that don’t create toxic waste include Spillman channels, CLTV channels, proposed duplex payment channels, and proposed LN-Symmetry.

• ● Two party only: old states may allocate more funds to a party than they are entitled to in the latest state. Penalizing the party who publishes an old state by allowing all of their funds from that old state to be seized effectively disincentivizes that behavior. That mechanism only works well for two users where funds either belong to Alice or Bob. For more users, say three users, Alice can publish an old state that assigns to her funds that the latest state assigns in part Bob and in part to Carol. If Alice publishes that state, both Bob and Carol must be able to act unilaterally to recover their funds (otherwise it isn’t a trustless protocol) but they each must not be able to seize the other’s funds (again for trustlessness). The inability to scale the protocol to multiple users in a single channel can be limiting in the context of channel factories, DLCs, and other protocols. PCPs that allow multiple parties include Spillman channels, duplex payment channels, LN-Symmetry, and multiple variants of the tunable penalty protocol.

• ● Fully penalized: the publication of an old state results in that party losing 100% of their balance in the channel. There’s no way to increase or decrease that penalty. PCPs that allow adjustable penalties include variants of the tunable penalty protocol and a variant of LN-Symmetry called Daric; it’s likely the case that an adjustable penalty mechanism could be added to some of the other PCPs mentioned in this article.

As of this writing, LN-Penalty is the only fully developed PCP compatible with LN.

Primary code and documentation

• Lightning Network
• Deployable Lightning

Optech newsletter and website mentions

2024

• Superscalar: a timeout tree channel factory proposal with leaves using LN-Penalty

2023

• Using covenants to improve LN scalability (compatible with LN-Penalty)
• Proposal for fraud proofs for providing outdated backup states to make LN-Penalty safer

2022

• Factory-optimized LN protocol compared to LN-Penalty
• Proposed new transaction relay policies designed for LN-penalty

2019

• Discussion of LN-Penalty versus proposed LN-Symmetry (eltoo)

See also

• LN-Symmetry (eltoo)

• Duplex micropayment channels

  
  

Previous Topic:
Liquidity advertisements

Next Topic:
LNURL

Edit page
Report Issue