Payment channels

Wonder

1. How can two parties conduct thousands of instant payments while only submitting two transactions to the Bitcoin blockchain?
2. What mechanism prevents a participant from broadcasting an older commitment transaction that favors their balance?
3. Why were the first payment channels, described by Jeremy Spilman in 2013, one-way only?

Explore

1. Bitcoin’s base layer settles transactions at approximately ten-minute intervals, creating latency and fee constraints for high-frequency micro-payments. Payment channels relocate the computational burden of state updates off-chain, allowing participants to exchange signed balance states indefinitely without network propagation. The on-chain blockchain serves solely as a settlement layer, verifying only the initial funding and final closure events.
2. The channel begins with a funding transaction: a 2-of-2 multisig output locking funds controlled by both Alice and Bob. If each contributes 1 BTC, the channel capacity is 2 BTC. Neither party can unilaterally spend these funds without the other’s signature. This structure binds the participants together in a shared account that exists only as a potential claim on the main chain.
3. During operation, Alice and Bob exchange commitment transactions representing new balance distributions. Each update replaces the previous state; if Alice sends 0.1 BTC to Bob, they sign a new transaction moving funds from Alice’s output to Bob’s. These signed but unbroadcast transactions are valid Bitcoin transactions, ensuring that any final state can be enforced on-chain without relying on counterparty trust.
4. Each side must also be stopped from broadcasting an outdated state that paid it more. Lightning does this with revocation secrets and timelocks, covered in a later lesson; in short, publishing a revoked state lets the other party take the whole channel balance, which makes cheating a losing bet.
5. Closure occurs either cooperatively or unilaterally. In a cooperative close, both parties sign a settlement transaction that pays their final balances immediately without timelocks. If one party disconnects or acts maliciously, the other can force-close by broadcasting the latest commitment transaction. On this unilateral path, the party who force-closes must wait out a relative timelock before spending its own share, leaving time for a penalty claim; the other party’s output is available almost at once.
6. Precursors include Jeremy Spilman’s one-way channels (2013), in which payments could flow only from payer to payee, and Christian Decker and Roger Wattenhofer’s duplex channels (2015). Joseph Poon and Thaddeus Dryja’s Lightning Network paper (2015 draft, 2016 version) added revocable commitments, so a channel can stay open indefinitely, and HTLCs, so payments can route across many channels. Mainnet implementations like LND, Core Lightning, and Eclair now manage these states, with significant adoption beginning in earnest during 2018.

Key facts

Exercise nine, six, three, one

Fact check

Dates, figures and claims checked against the published sources listed below.

• J. Poon & T. Dryja, 'The Bitcoin Lightning Network: Scalable Off-Chain Instant Payments' (2016)
• A. M. Antonopoulos, O. Osuntokun & R. Pickhardt, Mastering the Lightning Network (2021)
• Lightning Network BOLT specifications (github.com/lightning/bolts)