Skip to content
Crypto.world

The Ledger

Lightning force-closes cost 2–6 times more in fees, preprint finds

An arXiv study of 973,581 Lightning channel closures found that force-closing cost two to six times as much in fees as cooperative closure. It also measured delayed payment timeouts during high-fee periods and compared the older anchor design with zero-fee commitments.

Lightning force-closes cost 2–6 times more in fees, preprint finds

Force-closing takes a larger fee toll

Lightning force-closes cost two to six times as much in transaction fees as cooperative closures and locked the closing operator’s balance for a median of 1.7 days, according to an arXiv preprint covering nearly a million channel closures. The paper, submitted Nov. 12 by Saurabh Kumar of Bitcoin Data Labs, measures closures from 2018 through October 2026 and examines fees, delayed payment resolutions and the use of anchor outputs.

The study traced 973,581 channel closures and followed the transactions that spent their outputs. Its data included 523,815 force-closes, in which a channel is closed without both parties agreeing to a cooperative transaction. The author also checked public-channel closures against a separate chain index, which confirmed all 698,476 cases the two methods could compare.

Fee spikes did not bring a closing wave

In a fee-spike week and the following week, force-closes were 0.97 times their level in the preceding four weeks; ordinary weeks in the same years registered 1.02, according to the study.

The author says the cost of a force-close also depends on the time funds remain locked and on pending payments. The closer’s balance was locked for a median of 1.7 days before it could be swept, with that delay outweighing the transaction fees only in the study’s low-fee years, 2025 and 2026.

High fees were associated with more delayed payment timeouts. In 2023, 18.2% of hashed time-locked contract (HTLC) timeouts confirmed more than 40 blocks after expiry, the paper reports. But it found that forwarding nodes lost 38 payments since 2021, and each had its payment preimage on-chain already.

Anchor design and the shift to zero-fee commitments

When the block median fee rate at expiry was 50 sat/vB or more, 20.6% of HTLC timeouts on anchor channels confirmed more than 40 blocks late, against 30.5% on channels without anchors. The study also found that owners rarely used anchors to bump commitment fees, while miners took most of the value of what third parties swept.

The paper sets those findings against a change in Lightning’s fee design. Bitcoin Core 31.0 removed the mempool rule that anchor channels relied on in April 2026, and the Lightning specification adopted zero-fee commitments in May 2026. The paper says those commitments would save 12% of the anchor mechanism’s block space, based on its comparison of the formats.

The first zero-fee commitment close in the study occurred on Oct. 29, 2025, before the specification was adopted. The author says the new format should be evaluated by how promptly payments resolve during fee spikes, rather than by block-space savings alone.

Limits of the chain data

The analysis relies on on-chain activity, which does not show which party broadcast a force-close or whether attempted fee bumps failed to confirm. Private channels appear in the data when they force-close, but their cooperative closures resemble ordinary payments, so the author says their number and force-close rate cannot be determined from the chain.

It describes the measurements as a baseline for the transition from anchor channels to zero-fee commitments.

Sources

  1. Original source · Preprint arXiv arXiv:2611.16311

Related stories

Explore

More articles