Client diversity is often described as the Lightning Network's main defense against a bug in any single implementation. But a new paper on arXiv argues that this defense is weaker than assumed. The authors report that LND, one of the main Lightning implementations, runs 84% of public nodes and holds 84% of public channel capacity — a share that has barely moved since 2023.
The paper draws on the 2026 Core Lightning shutdown to examine what happens when a major implementation fails. While the abstract does not detail the shutdown's effects, the authors use it as a natural experiment to question whether diversity can meaningfully protect the network when one client so thoroughly dominates.
The findings suggest that the Lightning Network remains highly concentrated, and that claims about resilience through diversity need to be reexamined. The study does not propose a fix, but it highlights a structural risk that persists despite years of discussion about decentralization.