Supporting decentralized staking through more anti-correlation incentives

The article proposes extending anti-correlation penalties to routine failures, such as missed attestations, so that validators running many nodes on shared infrastructure face higher penalties when they fail alongside others. The author's preliminary analysis compares actual co-failures within publicly known staking clusters (such as "Lido" and "Coinbase") against expected co-failures under two baseline methods and a fake-cluster method, finding excess correlated failures within clusters. In a simulation, an "excess" penalty scheme reduced the penalty advantage of big stakers over small ones, from about 1.4x to about 1.3x under the basic dataset, and the article notes that the code is publicly available for independent replication.