EIP-8361 is being debated as monetary policy. Our dependency graph says it is also infrastructure policy for DeFi. We ran percolation analysis on every on-chain dependency in Ethereum DeFi, roughly 12,200 building blocks. Two findings matter for this debate: 1. Remove 1% of the network at random and connectivity barely moves (90.5% to 89.4%). Remove the top 1% of hubs instead and the connected web collapses to 21.3%. The system is robust to chaos and fragile to anything that lands on the hubs. 2. The hubs do not back each other up. Assortativity is negative (−0.24) and the rich-club coefficient falls to zero at the top. Each hub stands alone. Now look at what the hubs are. In our live systemic-token ranking, five of the top eleven are ETH liquid staking tokens. wstETH alone sits under roughly $1.6B of potential forced-sell pressure from positions that use it as collateral. Staking yield is the gravity that built that layer. Every leveraged staking loop, every wstETH lending market, every restaking wrapper prices off it. EIP-8361 would cut net consensus yield from about 2.6% to 1.2% today, and to zero if half of ETH ends up staked. There are many valid arguments on Ethereum issuance policy. The following is not one, but a view on potential systemic risk of the proposed policy. The debate so far weighs stakers' revenue against ETH's monetary premium, and the dependency graph adds a third item to the ledger: DeFi's collateral layer is concentrated in yield-bearing ETH wrappers with no redundant backbone behind them. Changing the economics of that yield changes the load on the load-bearing walls. Which walls, exactly? We ranked every building block by that question. That list is in Part 4, out next. The percolation analysis behind this post is Part 3, linked below: https://t.co/rA6wouAHHN