Actually, this is sometimes done exactly as you describe. See "AMD Anti-Lag" or "Nvidia Ultra Low Latency"/"Nvidia Reflex Low Latency".