I think this is a false dichotomy. The chain-of-thought can provide interesting information even if it isn't perfect. In any case, we should work to move in t

I think this is a false dichotomy. The chain-of-thought can provide interesting information even if it isn't perfect. In any case, we should work to move in the direction of increasing fidelity, not decreasing it.