One concrete data point for the thread. If you are comparing against somebody else’s result, check that you are comparing the same measurement taken the same way. Most of the apparent contradictions in these threads dissolve at that step.
Correct me if the detail matters more than I have assumed.