The point was that if the data is public, more independent observers can perform experiments (as in your other example, taking their own arsenic levels). This m

The point was that if the data is public, more independent observers can perform experiments (as in your other example, taking their own arsenic levels). This may be funded corporations or citizen watchdog groups or whatever. You're talking about arguing about whether certain experiments were valid and you can do that. Once you agree on a dataset, the mean is the mean and that's it. My point is that if ALL the data is released then some group will analyze it with all the outliers in and publish that.