[bhouston]

One sample can always be really "lucky."

I feel that we need to collect more data and have it open in some way so that people can look for "unexpected" correlations and figure out if people in one area or exposed to one thing are unnaturally lucky in general.

I wonder if in the future one could use purchases (e.g. harmful products which we do not know they are harmful) and GPS data (e.g. exposed to environmental factors) correlated with health records to do forensic large scale analysis? Still I wonder if the signals are strong enough to show up?

It is sort of like generic trait attribution via population studies.