[parennoob]

If you have tests that don't provide information, that's the problem of the test, not the person getting it. Doctors are still the people actually handing out prescriptions, so if they start telling people, "Look, this $10 test from Theranos isn't reliable as a predictor, stop doing it.", it will create a feedback loop that will stop people from doing those tests that, as you say "border on quackery".

Meanwhile, if I suspect that, for example, that my BMI veers into the obesity range when my TSH is high, it should be perfectly fine for me to try and get a $50 test to examine if there's a correlation. Better than spending that on a DVD of Keeping Up With the Kardashians, don't you think? Might even increase awareness about general health in the US.

[enziobodoni]

You misunderstand Bayes. The point is that the test is the test, and the information value is all dependent on the person getting it, not the test.