[logfromblammo]

The only acceptable way to kill unscientific nonsense is to test the hypothesis it advances, no matter how ridiculously untestable it seems to be, and then publish the raw data without redactions.

This one actually seems testable. Get a bunch of volunteer non-Celiac "gluten sensitives" and put them on a controlled diet. Measure some objective physiological metrics. Blindly introduce gluten to random subjects. Determine which metrics, if any, change in response to gluten.

And then, to check out placebo/nocebo, you tell all the subjects that gluten will be introduced into their diets that day, and continue monitoring the metrics.

A similiar experiment could be performed on a random sample of ordinary people, with glyphosate.