Percy points out that 5000 years of Ephedra use in China seemed alright, so what happened here?
If by "seemed alright" you mean "sufficient data by which one can judge safety/effectiveness", I think you are taking Percy's post out of context.
From Percy's
Message 10.
But while many herbs have a long history of use, what they don't have is any history of systematic data gathering in clinical settings. Anecdotal data gathering is almost always absent even the basics of such measures as blood pressure, temperature, blood analysis and weight tracking, just to mention a few. And since there's no data gathering, there's also no data analysis.
But let's run with it.
Centuries of use = safe/effective.
How many centuries count? 1? 3? 50?
After all, we have 2 centuries worth of "evidence" that homeopathy works.
We have a 1/2 century worth of "evidence" that Bigfoot exists.
We have 20 centuries worth of "evidence" that chicken soup is good for a cold (and, according to the Torah, leprosy).
We have at least 20 centuries of "evidence" that ghosts exist.
So. What's the cutoff point?