It's not really wrong. It makes the correct point that randomized controlled trials are only one kind of useful knowledge. It's possible to know stuff that can't be tested that way, so blind insistence on EBM is likely to throw out some babies along with all that bathwater.
Seth Roberts often makes similar points in his blog - people running experiments on themselves of size N=1 can often discover stuff that is useful, even vital, but can't be discovered or demonstrated in a big double-blind study.
Imagine that we had no scruples and were willing to throw people out of aircrafts with dummy or with real parachutes. The ones who had parachutes would still notice when the parachute opened and slowed their fall. So we can't rule out the placebo effect! :-)
Is it we know lesser falls cause death and as such we can extrapolate out.
Or since the differences are so pronounced there is no need for randomised placebo trails?