This is why science needs to formally pre register hypotheses. "People tested this ten times and it worked once" is infinitely more useful info than "someone published a paper because this works! But we have no idea how many people tried exactly the same thing and got a negative result".