Earlier quoted context omitted.
So has so called "evidence". Can't google right now, but recently someone tried and failed to reproduce ~60% of celebrated cancer results published in Science & Nature from the last 20 years. Also, ionaddis has shown most published research in medicine and nutrition to be wrong. "Publish or perish" has replaced the gold of gold standard with tungsten, with everyone related aware of it, but no one doing anything about…
I remember the amount of non-reproducible research was as high as 90 %? Google returns some hits from May 7th(and I remember the HN post from around a month ago, we talk about the same thing I think) or so with that title, but perhaps the actual number is somewhere between 60 and 90 %?
http://www.reuters.com/article/2012/03/28/us-science-cancer-...
Thanks for correcting - it was much worse than I remembered.