The journal Nature has a report this week from a former researcher at pharmaceutical giant Amgen that of the 53 promising, groundbreaking scientific findings that they attempted to pursue to develop new cancer treatments, 47 of them ended up being junk science. When researchers tried to replicate results independently, they could not. Another team at Bayer was only able to validate about 25 percent of the research that they pursued.

In many cases, the original researchers only shared data with the Amgen scientists on the condition that they not release any contrary findings to the public. As a result, those 47 junk papers in peer-reviewed journals are still circulated as valid science and are being cited by other researchers as support for further research in the field.

Retractions at scientific journals are up tenfold over the last decade, while the number of articles published in scientific journals rose only 44 percent. (I've read elsewhere that the vast majority of scientific research never gets independently verified by other researchers, meaning that the mistakes that are caught represent only the tip of the iceberg.)

The article in Nature focuses on medical research, and specifically on cancer research. The rate at which junk science is produced and accepted in the scientific community may say a lot about why we struggle to find cures. But I've read a number of other articles in recent years from other areas of science that amount to a growing chorus of people complaining that our scientific research is increasingly turning into a sloppy, unverifiable mess.

Science has given us many wonderful discoveries and made possible many fantastic practical applications. We can credibly believe the underlying science in many of these cases because the technology built on those scientific findings works. But if almost 90 percent of the promising research done in some fields today is completely wrong, and if scientists in many fields are demanding public money and radical policy changes based on findings that have not been verified or replicated and that may be 90 percent junk, then we have a problem.


Sorry, no video for this one, but it's an interesting and important bit of news. Read the Reuters article on this here:

http://www.reuters.com/article/2012/03/28/us-science-cancer-idUSBRE82R12P20120328

Read the Nature article here:

http://www.nature.com/nature/journal/v483/n7391/full/483531a.html