Search This Blog

Wednesday, November 29, 2006

The Hwang Paper

The very first Cargo Cult Alumnus was Dr. Hwang Woo Suk. He earned this reward by pulling the wool over the eyes of the editors and reviewers of the journal Science. He also gained international fame through the claims he made in his Science papers. And he was lying the whole time. The fraud was discovered not through peer review but because a whistle-blower in Hwang's lab spoke to a South Korean TV station.

In order to correct for their embarrassing inability to tell shit from shinola, the journal appointed a panel to review the review process that led to the incident. The editor of Science, Donald Kennedy, said that the journal accepts the panels major findings. What the panel decided was that this whole thing could have been prevented by extra review procedures. Is anyone seeing a pattern yet?

These panels are populated by reviewers. What good would adding more people and more reviews do? It's like trying to clean off a muddy windshield with a newspaper. No matter how much you scrub you just end up smearing the mud from place to place. More reviews creates more chatter but it doesn't have the power of going into the lab and repeating an experiment. What if this latest panel on the Hwang case would have made the recommendation to have key experiments verified by a non-partial contract research organization? Perhaps such a recommedation would diminish the perceived authority of a scientific panel.

The Science panel said that a risk assessment method should be developed to flag high-visibility papers for further review. Also, authors should specify their individual contributions to a paper, a reform aimed at Hwang's stratagem of allowing another researcher, Dr. Gerald Schatten of the U. of Pittsburgh, to be a lead author of one of the reports even though Schatten had done none of the experiments. The Cargo Cult Scientist would like to point out the no respectable scientist in this age ever does an experiment. They may run a lab where the experiments are done but they are as close to the work as any other co-author whose name is of value to the paper. As for the risk assessment method to be developed, we see another panel being formed in the future.

A scientific paper was once a thing of great beauty. A scientist was once a person who observed nature first hand and offered up explanations that made the world take notice. The ease with which Dr. Hwang hoodwinked this major journal sent out a message. We're doing something wrong. We have a system where highly ambitious people are told that the only way to accomplish there ambitions is to publish papers on how great they are at conducting research. Negative results are not very welcome. The best way to get published is to tell the editors what they want to hear. If, for example, RNAi is popular with a journal you must validate the concept by reportinig how well it knocked down your gene of interest. Failures are just not welcome. How many papers were published last year about RNAi experiments that failed? How many people just couldn't clone a stem cell? Dr. Hwang said he could and he was published. Hmm.

We have a dream here at the CCS. Just as the computer revolutionalized the way offices are ran throughout the world, we believe they can help medical science. A computer is a cold and impersonable entity. It doesn't care of you need to tell your boss that the ELISA assay worked out as planned. It can only report the data it receives. I'll end with a simple example of how this could work.

An ELISA can be developed to the point where all a technician needs to do is add the constituents and read the plate. The plate has a code indicating that the ELISA is, for example, one of three. If one of the ELISAs needs to be thrown out it must be justisfied. The cubicle scientist must understand the system and interpret the data based on the data. NO BEER AND PIZZA DATA HANDLING. If a scientist is forced to design experiment to the point where this will work, the scientist will become a better researcher. Reviewers will have to really understand the designs they are reviewing. Will this ever become a reality? We here at the Cargo Cult Scientist are forming a panel to discuss this very question.

No comments: