“Why Most Published Research Findings Are False”
by Yi Zhang
I found this essay from PLoS is quite interesting and, unfornately dismaying. Even though it is mainly addressing on the medical science, I found what it says is quite true in my field, with some other characteristics added. I always believe that the computational mechanics is more “engineering” than “science”, because what most people are doing here is to mimic the mechanical reality instead of searching for answers to hypothesis, as many other scientist are doing. Because of this nature, bias is more likely to be confirmed using human generated data, after all, all the data in computational mechanics field is “human generated”. Another problem lies in the fact that research groups around the globe are pursuing same/similar topics with thousands self-complied codes. Those codes are what most young computational mechanics researchers start their career. Their “weapon of massive data”, are often examined within small circle, sometimes just one group. Besides, I believe even those codes are open, most of them would be practically impossible to be peer-reviewed: it’s just too much work.
There are some other interesting and also dismaying points made in the essay, like “the hotter the scientific field, the less likely the research findings are to be true”. Oh, my, I hope they are not talking about turbulence simulation.