We seem to be stuck in perpetual loop of scientific papers across disciplines ranging from economics to medicine being repeatedly questioned for lack of reproducibility, plagiarism and outright fraud. The solution seems obvious - scientific notebook publishing replete with data and calculations. The technology exists and has done so for close to a decade - Wolfram's Mathematica came out with its first notebook release in 1988 and IPython notebook in 2011. This solution has been discussed ad-nauseum but the implementation seems to be ad-infinitum. Is it merely an adoption issue or are there other hurdles?
Except that there is no standard of a "scientific notebook". It's not a form one fills out. It is notes taken during experimentation. Data analysis generally takes place outside of a scientific notebook, which simply contains raw data and observations, the completeness of which is up to the experimenter. Notebooks contain proprietary, private data, owned by the organization overseeing the research. They have no incentive to publish their "unpublished" results that have been expensive to produce.
There is no easy solution.
A lot of people like to treat "science" as this unassailable "truth", but science is political. It is subject to bias and it is corruptible. Results are skewed; inconsistent results are discarded. "Science" is incentivized to produce results consistent with the desire of those who finance it. This is not a "China" problem by any stretch.
3 comments
[ 5.4 ms ] story [ 35.7 ms ] threadA lot of people like to treat "science" as this unassailable "truth", but science is political. It is subject to bias and it is corruptible. Results are skewed; inconsistent results are discarded. "Science" is incentivized to produce results consistent with the desire of those who finance it. This is not a "China" problem by any stretch.