Image duplication software is not the solution
Data sharing is a partial solution
More and more claims are coming out against Harvard. A key dei faculty member has plagiarized. There is more image duplication.
In response to these claims, some have suggested we need more plagiarism checks and better image duplication software.
That's not nearly good enough.
The only reason we're finding plagiarism and image duplication is because the raw data is directly available. You can look at the pictures, you can read the text.
Instead, most papers just present the findings. They do not show you the data set, and they do not show you the analytic plan.
Which is more likely: The only problem in academia is image duplication or all parts of science have his problems, and the only place we pick it up is when we can see the raw data?
It's the latter.
We need complete data transparency. People who publish papers should have to post the de-identified data sets. They should have to post the code.
Even with this, they can still cook the books. Data can be fake. But at least it would be one more way to catch them.
Ultimately science is the greatest thing humans have ever done.
… Incredibly competitive
… Full of false information
… A massive welfare program for rich people.
Because we all pay for science, there's an obligation to make sure it is not fraudulent. Data sharing should be the norm. Don't get distracted by the Western blots. The problem runs deeper.