I keep hearing a lot about publishing negative results in clinical trials, and there has been a lot of push towards it. I think its also time for the same to happen in engineering journals.
There are perhaps two broad categories of engineering research, one where the approach is to take an unsolved problem and find/develop the appropriate tools to solve it, and the other where the researchers find the right problem to apply the tool they have been developing. Both have their uses, and quite frankly, from my experience, it has been very frustrating how these two situations have been treated in the publishing world.
For the first kind of problems, I have seen many very good papers rejected because the claim from the reviewers was that there were no new techniques developed in the paper. Of course, the reviewer has completely failed to see the point of the paper in this situation.
The second situation, I think, can run into even bigger problems. Consider a researcher working on developing computational methods for some abstract mathematical field, and wants to find an appropriate problem to showcase the usefulness of such computational methods. In this situation, it is very likely that the methods doesn't outperform other existing methods, or that it turns out (due to many practical reasons) that the method is actually not applicable for problems where the underlying mathematical theory seems to imply it should work. In both these cases, the resulting work is very unlikely to get published. As a result, researchers feel often forced to exaggerate the usefulness of their computational methods, or perform and unfair comparison with existing methods. If the researchers are not willing to do this, then the research of the second kind is not viable in the long run.
There is a simple fix to this: publish negative results.
No comments:
Post a Comment