Recently one of my colleagues received an email from the editor of a physical journal, with the following statement at the end, I quote, “…please note that policy of XX as leading YY journal in ZZ physics is … being correct is not enough to be accepted.” The statement is true, and I am not going to elaborate on what is enough to be accepted (one sufficient condition is described in the article http://www.bbc.co.uk/news/science-environment-15697636). Is the above statement equivalent to: being published in leading journal is enough to be correct? Is it a true or false statement?
After this general comment, I return to one example of the importance of the rule of procedure, based on our recently submitted on arXiv paper. In a few published papers on Horava-type models we found the algebra of constraints, and we knew that such an algebra couldn’t describe spatial diffeomorphism invariance. The simplest way to demonstrate it was to check the differential identity (or Noether’s identity) without the need to reference the Hamiltonian results, and this is what we did. Because the Hamiltonian and Lagrangian methods must produce the same result, we looked for a term that destroys the differential identity corresponding to the spatial diffeomorphism gauge symmetry. We thought we had found such a term, and concluded that all models that have such an algebra of constraints do not have this gauge symmetry. However, a few days later, we found that the contribution, on which we based our conclusion, can be compensated (we immediately withdrew our paper), but all our attempts to find another term that destroys the DI failed.
Such a result for a Lagrangian analysis is not compatible with the algebra of constraints given in their papers. Therefore, if the DI is confirmed for the Lagrangian, then the algebra, published in leading journals, cannot be correct. It turns out to be almost obvious (we should do this from the beginning!) that the algebra cannot be true, although to find the correct one is a long calculation. This Hamiltonian analysis is presented in our replacement http://arxiv.org/abs/1111.2647.
I personally found this event with our comment (see first version) very embarrassing, not because a mistake was made (no one is protected from this), but because the published results were blindly assumed to be correct without checking them. I disobeyed the “rule of procedure” – do not blindly believe any result (irrespective of who the author of the result is or in which journal it is published).
However many “rules” seem to be in use nowadays, and here is another story:
A few years ago, one PhD student told me that when he found two articles with contradictory results and he approached his supervisor for help; the supervisor opened SPIRES (now INSPIRE) and without reading either of the articles made a conclusion based on simple procedure “Article with more citations is correct one”.