Can a response paper agree with the initial paper, or must it disagree?
In general, 'comments' can be comments in support or comments in disagreement. Usually people tend to only submit comment when they are moved strongly to point out a disagreement.
But, as question specifically referred to Nature's "brief communication arising" (BRA) the answer is NO. The BRA notes that "Submissions should challenge the main conclusions of the Nature paper and contain new, unpublished data to support the arguments.". (i.e., pointing out errors, reinterpreting conclusions, coming to a different end result, comments to refute). Showing that there is an alternative method that is better than that presented in the paper doesn't seem to fall under that 'challenge' to the paper. (Unless perhaps you can write it as challenging some final conclusion that their method is the best possible ever? )
They offer the option to add extensive comments under the paper on the website. But then it will be difficult for others to be able to cite or even find your work. It sounds as if you should find a journal and submit your your method as improvement in XX algorithm/method, citing the original Nature paper.
For comments and criticism there is a brilliant new service of pre- and post-publication reviews called Publons. For example, a recent and very harsh criticism of a sensational paper in Nature. Each review is assigned a DOI, so others may easily cite your comments. Also, it is well covered with Altmetric, which provides quite a visibility to your comment. Yet, no official impact factor for you reply publication.
Every paper, in some way, is a response to everything that it cites. You can respond in agreement as much as you can in disagreement. An improvement is a positive response and you might couch it more in terms of a follow-on or follow-up without having to be negative about the prior results.
To be direct,
Not always
A response paper with respect to an initial paper, X, can come under one or more of the following categories.
- A method that can outperform what is prescribed in X
- An improvement over the method implemented in X
- A conflict of results with that of X
- A different approach to the problem stated in X
- A sequel to X by providing additional information or different analysis
From the above, only 1, 3, and probably 2, applies so as to disagree with the initial paper, X.