One of the positive traits of Kei Kataoka's last book (Kumārila on Truth, Omniscience and Killing. A Critical Edition of Mīmāṃsā-Ślokavārttika ad 1.1.2 (Codanāsūtra), Vienna 2011) is how much of its methodological approach is explicitly offered to the reader to ponder about. Kataoka discusses his choices and describes his criteria, at the risk of raising criticism. This is apparent in the case of the interpretation of Kumārila, since Kataoka strongly upholds the view that commentaries may be used only as a third and last resource (after Kumārila's own other texts and Kumārila's predecessors' ones) and that one should be cautious in using them and always make readers aware of the fact that one is doing it:
A scholar who reconstructs a temple, if he has no choice but to use later materials, ought to explicitly mark the items so that other scholars will not be left in ignorance and confusion. Reconstructing an original idea of the seventh century through an interpretation in the eleventh century, for example, is anachronistic if unconsciously done, however labor-saving it is and however aesthetically attractive is the result (Part 2, pp, 108-9).I see Kataoka's point and I deeply appreciate the fact that he spells it out. However, I wonder whether an 11th c. paṇḍit is not often in a better position than a 21st century (Western) Sanskritist to understand a text. The paṇḍit has most likely an own agenda (but so does the Sanskritist, although s/he might not be aware of it —which is even more risky) and he might try to make the text say what he wants it to say, in order to justify a certain development within the school, for instance. Whatever the case, I am inclined to think that one should take commentaries seriously. One might disagree with them, but one has to show why they preferred a different interpretation (they wanted to make a Dvaita text into an Advaita one? there were no longer Buddhists around and hence they wanted to shift the polemics against another target?).
More in general, commentaries are more than welcome if what one is reconstructing is not the Ur-intention of a certain text, but rather the understanding of it by the tradition. In Kataoka's metaphor, not the 2nd c. b.C. temple, but the Middle Age church built over it.
How do you work with commentaries? Do you acknowledge you used them whenever your interpretations relies on them?
On acknowledging one's methodology, see this post. On this book by Kataoka, see this post (discussing his translation).