What's the right style when Release A credits composer X as composer of work W, and then Release B says that, no, in fact Composer Y composed that work W? Then backs it up with research?
Case in point: Work 21f580e, the Aria "Non posso disperar". One Release, Schirmer's "Twenty-Four Italian Songs and Arias for Medium Low Voice", includes a recording of this Work, and credits the composer as S. De Luca. The composer credit is based on an older book, Parisotti's "Arie Antiche".
However, in another Release, "26 Italian Songs and Arias", editor John Glenn Paton did research, and finds that Parisotti was wrong, the actual composer was Giovanni Battista Bononcini (1670-1747).
So, both Releases make claims about the same music composition, and those claims are different. They can't both be right. But both claims were actually made by Releases.
What do we put in the Composer Relationship on a work: our best information about what is true, or an accurate recording of what was credited, even if mistaken? And if two different recordings make different mistaken claims, what do we do about a Relationship between each recording and the Work entity?
This conundrum reminds me of Wikipedia's "Verifiabiilty, not truth" policy choice.