If you're asking about visualizing data, and part of the reason for the question is that you're generating the image (or sound in the case of auralization) from an open API. (vs. downloading it, having a chance to prep it for your software, and then use it).
I could also see questions being valid when they're asking for clarification or help with data that may have been released with poor documentation.
"Best [representation] to compare [data]" is never answerable without knowing who the audience is and what questions they're trying to answer. Scientists may have their standards that they're used to, but it may be inappropriate for a 5th grade class. Many visualizations are themselves a form of bias and are used to try to influence people. (see Darrell Hoff's seminal work).
I'm torn on questions about asking for norms for visualizing a given type of data, as we might to a case of them being too localized. I'd likely have to see individual questions before making a judgement.
I would personally be okay with questions about the possibility of bias in a visualization (although, there's then no '1 correct answer' that so many people on SE sites have been fixated on over the last couple of years). I guess you could ask for the worst bias in it to get around that rule, if they're going to be accepting of 'best' questions.