2nd-order Qualia and what-it's-like problems
It may indeed be true that we cannot know the what-it's-likeness of another person's experiences, what I shall call "first order" qualia. But does that mean we cannot know the what-it's-likenss of the relationship between first order qualia? There might be some reasons to think that we can indeed know such relationships, which I shall call "second order" qualia. Essentially, these are qualia of qualia.
Working with one perceptual modality as an example is helpful. Consider aural perception. I don't know what's it's like for you to hear C#. I don't know what it's like for you hear A#. Does that mean I don't know what it's like to say one tone is different from another -- i.e. that C# is different from A#? Suppose that I identify patterns in our verbal reports: (1) one tone is affirmed as different than another tone. (2) One tone is affirmed as the same as another tone, or (3) as higher, or (4) as lower.
Suppose I mathematically quantify predict (unexamed) tones, and do so even w/o a single exception. This would be justification that I have indeed understood the what-it's likeness of your qualia, at least of the 2nd order qualia.
Granted, for example, I don't understand "your C# perception." Granted, I don't understand "your A# perception." But it hardly follows that I don't understand "your C# perception in relatation to your A# perception."
Consider the following analog: Object A is liftable by me. Object B is liftable by me. Thus, Object A in addition to object B is liftable by me. One does not have to work with a weight set very long to see the flaw in this reasoning.
I think skepticism over qualia is over rated as a philosophical problem.