TL;DR: How does the current behavior improve the user experience compared to the behavior I expected?
------------
I think I understand, but just to be sure, I'll say it in my own words.
In each layer, there is exactly one pixel per "point." e.g., in a 200x300 layer, point (0,0) has exactly one pixel, point (1,0) does too, and this is true all the way to point (200, 300).
If you can't see the pixel, that is because it is transparent, but it is still there.
When you copy a selection, the transparent pixels of that selection are copied, too
When you move the selection, the transparent pixels will overwrite the pixels that existed in that location, even if the previous pixels were visible.
If I'm understanding that correctly, I'm glad I understand and this will prevent confusion for me in the future. I now see the rationale behind the behavior and appreciate the consistency with the way pixels work on layers in general. I can see now this isn't a bug but working intended. That said, I'd now like to change my "bug report" to a "feature request," if possible?
I'd like to ask, is this a good user experience compared to the behavior I expected? Namely, I expected the transparent pixels of my paste buffer/selection to leave the non-transparent pixels the way they were. Instead, they overwrite the visible pixels. If the current behavior is a good user experience, can someone explain to me why? I can't think of a situation where transparent pixels overwriting non-transparent pixels would be useful.