OK, well hopefully you understand now! This part is key:
> Now, it seems like you were saying that the text layers are just the pixels from the page moved up to a new layer. I said that I think that's surprising.
That's indeed all it is. It may be surprising, but that's how it works. Absolutely nothing about the pixels are changed as part of the process of that text layer separation. Later when saving the final PDF there's normal lossy compression, same as any JPEG, but the layer process actually preserves the text edges better than JPEG.
So I do hope you're satisfied now that everything about this is just normal image processing, and that nothing has been destroyed, there's no "missing evidence" or anything. And that the whole idea that the layers imply some kind of manipulation or forgery is false. What you're seeing is just the scan itself, saved (presumably) at a lower resolution and with the normal image compression scanners produce. The layer separation process is completely non-destructive and doesn't manipulate pixels at all.
> Now, it seems like you were saying that the text layers are just the pixels from the page moved up to a new layer. I said that I think that's surprising.
That's indeed all it is. It may be surprising, but that's how it works. Absolutely nothing about the pixels are changed as part of the process of that text layer separation. Later when saving the final PDF there's normal lossy compression, same as any JPEG, but the layer process actually preserves the text edges better than JPEG.
So I do hope you're satisfied now that everything about this is just normal image processing, and that nothing has been destroyed, there's no "missing evidence" or anything. And that the whole idea that the layers imply some kind of manipulation or forgery is false. What you're seeing is just the scan itself, saved (presumably) at a lower resolution and with the normal image compression scanners produce. The layer separation process is completely non-destructive and doesn't manipulate pixels at all.