Unproven citations

I’ve seen some Emyoulation’s posts about unproven citations and I’ve tried them.

Are they only researches citations (wich could stand into notes)?
Does they really unprove something, if it is possible to do so?
To what for are they usefull?

I’ve tried them for couple weeks now, combining proven and unproven citations, attributes and report notes to explained conclusions. Evidence Explained is not very verbose about them. I know false-fail testing in IT to prove a test will be correct but I’m looking to learn more about that practice in genealogy context.

Do you use them and how, in general and with Gramps? Do you know any literature available on the net on that subject?