Hi Paul,
I'm not a DNA expert, but I'm learning more and more as I go along. My understanding is that your 2 bullet point statements are correct. And it makes sense to me that a line had to be drawn somewhere to say that additional evidence is needed to support DNA confirmation for more distant relationships. So here at WikiTree we have triangulation as criteria for >3C auDNA confirmation.
My oversimplified view of triangulation is that we're essentially trying to assign ancestors to a particular segment. If we have 3 or more potential distant cousins (>3C) that all share a common segment with each other and they all have evidence of the same potential MRCA(s) in their trees, then we are likely correct in attributing that particular segment to the MRCA(s).
If 2 potential distant cousins that share a common segment, also share another common segment, that may be a stronger indication that the 2 individuals are actually related. But IMO, it doesn't provide any additional evidence to support that a particular candidate ancestor is indeed the correct MRCA, which is what we're really trying to do at this point.
Also keep in mind that while the MRCA(s) are only 4 generations away from 3rd cousins, the MRCA(s) are 5 and 7 generations away from 4th cousins 2x removed.
In this case I believe the relationships should be marked as "confident" vs "confirmed with DNA" (assuming that you have other sources that support "confident" for each of the parental relationships in question).