Private User, I hope this doesn't sound like hedging but I'm afraid it will
There's a difference between where I would sever it and where I think WE should sever it.
I think experienced genealogists know this problem. I might believe something but that doesn't mean I believe it has been proven. So, I deal with that and so does everyone else on Geni. The trick is to look for common areas of agreement using good genealogy standards and try not to push our own pet theories.
So, if it were just me I'd look at those connections and cut the ones that look dubious and where a quick 10-15 minute search doesn't turn up supporting evidence. The burden of proof is on the person proposing the connection. If the proof isn't there, then I wouldn't accept the connection.
What WE should do might be very different. I'm always open to ideas about that.