I dunno....

When I look at the GFDL solution it is predicated on the ridge breaking down and the high retrograding NE by several hundred miles.

I find the CMC's solution rather interesting too - but not persuasive. It keys on a couple of VERY strong impulses north of the Canadian border that allow the ridge to buckle northward and while it takes Bertha almost to the Bahamas before it turns her, it then takes her poleward from there.

NoGaps has an even odder solution, with a massive high coming off the eastern seaboard, basically "pincing" Bertha between the Bermuda high and it, forcing it northward. But that one I buy - if the high shows up in 4-5 days and intensifies as it predicts.

The Wv loop gives some cred to this latter view - but IMHO the movement is too fast. And that, ultimately looks to be the key - if the connection is missed and the trough exits before Bertha gets where it can be influenced, the ridge will rebuild behind that and block polward movement.

So the slower and weaker she is, the more likely she remains westward-bound. Strength and especially speed likely gets her into the connection zone where she can be sucked into the gap as the ridge breaks, and out she'll go.

This far out I wouldn't even attach a probability on an impact on the ConUS, although were I in Bermuda I'd be a lot more nervous than being on the East Coast!