I hope that the BAMS and BAMM computer model tracks are the correct ones, but again there is a huge variance in the individual Computer model tracks.
Even at 42 hrs from now

BAMS has Earl at 31.1N 72.9W
BAMM has Earl at 32.5N 72.5W

Where as NOGAPS (US Navy) has Earl at 39.0N 72.9W (approx around 120Nmi from New York)

This is a huge discrepancy between the BAMS and NOGAPS tracks at 42 hours of over 290 Nmi

The problem with these fast moving Hurricanes (as they track northward or northeast) is that it would appear that the consensus of computer model tracks shows that only approximately 12 to 18 hrs warning is available for high density urban populations on the east coast such as New York. You will have to wait a good 6 hours after Earl has passed Cape Hatteras to get a more reliable indication of whether Earl has begun to vector away into the North Atlantic.

When do you Bug out from New York?

e.g. 18hrs out with 10% probability of a direct hit
or 12hrs with a 50% probability or 6hrs with a 90% probability.

This is going to be a close one. The lives of many thousands of unprepared people could be reliant on a computer model.