Sorry, I can't just keep this in any longer.
The models have NOT shifted this way or that way.
Models are just models - they estimate what will happen based on a variety of inputs, from readings taken in the upper atmosphere to the surface currents and temperatures.
So what's happening? When most people (including the talking heads on TV) are focusing on that big thick line that the NHC plots, and ignoring the 'cone' surrounding it. But that solid track line is plotted after looking at about two dozen outputs (models) from several different forecasting agencies. Each model has its own confidence interval.
It's a mistake to think that a hurricane will (or should) take the average path between these models. All models being accurate (they're not btw) the model predicting the most western course is just as likely as the one predicting the eastern-most course.
Here's how the NHC addresses the models:
After that time, Irma will reach the southwestern edge of
the subtropical high and begin to turn north-northwestward and
northward. This turn will occur, but the precise moment is still
uncertain, and that is why NHC emphasizes that nobody should focus
on the exact track of the center. The new NHC forecast was adjusted
just a little bit westward and is on top the latest ECMWF model and
the HFIP corrected consensus. In fact, these two aids are also very
close to each other.
Why does this matter? Because people keep staring at that solid line, and say "gee, now it looks like it's going to pass slightly to the east, I should be alright". That's a dangerous assumption to make. If you're inside "the cone" at least some models are showing the hurricane going right over you. It's the wrong way to look at these predictions.
Here's the tracks postulated by 20 different models as of noon today. Notice some still predict a path that stays over the Atlantic, others into the Gulf.