Isn't it likely that that one's already in failure? The 2/3 convention delegate threshold isn't itself what's significant ...

Like any mechanical method, you can't go willy-nilly swapping your criteria. You'd need to back test to be sure the substitute works the same. I agree it's a measure of party unity, but the measure for this particular formula is the objective vote of delegates at the convention. Your suggested replacement is a more subjective measure. Coming from a more scientific background, Lichtman was looking for objective measures to use in his model. The more objective you can make it, the more you are not going to be blown around by the fickle winds of emotion. (As in the change from 3 pm on June 27 to 11 pm on the same day.)

But even if that particular measure is wrong this time, that would be the 5th false answer and the model would still predict Biden as the winner if he continued to run. Replace him, and you're now at 6 false answers and the model predicts Trump as the winner.

--Peter