The only "objective" way would be to use a computer algorithm, which we know generates "unusual" results at times, even end of season.
I think we need human "intervention" when that happens. Take three decent computer polls and average them to get the top 4, or whatever, might work most of the time. By "work", I mean generate a reasonable output, but then, the committee does that anyway in my view.
Perfection is the enemy of good enough ...