It's an interesting idea, but how would a user make that decision on who to trust? As in, what 'things' could we look at to make that call?

Doesn't ACME+CT solve this threat?