Generallly my approach if engineers disagree significantly on a timeline is to multiply the worst case estimate by a large factor, on the assumption that odds are good it means neither one has a clue what the actual complexity is if they don't manage to get to rough agreement...