That's as I remember, flame war and rapid short volatile exchange detection leading to down ranking weighting .. I had the impression that explicit [flagging] was the visible mechanism for explicit core user disapproval (for inclusion, not necessarily of specific content) and kept separate from the invisible hand of algorithmic weights.

The repeated takeaway I see is that they are still open for discussion on aspects of all that:- https://news.ycombinator.com/item?id=39233617