I know, I understood what is being proposed, but you probably didn't get my point. Let me explain.
Let's take yourself and me as User A and User B respectively in this example. So you and I had some argument in the past, and since then maybe you dislike me for whatever reason best known to you based on the argument we had. However, I'm not an obnoxious, rude, or bad forum member in general. When such a system is implemented, you decide to give me "obnoxiousness" points based on what happened in the past. So, this happens:
User A (You) who has 0 obnoxiousness and more than 1000 Merits gives User B (Me) obnoxiousness.
My "obnoxiousness" rank gets to -3 because you have 0 obnoxiousness yourself which gives the person -3 points when you give them obnoxiousness and it is multiplied by 1 since you have more than 1000 merits. This is based on the explanation from @apogio above.
Now, if a new user looks at my profile, does some research to find out what "obnoxiousness" is for, and finds out it means the user is rude or bad, he would have that image of me. Does that sound or seem fair at all? I don't think so, because, even if we had problems, I'm not a rude forum member or a bad person in general.
Besides, some might even do it out of jealousy or other emotions without even having any personal grudges or a history of arguments, etc.
As User A, I will have a counter under my name like so:
obnoxiousness: -7.5 (9)
the number in () will portray the number of users who have given user A an obnoxiousness feedback.
1. The larger the number, the more serious the accusations are.
2. The final obnoxiousness level can scale more rapidly if the number in () is larger, or more slowly if it's smaller. Thus, the number in () can be a multiplier as well.