I like the general idea, but 10% rep for a single downvote seems too stiff a penalty.
While I generally like the idea of rewarding people for explaining their downvotes, I’m not sure how to make the mechanics work. What keeps someone from simply typing “lkjsd jlksj dfklj sdlkj ytiu yoij we lm” as the reason? How would the system detect fraud of #2 looking like #1?
Good downvotes should be encouraged, as you are trying to do with #1 and #2.
Now consider the likely quality of anonymous versus signed downvotes. Those willing to publicly stand behind their downvotes are much more likely to have a good reason for them.
Some people here argue that disallowing anonymous downvotes will reduce downvotes, and thereby reduce an important feedback mechanism. But, consider the quality of such lost votes. These votes are cast because:
- Not really sure, doesn't want to look like a fool for being wrong.
- Revenge for downvote or close elsewhere.
- Doesn't want to spend the time explaining.
- Afraid of revenge for disagreeing.
Properly addressing the problem of revenge downvotes eliminates #4. #2 is clearly something we don’t want. That leaves #1 and #3. Some of those might be legitimate, but how much weight do you really want to give those compared to the expert that explains what is wrong and puts their reputation behind their conviction?