Please also review our SoylentNews Moderation Guidelines.
As always, we are willing to make changes to the system, but please post examples *with* links to any cases of suspected mod abuse. It's a lot easier to justify changing the system when evidence is in black and white. I also recommend that users make serious proposals on changes we can make. I'm not going to color the discussion with my own opinions, but as always, I will respond inline with comments when this goes live, and post a follow up article a few days after this one
(Score: 1) by albert on Thursday May 21 2015, @04:52AM
The solution is non-linear mod points. Count the up mods and down mods separately. Before subtracting to find the result, apply non-linear functions such that the strength of up mods grows faster than the strength of down mods. Problem solved!
There are numerous functions that can do the job. Examples:
* Take the square of the up mods.
* Take the inverse tangent of the down mods.
* Take the log of the down mods.
* Take the factorial of the up mods.
Apply any one or two of those and the problem goes away. Choosing the easy one for example:
score = up*up-down