current math says that the linear value of the positive vote must negate the value of an existing positive vote
if a new math comes into place to reserve say any where between equal relative curation (around 25% - it can be set between the total flagged and negated flagged amount which is fundamentally 0% to 50%) value as curation, then so be it. Computation for a non linear or super-linear will be more intensive but it can still be simplified and done.
fundamentally, the answer is yes, it can be architect-ed and implemented.
Then serious coders have to code in the curation amount for flags.
I actually propose that Witnesses and Coders support a variable say between 5% to 50%, and let all the witness agree to set that amount much like the SBD to Steem ratio (Note the minimal 5% is token amount to give value to flagging which we do not today, and depends on the good whale to do so). This will spur the blockchain community leaders and witnesses to increase incentive to auto-correct when there are a multitude of bad-actors.
However, once again, witnesses supported by Steemit Inc (due to large bias of whale & Steemit ownership of Steem) has to agree to such a radical idea to fork.
Yep, and stinc prefers scammers to real users,...