Sort:  
  1. you could use cheetah and similars as a flag.
  2. You could use community to spot non quality post. For example if i see a bad post upvoted by smartsteem i could flag it or post a special comment. Then someone reviews the case. Then we use machine learning from a series of such cases to spot them (semi)auto

Hey there. Out of pure curiosity, could someone get overbid so hard, or the volume so huge, that they get pushed back multiple rounds? also, if someone overbids a HUGE amount, which would normally take up the entire vote and then some, would you just refund the amount back?