Sure enough, there are two different issues here. The scoring would be objective based with characters being judged favorably for performing suitable tasks and penalized for unsuitable. The list of task rewards/penalties, per character or type, could even be public. Such a system would indubitably reward those more experienced with the system better than newcomers but most things in life are like that.
The referees would be needed to deal with those players deliberately abusing the system in whatever way they might dream up. If this sounds a little generic it is because the system is imaginary and there are no specifics to speak