Also, I think it should be noted that different styles of beers have different difficulties to make and expectations. I think most would agree that an average big Imperial Stout should be a pretty good beer because it is so big and powerful that it might overcome shortcomings in fermentation, recipe, balance, etc. However, an oatmeal stout or sweet stout requires more balance and finesse to make well, so in that respect maybe the judges will give credit to a sweet stout that scores a 35 when all the others were in the 20's over a RIS that scored a 38 over a field in the low to mid 30's.
Another alternative is if 2 beers get similar scores 34 vs 38, but 1 beer got better scores for flavor and overall impression, whereas the other was a point or two better in mouthfeel, aroma, and appearance. In my mind, the beer that tastes better should be the winner even if it isn't as technically perfect, especially when it pertains to inproper carbonation. Undercarbonating a beer will drastically diminish aroma, appearance and mouthfeel which can take 5-10 points off of a beer that was otherwise better. If you have the perfect recipe and process but mess up carbonation, your score takes a huge hit for something that is sometimes an afterthought for me.