It might be simpler to say that the interaction between the reviewer's motivation to believe that 50% of everything is crap and the fact that 90% of everything is crap has emergent effects.
The major one is that bad qualities that randomly happened to occur more commonly in products that have previously made the 50% cut become misindicators of quality, distorting the evaluation of a constantly drifting subset of products that have or lack those misindicators. This occurs asymetrically (between misindicators of positive quality and misindicators of negative quality) because there are more products misidentified as being good, therefore it's relatively far more common for particular bad qualities to randomly be more associated with works above the median then the opposite. Works that fall below the median in a world of 90% crap would have very few good qualities at all.
The major one is that bad qualities that randomly happened to occur more commonly in products that have previously made the 50% cut become misindicators of quality, distorting the evaluation of a constantly drifting subset of products that have or lack those misindicators. This occurs asymetrically (between misindicators of positive quality and misindicators of negative quality) because there are more products misidentified as being good, therefore it's relatively far more common for particular bad qualities to randomly be more associated with works above the median then the opposite. Works that fall below the median in a world of 90% crap would have very few good qualities at all.