First of all, I think we focus way too often on the "average" or "median" behavior when trying to define how humans behave. In a lot of cases (optimizing a website or arranging a retail outlet), this makes perfect sense. In many areas, this is misapplied.
Imagine that a researcher did a study on dunking a basketball. They are very careful to take an appropriately-representative sample of a country's population, and they find that something like 0.5% (totally made up) of the people can dunk on a standard 10-foot basket. They find that lowering the basket to 9 feet allows about 5-10% to be able to dunk, and they provide arguments for why this 5-10% figure is an ideal number for maximal excitement of a basketball game. We can clearly see that this is a bad reason to use for lowering an NBA hoop to 9 feet. NBA players are extraordinary, and appealing to the average is silly.
In the same way, I think this could be regarded as a similar case. Clearly only a small percentage of the population is capable of writing good reviews. It is automatically narrowed to people who are capable of writing not only clearly, but also in a way that makes a somewhat dull topic more interesting. We should be able to further narrow "good reviewers" to that group of reviewers who are always honest in their review, those (perhaps rare) people whose reviews are not swayed by potential monetary gain. And there are certainly people like this.
Perhaps what we really need is a way to differentiate the best across all kinds of categories, something like an NBA of professionals. Currently we rely on various magazines and news sources to select the best reviewers, but this narrows the field too much and too arbitrarily, while irrelevant things like seniority and past results affect judgment of a person's abilities. Perhaps it's time for the business/information world to take something from the sports world.