Dear Sergey,
You are right that rating procedure is at least tricky. Because it is only partly hided (present score not known) many participants try to get a better position by low rating (1) the essays with a higher score than theirs. This is a common phenomenon and it is getting worse during the last days and specially for the essays close to the first 35 (in order to get into the heaven...). A simple solution would be to hide community rating completely, until the end of rating period, so that most of various rating strategies will be useless.
After all the benefit that any author can get participating to this contest is not the rating results but the discussions on the different views the essays offer and the opportunity to publish your thoughts without the filtration of a peering council (for authorization according to the established views).
Best wishes to everybody, Ioannis