I might be wrong but I think the way GOG decides who has the most helpful review is simply how many people say it is helpful instead of the overall yes to no ratio. Overall I think this is a bad system because it ensures the oldest reviews are always the most visible.

I was looking at the reviews for Diablo and the current most helpful review has (474 of 1603 users found this helpful). While the next most helpful has (416 of 464 users found this helpful). So even though more than 66% of the people who read the first review said it wasn't helpful while only 10% of people who read the second said it was unhelpful; GOG will favor the first review as the most helpful. There has to be a better algorithm, but with the state of the rest of the site since the update this is probably not even something they will address this decade.

Just something that's been rattling around in my head.