04.11.2022, 18:59
Something very similar has already been suggested in the past. I did not like the idea back then, and I still do not like it now.
The rating system already has some major defects, and I believe such changes would make it worse.
A few statistics. The Portal was created in 2008. In the beginning, a puzzle was rated by about 180 solvers on average. This number has dropped over the years, and currently the average number of solvers lies around 40. Since the number of Portal users has exploded recently, for any two given puzzles it is now far more likely that the groups of solvers are essentially disjoint. Unless the users have the same standards for the ratings they give (and I am struggling to find a reason why that should be the case), the ratings are far less comparable than they used to be.
In 2008 the average rating for puzzles was around 4 stars (which corresponds to 75%). This average has also increased consistently, and right now the rating average lies above 4.6 stars (90%). Most ratings that today's puzzles receive are 5-star ratings, which means that removing the best ratings would hardly make a difference. But removing the worst ratings would serve to increase averages even further. This makes no sense to me.
On the contrary, we would filter out the few users who are still using the full rating spectrum. As far as I am concerned, lower ratings are still legitimate ratings. In fact, there is a significant number of solvers who only give 5-star ratings and nothing else; one could even argue that those users are actually the ones who are abusing the system. For some reason, nobody is complaining about them, though. As I see it, people do not want fair ratings; they only want good ratings. And I see no point in manipulating the rating system to satisfy this desire.
The rating system already has some major defects, and I believe such changes would make it worse.
A few statistics. The Portal was created in 2008. In the beginning, a puzzle was rated by about 180 solvers on average. This number has dropped over the years, and currently the average number of solvers lies around 40. Since the number of Portal users has exploded recently, for any two given puzzles it is now far more likely that the groups of solvers are essentially disjoint. Unless the users have the same standards for the ratings they give (and I am struggling to find a reason why that should be the case), the ratings are far less comparable than they used to be.
In 2008 the average rating for puzzles was around 4 stars (which corresponds to 75%). This average has also increased consistently, and right now the rating average lies above 4.6 stars (90%). Most ratings that today's puzzles receive are 5-star ratings, which means that removing the best ratings would hardly make a difference. But removing the worst ratings would serve to increase averages even further. This makes no sense to me.
On the contrary, we would filter out the few users who are still using the full rating spectrum. As far as I am concerned, lower ratings are still legitimate ratings. In fact, there is a significant number of solvers who only give 5-star ratings and nothing else; one could even argue that those users are actually the ones who are abusing the system. For some reason, nobody is complaining about them, though. As I see it, people do not want fair ratings; they only want good ratings. And I see no point in manipulating the rating system to satisfy this desire.