Ratings "Settling"


#1

I greatly enjoy CC, in fact I love it, and it’s helped me improve quite a bit. However, I’m pretty much ignoring ratings for a while, as they are pretty erratic. Some easy problems have a high rating and some rather tough ones have a low rating. So my own problem rating surely doesn’t express my (lack of) skill.

That’s probably because the ratings are far from reaching anything near steady-state. Some of these problems have had only two or three attempts— hardly enough to settle down to a realistic rating. Obviously the K factor (or whatever it’s called in the Glicko method) is high.

I guess the best solution is to attract a few thousand more players to the site and give the problems a real workout!


#2

Thank you for the kind words. Indeed there is a lot of noise in the ratings. It’s a little better every day, but yes, we need a lot more people to help “sort” the puzzles. It’s one of the reasons I haven’t introduced more puzzles yet. I know some heavy users are seeing a lot of repeats which stinks. But there are some 3,000 puzzles that have never had a rated attempt. And adding more puzzles spreads users even thinner.

There are a good number of people, like 200-300 a day, that play the puzzles but don’t have an account. And because they don’t have an account, they don’t have a rating, and they don’t don’t affect the puzzle’s rating. I’m working on some site layout changes and some features that will hopefully encourage people to create accounts. But development has been slow recently.

Thank you again for the encouragement. I’d love to attract a few thousand people. Any ideas on how to do so?


#3

As to how to attract a few thousand new, registered players, I wish I knew!

I am not sure that repeat problems are as bad as all that. The way the game Go is taught in Go academies includes doing graded series of problems over and over again until the patterns are completely ingrained and can be instantly and reflexively solved. I think this applies at least in some degree to checkers.