I find that one of the most interesting things around here. For instance, my GC ratings are strong, but they're never in like the top 10 or really high up. For instance right now I'm at #15. But if you were to look at total score per set, I would often feature in the top 3-5. So a person might look at ratings and see that I have a 1762 and compared to eyhung's 1992 might conclude that I'm not a threat. When in reality, on any given day and any given set, I might have scored almost as many advancement points as eyhung (I usually run a few questions margins below him, he has mad skills and I wouldn't compare myself at all to him). But that 230 points is suggestive of a much wider gap than actually exists.

So I think the answer is basically that the rating is generated by scores in the games that you play, it's irrelevant how someone plays across all 6 games. People that score in the higher percentiles of most or all of their games are probably scoring well beyond that in total score, as anyone who's looked at SAT, ACT, and GRE score charts can tell you. So I think that's what's happening. It's like it's capturing I'm a 75th percentile player at each point but then that is NOT informative in the final analysis of what kind of player I am in six game sets. Thus it's an odd feature of statistics that make this FIDE style rating actually not that informative a metric.

So with that being said, have I missed anything and is that the gist of it? Does it include Impossible because I would think that my Impossible would raise it because I've become quite good at that game after about 7 or so GCs.

Oh, I should probably add, informative at the extremes, but not informative towards the middle.

The rating is based on ELO, which in turn compares your scores against those of all others you played against in terms of win, tie or lose.

If your score is 1443 and everyone else's in that set is 1444, you'd lose a lot of rating points, especially if some of those players are rated much lower. I suggest you look up ELO yourself as the math can get quite complex when it comes to ranking against a field, not just one on one.

By the way, the GC (and the champion ratings) use a fixed K-value. No lower K the higher you rise. (This sentence makes sense once you looked at the definition).

I think Impossible does count, but I'm not 100% sure and would need to test (or anyone else can if you're not actively competing in the GC: Simply play only Impossible and see if your rating changes afterwards.)

OK, I'll be the guinea pig.
current global rating is 1627
I played only the Impossible. Will check back after lunch to see if there's a change.

My rating is still 1627. Not sure if that's definitive or not - if it was a large enough change to make a difference - I have no idea about the workings of that. My Impossible score was 425 in case that matters.

OK, this should complete the test. I played one plain (not starred) game, Television, in the 2nd set and managed to score 437. Now if I can just remember to check my rating tomorrow and see if such a small amount changed it.

Rats, there goes the 2nd test - I just played the Brain Teaser without thinking. Realized when it asked if I wanted to return to the home page or the global challenge.

Brain Twist gives you progression points, but I don't think it contributes to your rating. The ELO system needs to rank your score against the other players in your set, and Brain Twist is not scored that way - just a pass/fail for points.

Impossible doesn't affect the ratings: my rating was 1536 before and after (with a quite good result in Impossible).

I've just tried *only* Brain Teaser to check.

What Looney_tunes and JanIQ tracks with my failed test. My current global rating is 1613, a drop of 14 points after a poor score in one category and a perfect score in Brain Teaser.