Title says it all. It would be cool to have the option to view the score plots in the user profiles by number of games instead of time. This would be a neat way to visualize progress.
<Comment deleted by user>
yes this would be interesting. I always wondered how often was one profile updated for rating. Was it at each game out come, or only daily (it is actually updated daily, i think). Internally, each game does it. but the profile?
But I would not say instead, but also. And this may not be working with long time controls and parallel games with overlap. or would it? Also, the rating variance is affected by the number of games in some time period (still not clear to me, if it would grow back after a while of not playing and when, but at my pace it is hard to probe by my own perturbations in behaviour).
In bullet somebody could test by just taking one day or 2 of a break. Or is rating characteristic time period for variance evolution independent of time control. i have no clue really how long of a break a bullet player should take to see the rating grow back a question mark (if at all), that shows my misunderstanding of the rating rules of evolution (glicko-2) perhaps, or its practical implementations in various time controls.
But still that would be an interesting angle. take time out of the picture.
But I would not say instead, but also. And this may not be working with long time controls and parallel games with overlap. or would it? Also, the rating variance is affected by the number of games in some time period (still not clear to me, if it would grow back after a while of not playing and when, but at my pace it is hard to probe by my own perturbations in behaviour).
In bullet somebody could test by just taking one day or 2 of a break. Or is rating characteristic time period for variance evolution independent of time control. i have no clue really how long of a break a bullet player should take to see the rating grow back a question mark (if at all), that shows my misunderstanding of the rating rules of evolution (glicko-2) perhaps, or its practical implementations in various time controls.
But still that would be an interesting angle. take time out of the picture.
<Comment deleted by user>
I am not sure that playing many games against weaker opponent and winning is going to increase your rating.
The point would be to see perhaps that different people learn from a certain amount of games differently than others, and perhaps same person could show different information gain per game (on average) over different time controls.
Or perhaps having both the time line and the game count line together could show some optimal pace of games (interval of non-play time to digest previous game). Something to see by having both that none alone can give. would not be the first time that embedding a question in a bigger set (or dimensions) would allow new patterns to be seen...
The point would be to see perhaps that different people learn from a certain amount of games differently than others, and perhaps same person could show different information gain per game (on average) over different time controls.
Or perhaps having both the time line and the game count line together could show some optimal pace of games (interval of non-play time to digest previous game). Something to see by having both that none alone can give. would not be the first time that embedding a question in a bigger set (or dimensions) would allow new patterns to be seen...
<Comment deleted by user>
I find that the deleted posts, could just take the line need in vertical space. Also, I have deleted posts of mine I did regret, for one reason or another. But here, I thought they were ok points to bring. Arguments don't have to be definitive to be made, in a discussion at least. (other threads don't even bother with the notion of arguments, opinions being enough). I did like the puzzle and graph experiment. And the suggestion to use home computer office suite (MS, or libre-office, or open-office, all have basic scripting ability), for limited personal data sets from own rating evolution per time, versus per game, to test the op display suggestion.
The question of winning streaks against weak enough opponents, was about where can improvement be seen better: time or game count, which while being in monotonic relation with each other, do have non linear regions, possibly plateaus with no games at all. Clearly not just game count would explain improvement, as learning does not occur that fast and needs time to actually happen while not playing even (sleep for example, or day periods when thinking of other things as well, that could even be the purpose or one consequence of the op question, to see such effects). But the point was made about having chosen to play only weaker opponents that could a false increase more visible if looking at game count versus time.
The question of winning streaks against weak enough opponents, was about where can improvement be seen better: time or game count, which while being in monotonic relation with each other, do have non linear regions, possibly plateaus with no games at all. Clearly not just game count would explain improvement, as learning does not occur that fast and needs time to actually happen while not playing even (sleep for example, or day periods when thinking of other things as well, that could even be the purpose or one consequence of the op question, to see such effects). But the point was made about having chosen to play only weaker opponents that could a false increase more visible if looking at game count versus time.
i have the feeling the plot is stuck for some days
This topic has been archived and can no longer be replied to.