"Intentionally withheld" is maybe a little strong. We are trying to keep things simple in terms of what we're releasing, at least initially.
You can pretty much figure out where people's means and variances are, at least relative to those near them in the rankings, just by looking at the rankings. The players with more games tend to have lower variances, although the more consistent your play is, the quicker your variance drops, as well. So, for example, TSIYW has a higher mean to his ranking than I do, but I have a lower variance, which is why I am (slightly) ahead of him.
There are some dynamics in the algorithm that also govern speed of convergence. There's a variable that basically talks about how likely it is that a higher-ranked player defeats a lower-ranked player. The higher that probability is, the faster a player's variance drops (because the algorithm is more confident that each game result reflects playing skill and not just luck or matchup or whatever). We have tuned this variable based on our results so far, but we still have not done a sophisticated optimization of that convergence dynamics. That is in the works.
You can pretty much figure out where people's means and variances are, at least relative to those near them in the rankings, just by looking at the rankings. The players with more games tend to have lower variances, although the more consistent your play is, the quicker your variance drops, as well. So, for example, TSIYW has a higher mean to his ranking than I do, but I have a lower variance, which is why I am (slightly) ahead of him.
There are some dynamics in the algorithm that also govern speed of convergence. There's a variable that basically talks about how likely it is that a higher-ranked player defeats a lower-ranked player. The higher that probability is, the faster a player's variance drops (because the algorithm is more confident that each game result reflects playing skill and not just luck or matchup or whatever). We have tuned this variable based on our results so far, but we still have not done a sophisticated optimization of that convergence dynamics. That is in the works.