I've noticed a pretty weird behaivor on the ladders - many people that are close to 50% wins have horribly low experience (elo rank), sometimes even as low as the rock botom 100 exp.  Basically the more games you play, the more your experience sinks if you're around 50% or less wins.  This is definitely wrong.  Anyone who is staying close to 50% wins should at least have close to 1000 experience.

Additionally, it seems that the loss of points is usually much greater than the win (probably causing behaivor #1).  I think it would be very beneficial to add the experience change info to every game so that we can analyze and help improve the ranking system.  Something like "starting exp, exp change" info for every player in a game.  It will allow to quickly see how well the teams were matched and how much experience was awarded.  I have a strong feeling that experience awarded for wins should be greatly reduced for winning against much weaker teams and probably increased for winning against better teams.  Probably the experience lost for losses should be reduced for near even matched teams - but these are all speculations without insight into the real numbers.  Also, a journal explaining the ranking logic would be very welcome (I hope you dont consider the elo system used a trade secret )

Finally, people with no games played should be unranked.  The ladders are populated with tens of thousands of people with no games played which horribly screws the ranking.

SD/GPG - I think these things need to be serously looked into along with the planned ranking/ladders cleanup and reporting bugs fixes.  In my opinion this would be a perfect opportunity to include the community in a meaningful way.  Providing the exp info should be something that is a breeze to make and I think you may be surprised by the quality of the discussions that will come out of it.