Board Thread:Game Discussion/@comment-26829117-20160522124555/@comment-27170954-20161020114518

after 10 days of using the restore technique i can report that it dosn't appear to make much of a difference.

in the first 10 days of the month i averaged 8.10 per day with a standard deviation of 5.75. had 2 GC 3 times and more than 10 (as high as 19) GC 4 times. i did not use the restore trick at all in the first 10 days.

For the second 10 days of the month i averaged 8.20 per day with a standard deviation of 3.88. No chage in the average to speak of, but noticeable reduction in variability. In this 10 days i initially got 2 ads 7 times before restoring (several times!) to get more. on those restore days i only got it above 10 twice.

so, my conclusion is the resotre trick does not have a material impact on the long-running average. i still think the model is the ad quantity is randomized around a target average. using the restore trick limits the low values in the sample, so the game will respond by reducing the high values to maintain the average over time. Admitedly, it is only 10 samples. but that's my theory and i'm sticking to it.

i'm going back to not using the restore trick. it is extremely annoying and doesn't seem to do anything. i'll probably get low values for a few days before the wider variability returns. i'll report if it doesn't behave as i expect.