I understand what you guys are saying with regards to average counts in a shoe, but I look at it a little differently.
When you walk up to a partially dealt shoe, it doesn’t really matter what the average count of the used cards is because you are not considering them. You are treating it like a freshly shuffled (undealt) shoe. You are ignoring the cards that have been dealt so it doesn’t matter what their average value is. You would start out at a RC of zero for balanced systems or the IRC for unbalanced systems and assume that none of the cards have been dealt yet. Basically we are treating it like a fresh shoe with worse penetration.
You could use the “average count” if you wanted to but it would be less accurate. Using an unbalanced system you could assume that the dealt portion of the shoe is of average value and just add the imbalance to your IRC. This would be the same as using a balanced system and assuming that the count of the dealt portion is zero. You would then be making up for the lost penetration by assuming that the unseen portion is average, but you cannot expect that to be very accurate.
Simply speaking, when you approach partially dealt shoe you have two options:
1) Assume that the dealt portion is average and continue counting from there.
2) Ignore the dealt portion and start from scratch.
I think that most players choose the latter in order to avoid the errors associated with making assumptions. If only a few rounds have been dealt then you might be safe choosing the first method, but the errors will get larger as the size of the dealt portion increases.
-Sonny-