Green Baize Vampire was pilloried by Don and the clones over his recomendations, on how high to bet when you know that one of your first two cards will be an ace. The "funny thing" is that Don and mathprof then actually learned some things when it came time to go over recomendations on how much to bet when you know one of your first cards will be a ten. They were correct, and GBV wrong, when they evaluated the single hand varriance of a hand that is known will have at least one ten.
This goes back to the time that counting was first considered, with the first published material coming from Frank Garcia. He quickly noted the possibilities of endplay. But only endplay was considered in that it brought definite information about the end hand and definite playing recomendations.
Then came the Cantley, McDermot etc., group working out the first basic strategy and first methods to find "best stategies" based not on definite information coming from knowing which cards were gone, but RELATIVE information on the likelyhood of cards to BE DRAWN. Thorp developed this and WAS initially opposed by Garcia, until Thorp and Garcia made the test plays mentioned in Beat the Dealer.
Normal Counting is based upon the cards remaining to be dealt having normal distributions, being equally probable in coming up in the next draw. Shuffle Tracking changed that, making it clear the mechanics of which cards are mixed in a shuffle provides a limited "opening up of the closed box," in regard to analysis of the draws available as the pack is depleted.
I have been pilloried by Don for pointing out that the validity of his trip ruin is another example of opening the closed box (it is interesting that if you read Blackjack Attack he was actually aiming to do this when you consider that he limits his own discoveries---hence my comparison of him to Max Planc discovering Planc's constant and opposing the importance of his own discovery.) in that you actually can link the degree to which the path of your results twists and turns before reaching an endpoint, to the spectrum of end-point results. The closed box is thus a bit fuzzy and see-through if you will.
This is important because all sorts of exciting statistical opportunities exist where the normal understanding of the Central Limit Theorem blocks the proof of such concepts.
Another such example has arisen here with the post below by "alienated," (the handle is very very understandable considering the above) where he points out that some degree of shuffle tracking type information can derived without actually tracking a prior shoe, by simply considering that a segment can be considered, NOT just as a removal of cards from the ENTIRE remainder, but as a removal from the SPECIFIC REGION, of the unshuffled discards, that that segement was originally part of.
The common understanding of the CLT, of TODAY, however would tend to downplay this. I myself am unsure that alienated has everything worked out. But partial knowledge of cards removed from a specific region of a shoe DOES provide some insight into THAT region, especially if YOU ARE ABLE TO ADMIT, that we know more about the CLT and can limit the way prior understanding of the CLT censored (I mean exactly that) such possibilities.
GBV learned what happens when an opponent gets it. Will Don? Ever? And as noted as Max Planc was, he ended a sad and bitter figure not willing to admit the potential of his own discoveries!

