Hi, im a long time lurker,
This post is old, and my opinion doesn't hold much weight as compared to the senior members of this board. But I would like to throw in my two cents.
If Derren Brown (DB) did memorize the cards seen he would intuitively be 'counting' the cards.
However, all CCs know that no matter your knowledge of the deck composition, you do not know the order of the cards; DB could not have possibly known the order of the cards.
Hence, even if DB knew all the Fours were left in the deck, he could not have known if a Four was the next card. Mathematically speaking, by hitting on a 16 v 6, DB was incorrect-he was lucky. If you watch the clip, the dealers next two cards are 9 and 10, either of those cards could have been DB's Four card, but instead he caught the Four.
DB's memorization of the deck is an inaccurate and imprecise way to play blackjack. Although it is still a cruder method of AdPlay with a subconsious application of Indicies; DB will still lose money in the long run.
Another prime example of why DB will lose overall but won due to variance is this.
There is 52 cards remaining; there are 10 Sixes and 10 Fives; 32 Tens left. For a HiLo count this is in a 4D game with no cut card: (80 lows minus the 20 lows left) minus (80 tens minus 32 tens left) = RC 12; converts TC12 (12/1). Using indices of 16 v 10 as TC0 and 15 v 10 as TC4.
In this scenario, DB would bet big because he knows theres lots of tens left in the deck. Essentially doing a RC to TC conversion with bet progression. However, during play DB gets 16 v 10 or a 15 v 10, what does he do?
If he knew there were only Tens, Fives and Sixes left, he will hit, to either improve to 21 or 20 to beat a highly probable Dealer 10 10.
BUT, mathematically this is incorrect. In fact DB should SIT on 16 v 10 and 15 v 10. DB is more likely to BUST if he HITs. The fact that the dealer can get a 10 on the next card is irrelevant; because if the dealer gets anything OTHER than a 10, the dealer is MORE likely to BUST on the THIRD card by being forced to draw to 17. There is approx 38.5% of a Five or Six draw or 61.5% of a Ten on 1st draw. While there is approx 44.5% chance 2 Fives, 2 Six, or combination of Five and Six occuring of the 2nd draw. In these both scenarios, the Dealer will have a higher chance of busting on the second draw.
Therefore by not BUSTING FIRST, DB is more likely to WIN.
In conclusion, the mathematical foundations of BJ is more valuble than memorization of the cards and by CC you maintain a better accuracy and decision than 'fuzzy' logic employed by DB.