Can Computers Simulate Real-World Conditions?

23 November 2001

By John May

A crucial debating point between card counters and proponents of "unscientific" methods such as card clumping and progressions concerns whether computers can simulate the real-world conditions gamblers play under.

The answer is yes...and no.

A computer can simulate a randomly shuffled game of blackjack perfectly. No conceivable difference could exist between a computer plucking numbers from some algorithm and a dealer plucking cards from a deck. Provided the variables are the same (52 cards in a deck, equal chance of each appearing) it does not matter.

The problem is that games of blackjack are not shuffled truly randomly in the real world. Many computer whizzes have succeeded in implementing many of the elements of shuffles, such as the nature of the riffle, into their programs.

The real problem with computer simulation is that ultimately everything has to be reduced to numbers--in fact 0s and 1s, the computer's own binary language. If you can't be 100% certain the figures you are allocating are accurate, you can't be 100% certain your simulation is accurate. With the shuffle process it is really impossible, for example, to simulate the level of muscular tension in a dealer's arm at the end of a shift, or even something as simple as where the player will cut, which is not truly random.

Nonetheless, that does not mean a computer simulation cannot provide a very useful indication of which strategies are and which strategies are not successful. A computer simulation can reflect reality probably greater than 95% of the time. So it is not correct to say: Computer sims are not always accurate; therefore, if the computer sim disproves my method, the sim must be incorrect. You would have to come up with a plausible hypothesis as to why this should be the case and produce contrary data.

I never trust computer data alone when exploring the viability of a gambling technique. I always back it up with a theoretical explanation and calculation as well as actual experience at the tables. But simulation is undeniably a useful tool no researcher should dispense with.