by John Robison
The god in the machine. That’s what Frank Scoblete calls the slot machine’s Random Number Generator (RNG) in his book Break the One-Armed Bandits. Like Zeus on Mount Olympus, the RNG is the arbiter of your fate at the slot machine. The RNG determines whether your spin will result in Megabucks, a bust, or something in between.
Nothing in a slot machine is more mysterious than the RNG. Slot machine manufacturers perpetuate and encourage the mystery by revealing very few details about how the RNGs in their machines operate. But the truth about the RNG is far more mundane than any of the legends that superstitious slot players have spread about it.
Before we look at how the RNG in a slot machine generates random numbers, we have to define what a random number is. Let’s try an experiment. As quickly as you can, think of three random numbers between one and 10.
Most people try to space out the numbers, so they pick one low number, one medium number, and one large number. Very few people pick three of the same number or three numbers in order. Yet if the numbers were truly chosen at random, one-one-one would be just as likely as one-two-three, which would be just as likely as one-four-eight and eight-four-one.
Our perception of randomness is that there should be no order to the numbers, so we avoid combinations that have a pattern. We also don’t pick combinations that have a meaning to us, like an area code. We eliminate many combinations that should be included and then unintentionally impose an order on the numbers by spreading them out.
Human beings are disturbed by randomness. We try to impose an order when we experience random events. As much as we may say we want variety and adventure, we take the same route to work each day and we do the same things in the same order when we get up each morning. We are creatures of habit, and we want the world around us to be predictable.
Randomness upsets our world order. We take comfort in cause and effect, so we try to find an order in random events and a reason for why things happen. We say that bad things always happen in threes. We say that a bad blackjack player at third base always makes the rest of the table lose money. We say that the machines are looser on the weekends or near the end of the month or when you don’t use your slot club card.
My dictionary defines random as“having no specific pattern, purpose, or objective.”Applying the definition to a slot machine, it means that each spin is an independent event. The outcome of one spin does not influence the outcome of another. If the outcome of one spin did influence another, there would be a pattern.
So, what is a random number? Is 7 a random number? How about 2,385,243,867?
A number cannot be random. Random is not a property of a number like parity (odd or even). A sequence of numbers, however, can be random. In a random sequence of numbers, there is no pattern to the numbers. One number does not affect the next number in the sequence. There is no relationship between the numbers in the sequence. Knowing the seventh number in the sequence, for example, does not help you predict the eighth. For that matter, even knowing all of the numbers in the sequence, so far, does not help you predict the next number.
The RNG in a slot machine really generates a random sequence of numbers, not random numbers. Let’s take a look at a few characteristics of a random sequence of numbers and how they apply to a slot machine.
One characteristic of a random sequence of numbers is that each number in the sequence is equally likely. For the slot machine, this means that every virtual stop is equally likely to land on the payline.
All electronic gaming devices are tested for proper operation and fairness by a state or independent testing lab. The testing labs also test the performance of the RNG.
A testing lab may test how well a machine’s RNG does in this characteristic of randomness by counting the number of times each number appears in the number stream from the RNG and comparing the count with the number of times they expect to see each number. Some deviation from the expected count is allowed–required, actually.
Randomness requires that the results not be perfect. If the RNG always generates the exact counts of each number expected, the lab would doubt the randomness of the RNG. There is a mathematical test for the deviation called the Chi-Squared Test for Goodness of Fit. If the result of the calculations in the test is too large or too small, the RNG will fail the test.
How do the testing labs access the output from the RNG? Devices with embedded computer systems, like a slot machine, typically have more than one way to control the computer in the device. The first way is the way a user controls the computer. For example, you control the computers in your car using the knobs and switches on your dashboard, the steering wheel, the shifter, and the pedals.
The second way to control the computer is the way a technician controls it. Your mechanic connects a test jig to a special connector in your car and he can see much more information from the computers than you can see.
We control the computer in a slot machine using the bill validator, the coin acceptor, the buttons, and the handle. The testing lab controls the computer in the slot machine by hooking up a test computer to a special connector, typically an RS-232 serial port, in the slot machine. The program running the slot machine responds to commands sent to it via this special connector. One of those commands is to send the numbers generated by the RNG back to the test computer.
Next time, we’ll look at more characteristics of random numbers and take a look at an actual RNG.