People in the gaming world often ask if the results that random number generators (RNGs) produce really random? There are those who question how do these RNGs work. But if you go down and dirty with the details of generating and programming an RNG our average player will have to take a crash course in programming. But not everyone has time to study the really dark details of RNGs. At best we can ask if RNGs are really reliable and honest.

There are actually three types of RNGs that we'll usually stumble upon. The first one is the truly random RNG, next is the quasi-random RNG, and finally we have the pseudo-random RNG.

Truly random RNGs really produce unpredictable sequences in which we will not find any patterns at all. The only drawback to these truly random RNGs is the fact that they just can't be used in every day life. Truly random results can only be generated using a phenomenon that naturally occurs in nature. An example of this is the decay of isotopes which can be recorded and fed to a computer and thus producing a truly random RNG.

When one initially studies how to make an RNG, a wannabe programmer will initially use textbook algorithms to produce an RNG. What this simply means is taking the simplest set of logical instructions that can be given to a computer and then produce a quasi-random RNG. The problem with quasi-random RNGs is that they tend to show a pattern in the results.

Let's say you measure the results of a quasi-random RNG when it mimics dice rolls. At first it would seem to produce random results. But that will usually fall in the short term, but by measuring the results in the long run you will then see patterns arise in the sequence of dice rolls the RNG produces. This isn't totally reliable if you're going to use it to produce games like backgammon, blackjack, or roulette since people can find a way to guess the results made by the RNG.

What we have today is a compromise between these two RNG models and have come up with what is called a pseudo-RNG. These are by far the most widespread model used today. Statistics show that the results made by this type of RNG is statistically independent. Tests also show that they are produce uniformly distributed results. Thus we now have RNGs that can be used day to day.

Today's RNGs are subjected to statistical tests to check if they produce results that are up to standard. That includes a certain percentage of the results should let players win. The percentage of wins varies from state to state (e.g. required 75% to 90% winnings).

In the end it is all a matter of trust knowing that random number generators are subjected to routine testing and that they are certified to comply with standards set by the law.