Understanding RNG Certification at Online Poker Rooms

A poker site displays a logo. The logo says "Certified RNG" or "Approved by eCOGRA" or "GLI Certified." These logos claim that the random number generator has been tested and verified as fair. This is what the logo says. What it means is more limited.

An RNG certification means an independent laboratory has reviewed the code. The laboratory has run the generator through a battery of statistical tests. The generator has passed those tests. The generator produces output that appears random by mathematical definition. This is not the same as fair.

Fair requires that no one can predict the next card or manipulate the deck. Certified RNG requires that the sequence appears random statistically. A certified RNG could be seeded with a predictable seed. The output would still appear random in retrospective analysis. But the user who knows the seed could predict the next card in advance.

0x01What the Tests Measure

The standard test for RNG certification is the NIST test suite. NIST stands for National Institute of Standards and Technology. The tests include entropy checks, distribution checks, and correlation checks. The generator must pass approximately fifteen separate statistical tests. If it passes them all, the certification is issued.

The tests do not measure predictability. They measure whether the output looks random. A predictable generator with the right seed can produce output that looks random. The certification would still be issued. This is not deception on the certifier's part. The certifier is testing what they claim to test. But the consumer reading the certification might believe they are getting something broader.

0x02The Practical Meaning

A certified RNG means that the poker site is not using a simple formula to determine card ranks. The site is not using a broken algorithm. The site is using a generator that has passed scrutiny. This is good. It excludes the worst form of fraud: using a deterministic algorithm without randomness.

What certification does not exclude is subtle bias. A certified RNG could favor the house in ways that are not mathematically obvious. The tests check statistical properties. They do not check whether the distribution favors certain outcomes. A test for fairness would need to simulate millions of poker hands and compare the distribution of results against theoretical expectations. Most certifications do not go this far.

0x03The Seal Shopping Problem

Poker sites can choose their certifier. This creates a dynamic where sites shop for certifiers with less rigorous standards. A lab that issues certifications easily becomes the preferred lab. A lab with high standards issues fewer certifications and loses business. The market thus penalizes rigor.

No major certifier has actually failed an RNG test conducted by a reputable laboratory. The absence of failures might indicate that the tests are thorough. Or it might indicate that the tests are not rigorous enough to catch subtle manipulation. The history of gaming certification suggests the latter. Major casino companies have been caught manipulating RNGs despite having passed certification.

0x04What to Trust

The certification proves that the poker site paid for the certification. It proves that the generator passed statistical tests. It does not prove that the poker site has not rigged the generator in ways the tests cannot detect. It does not prove that the site has not implemented special rules to favor certain players. It does not prove that the site has not introduced side channels where insiders could predict the next card.

A better question than "is the RNG certified" is "can the player verify the result." A provably fair system lets the player download the code and run statistical tests themselves. The site publishes the seed. The player can confirm that the cards dealt actually came from the seed. Certification cannot provide this. Only code transparency can.