The Ritz Casino in London, March 2008. Three men walked in separately. They each placed bets on roulette table 14. Over the course of the evening, they won approximately 1.3 million pounds, roughly 1.5 million dollars. Then they left.
The method was straightforward: one man used a hidden laser scanner to measure the speed of the roulette wheel and the position of the ball. The scanner transmitted this data wirelessly to a second man, who performed calculations and transmitted a prediction back to the bettor. The bettor used this information to place bets.
Roulette is deterministic. If you know the speed of the wheel and the speed of the ball, you can calculate where the ball will land. The wheel will make a certain number of rotations. The ball will make a certain number of rotations. The physics is mechanical. Prediction is possible.
The Ritz Casino had no defense against this because the Ritz Casino did not know this attack vector was possible. The casino had security cameras. The casino had trained staff. But the security theater was designed to detect weapons, marking cards, and cash switching. Nobody was designed to detect wireless data transmission from a laser scanner.
Why This Interests Me
The Ritz attack is an example of system design failure. The casino's security model assumed a particular set of threats and optimized against those threats. The casino did not model the threat of wireless data transmission from a hidden device. This is not a failure of intelligence. This is a failure of threat modeling.
A good system design process includes: what are the potential threats? What is the likelihood of each threat? What is the cost of defending against each threat? The Ritz did not perform this process. They assumed roulette was secure because nobody had predicted the outcomes before.
The Detection and Response
The casino eventually noticed because the betting pattern was unusual. Three separate players, no connection to each other, all winning on the same wheel. The odds of this happening by chance are very low. The casino reviewed surveillance footage. They found no evidence of cheating.
But the casino suspected something was wrong. They reported the incident to law enforcement. Investigators eventually identified the three men and found the laser scanner. The case went to court.
The prosecution charged the men with conspiracy and cheating. But the legal question was: is predicting roulette outcomes cheating if you do not mark the wheel or interfere with the wheel's operation? The men argued they merely observed information that was publicly available (the spinning wheel) and performed calculations. This is not cheating. This is applied physics.
The court ruled against the men. They were convicted and sent to prison. The casino kept the money.
What Changed After The Ritz
Casinos updated their security procedures. They now check for wireless devices. They use radio-frequency detection. They have trained staff to identify suspicious activity. They limit the time during which bets can be placed after the wheel is spun. If the ball is released and the betting window closes, prediction becomes difficult.
The Ritz incident is studied in security circles as a case of threat modeling failure. A system that was considered secure revealed a vulnerability that nobody had previously considered.
The Larger System
From a systems perspective, roulette is a closed system. The outcome is determined by physics. Prediction is theoretically possible with sufficient information. The Ritz Casino had not implemented controls to prevent information leakage. The laser scanner was a method of obtaining that information.
A better security model would assume that prediction is possible and implement controls to prevent it: random wheel speed, random ball release points, frequent equipment changes. These controls are now standard in major casinos.
The three men did not invent roulette prediction. They just assembled the tools that made it practical. And they did so in an environment where nobody was watching for those particular tools.
