Tommy Glenn Carmichael: the slot cheat and the industry's response

Tommy Glenn Carmichael was running a television repair business in Oklahoma when, according to the usual account, a friend showed him a device for cheating older slot machines in 1980. Carmichael had the electronics background to understand what it did and why, and he quickly became skilled at using it.

He was arrested within a short time and served a prison sentence in the early 1980s. By the time he was released, the machines had changed: casinos were replacing mechanical slots with electronic ones controlled by microprocessors and optical sensors. The old device no longer worked. Carmichael's response, in the version of events he later gave in interviews, was to buy machines, take them apart and work out how the new designs could be fooled.

Why his case is remembered

Many people have cheated slot machines. Carmichael is remembered because he kept adapting as the technology changed, and because the devices he built became well known to casino investigators and regulators. His methods are now obsolete, which is also why they can be described at a high level: the machines they targeted have largely been retired.

Later accounts say he eventually turned to designing equipment meant to detect or block the kinds of cheats he had once used. How far that work went commercially is less well documented, so it is best read as a reported postscript rather than an established second career.

The details of his career, including exact dates and the number of arrests, vary between sources, and much of the colour comes from his own later interviews. This page sticks to the broad, well-reported outline. For other profiles from the same period, see the famous gamblers overview and the casino knowledge topic hub.

Carmichael's devices did not change which symbols appeared on the reels. They attacked the part of the machine that counted out coins. That distinction is the key to understanding them.

How coin-era payouts worked

Older slot machines stored coins in a hopper. When a win was due, a motor pushed coins out one at a time, and a sensor counted them as they left. When the count reached the amount owed, the hopper stopped. In early electronic machines, that counting was done by an optical sensor: a light beam that each coin broke as it passed.

The light wand

The light wand, in broad terms, was a small light source that interfered with that optical sensor. If the sensor could not register coins leaving, the machine kept paying. Carmichael reportedly developed it in the 1990s and used it on machines across Nevada before it became known to investigators.

The monkey paw

The monkey paw, in broad terms, was a bent rod with a length of wire attached that could be pushed up through the payout chute to reach and trip the mechanism that released coins. It was used on machines where the sensor design made that possible.

Why the payout was the weak point

By the 1980s, the decision about whether a spin won was made inside the machine's electronics, which a person at the front of the cabinet could not reach. The payout hardware was different. It was mechanical, it moved real coins, and it trusted a single sensor to say when to stop. A cheat did not need to know anything about the game's odds or its random number generator; they only needed to confuse the counter. That is why the most successful cheats of the coin era went after the hopper rather than the reels.

This is a historical description, not instructions. Modern machines are ticket-based, heavily sensored and monitored, and possession of such devices is itself a crime in major gaming jurisdictions.

Carmichael's devices were effective because casinos, in that era, relied on the machine to count its own payouts. The industry's response came on several fronts at once.

Regulators and the law

Nevada's gaming laws make it a crime to cheat at a gambling game, and a felony to use or even possess a device designed for cheating. The Nevada Gaming Control Board's enforcement agents investigate slot cheating, and people convicted of it can be added to Nevada's List of Excluded Persons, which bars them from licensed casinos. Similar offences exist in other major gaming jurisdictions.

Manufacturers

Slot makers redesigned the parts Carmichael targeted. Payout chutes were changed so tools could not reach the mechanism, and coin-counting sensors were redesigned so outside light could not blind them. Machines gained more internal sensors and event logs, so an abnormal payout pattern could be spotted and investigated.

Ticket-in, ticket-out

The largest change was the move to ticket-in, ticket-out (TITO) machines, which spread through the late 1990s and 2000s. A TITO slot prints a barcoded ticket instead of dropping coins. With no hopper and no stream of coins to count, the entire class of payout-mechanism cheats had nothing left to attack.

Surveillance

Casinos also expanded camera coverage and trained surveillance staff to watch for the body movements associated with device use. Modern systems link machine data to video, so an unusual payout can be reviewed within minutes. For how that works on today's floors, see casino surveillance.

Carmichael was one of several well-known slot cheats of the late twentieth century. The others help show why the industry redesigned so much.

Attacking the reels

Some cheats worked on the reel mechanisms or the logic that decided outcomes, rather than the payout. Dennis Nikrasch is commonly described as the most successful of these, reportedly using stolen or purchased machine components and teams of accomplices to force jackpots on specific machine models in the 1980s and 1990s. The details come largely from later journalism and prosecutions.

Attacking the software

The most serious case involved an insider. Ronald Dale Harris, a software engineer for the Nevada Gaming Control Board, used his access to hide code in gaming machines and to exploit a weakness in a keno game. He was caught after an associate's large keno win in Atlantic City in 1995 drew scrutiny, and he was convicted and imprisoned. His case is a key reason regulators tightened controls on who can access and verify machine software.

What they had in common

All of these schemes targeted the gap between what a machine was supposed to do and what it actually checked. Carmichael fooled the counter. Nikrasch fooled the mechanism. Harris hid logic inside the software. In each case the response was the same kind of fix: more independent verification, more logging and fewer places where a single component was trusted blindly. That principle is also at the heart of how random number generators work in modern games.

Slot cheats existed because honest slot play has no winning strategy. The maths shows why.

A slot's return to player (RTP) is the average fraction of each wager it pays back over the long run. On a machine with a 92% RTP, the expected loss is 8 cents per $1 wagered. At $1 a spin and 600 spins an hour, a player wagers $600 an hour and expects to lose about $48 of it.

No choice the player makes, such as timing, stake or which machine to sit at within a bank of identical ones, changes that percentage. The outcome of each spin is decided by the machine's random number generator, and the RTP is set by its programme and paytable.

Carmichael did not try to beat that maths. He bypassed it by forcing the machine to pay more than the outcome owed. That is why his activity was fraud rather than advantage play: advantage players use legal information, such as card counting in blackjack, while slot cheats alter the machine's behaviour.

For how RTP is set and measured, see RTP explained, and for the machine logic itself, how slot machines work.

The lesson regulators drew from Carmichael's era was that trust should not rest on a single component checking itself. Provably fair games apply a version of that idea to the result itself.

PVPspinArena runs three player-vs-player games: Jackpot at /, Coinflip and Roulette. Each round's result is derived from seeds committed before the round and revealed after settlement. Because the commitment is published first, the outcome cannot be quietly changed afterwards, and any player can recompute a settled round on the fairness page. For how commitments work, see the provably fair topic hub.

There is no coin hopper to fool and no physical payout to intercept; payouts are balances credited by the system. Verification shows that the result was not changed. It does not change the odds. Coinflip is a fair 50/50, with the winner taking the pot minus any fee shown before entry. In Jackpot your win chance equals your share of the pot. On Roulette's 15-slot wheel, every bet returns 14/15 ≈ 93.33% on average, a 6.67% house edge.

Carmichael's story ends where every slot-cheating story ends: the only reliable edge was illegal, and it stopped working once the machine was redesigned. Honest play has a known price. Gambling is 18+, and limit tools are on the responsible gambling page.

In the same cluster, see also phil ivey and edward thorp.