Random

True stories

When the lottery wasn’t random

Painted ping-pong balls. A drum nobody stirred. A “random” game board on a loop. The insider who wrote the code. Four true stories about what happens when chance gets a helping hand.

Skip to the experiment

By Stuck at Home, LLC ·

Picture a television studio in 1980. Three clear tubes, ten numbered ping-pong balls in each, a blower underneath. The announcer reads the night’s number as the balls dance: six… six… six.

A result like 666 is no less likely than 174 or 925. Every three-digit number has the same one-in-a-thousand chance — if every ball weighs the same. That evening, most of them didn’t. Someone had injected the unwanted balls with paint, so only a handful could float to the top.

Fixed draws are rare, and they are instructive, because they show what fairness is made of. Equal weights. Proper mixing. Nobody with their hand on the code. Take any one away and the result can still look random.

Three tubes, ten balls each
TUBE ONE6heavy: 0 1 2 3 5 7 8 9TUBE TWO6light: 4 and 6TUBE THREE6eight results left
Only 4s and 6s can fly.
The arithmetic of a fixed draw:
a thousand outcomes shrink to eight.

Try it. Draw a few numbers from an honest drum, then tick the box, paint the balls and draw again.

Your own three tubes

Rig the drum. See what survives.

An honest draw gives every ball the same chance. Paint the balls and only the 4s and 6s can rise.

  • 0
  • 1
  • 2
  • 3
  • 4
  • 5
  • 6
  • 7
  • 8
  • 9
  • 0
  • 1
  • 2
  • 3
  • 4
  • 5
  • 6
  • 7
  • 8
  • 9
  • 0
  • 1
  • 2
  • 3
  • 4
  • 5
  • 6
  • 7
  • 8
  • 9
???

Honest drum

No draws yet

    Painted drum

    No draws yet

      Start with a few honest draws.

      Gold entries use only 4s and 6s. In the honest drum that happens about 8 times in 1,000; in the painted drum, every time.

      1980: the Triple Six Fix.

      The game was the Pennsylvania Lottery’s televised Daily Number, and the date was 24 April 1980. Announcer Nick Perry and lottery official Edward Plevel had injected white latex paint into every ball except the 4s and 6s. Heavy balls sink; light balls fly. Only eight results were possible that night, and the conspirators and their partners had bet heavily on combinations of 4s and 6s. The number came up 6-6-6 and the draw paid out $3.5 million. Perry was convicted in 1981 and served about two years in prison plus time in a halfway house; six people were sentenced in a scheme reported at $1.2 million.

      Notice what the fix did not touch. The blower still blew, the balls still danced, the draw still looked random on television. Equal weights were the only thing missing, and that was enough to shrink a thousand outcomes to eight.

      1969: the drum nobody stirred.

      Not every bad draw is a crime. On 1 December 1969 the United States held its first draft lottery since 1942: 366 capsules, one per birthday, drawn from a glass container to decide the order in which young men would be called up for Vietnam. The capsules had been put into the box month by month, January first, December last, and the mixing was not enough to undo that order. Late-year birthdays came out early. January birthdays averaged a draft number of 201; December birthdays averaged 121. Statisticians noticed within weeks — the New York Times in January 1970, Stephen Fienberg in Science in 1971 — and the next year’s lottery used a revised procedure. Nobody cheated. The drum simply hadn’t forgotten the order the capsules went in.

      1984: the board that repeated itself.

      On 19 May 1984 a contestant named Michael Larson won $110,237 in cash and prizes on the CBS game show Press Your Luck, then the largest one-day haul in game-show history. He had studied taped episodes frame by frame and found that the “random” prize board lit up in a small number of repeating sequences. Memorise the sequences and the Whammy never comes. His 47-spin run was so long it aired across two episodes. The board was not rigged; it was simply not random, and one careful viewer had the patience to prove it.

      1994 and 2010: when the computer is the weak point.

      In April 1994 a computer analyst named Daniel Corriveau beat the Montreal Casino’s brand-new electronic keno three times in a row — reportedly picking 19 of 20 numbers. The casino shut the game down and the provincial police fraud squad investigated him, polygraph included. The fault was the casino’s: the machine was missing the clock that should have reset its pseudo-random generator, so it restarted the same sequence every time it was switched on. Corriveau had spotted the repeat. Loto-Québec admitted it had not tested the game and paid his family more than C$620,000.

      The modern version is an inside job. Eddie Tipton rose to director of information security at the Multi-State Lottery Association, the body that runs drawings for dozens of American lotteries. He installed code in the drawing computers that let him predict the winning numbers on three days of the year. A ticket for the Iowa Hot Lotto jackpot drawn on 29 December 2010 was presented anonymously through lawyers less than two hours before it expired a year later, and was never paid. An Iowa jury convicted him in 2015; in 2017 he pleaded guilty to rigging drawings in Colorado, Wisconsin, Kansas, Oklahoma and Iowa and was sentenced to up to 25 years. The machines he compromised looked exactly like honest ones.

      What these four have in common

      Every fixed or failed draw removes one ingredient of fairness. Pennsylvania lost equal weights. The draft lottery lost mixing. Press Your Luck lost unpredictability. Montreal and MUSL lost an honest, well-seeded generator. The results in each case were still “random numbers” in the everyday sense — they just weren’t equally likely, and somebody knew it. That is why serious draws publish their procedure, test their equipment, keep witnesses in the room and, increasingly, use randomness that strangers can verify after the fact.

      What this means for your raffle.

      You are not going to be defrauded by a lottery insider at the school fair. But the same ingredients apply at any size. Make sure every entry is the same “weight” — one slip per person, or a clear rule if people may enter more than once. Mix properly, or let software do it. Decide the procedure before the draw, not after. And draw in front of people. Our raffle picker uses the same Web Crypto source as everything else on this site, generates the pick on your device, and has no idea who is in the room — which is exactly the kind of indifference you want.

      Run a draw nobody can rig.

      Number the entries, draw in front of everyone, keep the list. Our raffle picker does the choosing; you do the witnessing.

      Open the raffle picker
      Where we looked things up