The Maryland Lottery provides a good example of why true lottery ball machines are much better than a computerized drawing for choosing lottery numbers.
"This is the super-secret lottery vault."
It's 90 minutes before Tuesday's lunchtime picks, and Patrick Morton, the 35-year-old drawing manager for the Maryland State Lottery Agency, is only half joking as he hovers over a small keypad deep inside the studios of WJZ-TV.
Beside the keypad is a large metal door - locked, alarmed, monitored by camera. With a quick peek over his shoulder, he briskly taps in a code known only to four others at the agency. Then he slides a key into the door and swings it open. Inside the shed-sized room are four draped lottery machines, a Compaq computer and a small safe.
The safe protects eight custom-made sets of numbered pingpong balls.
The elaborate security may seem extreme, but not to Morton. After all, he says, these tools create a commodity so precious that it earned the lottery agency nearly $1.4 billion last year: random numbers.
Valuable, elusive and often misunderstood, randomness has never been hotter. "There's definitely an increasing demand for random numbers," says Mads Haahr, a Danish computer scientist who operates a busy online service that creates and delivers them.
And it's not just lotteries and casinos. Much of the demand for randomness is driven by the Internet and the need to encrypt sensitive data.
Every time you buy a book on Amazon or bid on an eBay auction, the store's computers must generate hundreds of random numbers. These numbers, in turn, serve as mathematical code keys for scrambling credit cards and other important information.
"Randomness is really the key to all online security," says Avi Rubin, technical director of the Information Security Institute at Johns Hopkins University.
Scientists are also increasingly turning to random numbers to solve tough problems. Biologists, for example, tap randomness to help them predict the location and function of genes within DNA. Astrophysicists use it to gain insights into the birth and death of stars.
And without randomized clinical trials, in which some patients get experimental drugs and some get placebos, cancer researchers couldn't be sure whether a new treatment really shrinks tumors.
As the demand for randomness grows, some researchers are even dreaming up new ways to create it, experimenting with exotic sources ranging from lava lamps to radio static.
It's just the latest in a centuries-long quest for new ways to generate sequences free from predictability or pattern.
Gamblers in ancient Sumer and Egypt were the first to seek sources of randomness. Their solution: dice.
Coins, cards and numbered balls weren't far behind. But when scientists and statisticians first became interested in random numbers, they found traditional tools too limiting.
The famed Scottish physicist Lord Kelvin grumbled in a footnote to a 1901 paper that his attempts to generate random numbers by tossing chits of paper in a bowl were "quite insufficient." Some chits, he found, were always less likely to be picked than others.
After a similarly unsuccessfully attempt to draw cards from a bag, British statistician L.H.C. Tippett hit on a more creative method in 1927: He dug up church records and recorded the middle digits from the measurements of the area of each parish. Tippett ultimately published a table of 41,600 random numbers generated this way - the first example of an increasingly popular scientific genre.
That genre peaked in 1955, when the RAND Corp. unveiled what is still considered the magnum opus of randomness reference books: A Million Random Digits with 100,000 Normal Deviates.
The 600-page tome sold 500 copies in its first four months and quickly became a favorite of scientists, pollsters, lottery officials and others who required randomness in their work. RAND even received a fan letter from a Navy submarine captain who reported that he consulted the tables to avoid predictability every time he needed to take evasive action.
RAND researchers, who required nearly a decade to ensure the numbers in the book passed statistical randomness tests, did confess to cutting one corner.
"Because of the very nature of the tables, it did not seem necessary to proofread every page of the final manuscript in order to catch random errors," they wrote.
Four decades later, statistician George Marsaglia at Florida State University packed a CD-ROM with 4.8 billion randomly produced 0's and 1's. Marsaglia's recipe for randomness?
"Rap music," he explains.
After generating random digits using traditional methods, the statistician digitized several rap recordings - turning the sounds the artists produced into more digital ones and zeros - and then mixed the result with his previously created sequences.
The CD-ROM, he says, has been a big hit among scientists ever since.
Today most people turn to computers when they need a string of random numbers. There's just one problem: Logic circuits are not good sources of spontaneity. "If a math chip inside a computer does something unpredictable, we call it broken," notes Landon Noll, a mathematician and cryptographer.
That's because computers are what mathematicians call "deterministic." In other words, they do only what they're told to do. To overcome this handicap, researchers are pairing computers with more reliable sources of disorder.
Noll and former colleagues at Silicon Graphics invented one of the hippest of these hybrid random number generators in 1996 by training a digital camera on six oozing lava lamps. The camera captured the movement of the heated oil blobs and converted it into digital sequences that served as seeds for a list of random numbers.
Noll, who now works for the computer security firm System Experts, recently collaborated on a new version of the system called LavaRnd (www.lavarnd.org) that does away with the lamps. It turns out that the chips inside a cheap digital camera with its lens cap generate random electronic "noise" that can do the job, says Noll.
Others are turning to quantum mechanics in the quest for genuine randomness. Their logic: quantum events such as radioactive decay are by definition unpredictable.
So John Walker of Fourmilab in Switzerland has rigged a Geiger counter to his computer and launched a free service called HotBits (www.fourmilab.ch/hotbits). The instrument is trained on a decaying capsule of Krypton-85.
Haahr of Trinity College, meanwhile, is relying on chaos derived from a different source: the atmosphere. Or more precisely, the static it produces on the radio dial.
Haahr hit on the idea while briefly working on an effort to launch an online casino. But he quickly hit a snag: Better portable radios electronically smother between-station static. So Haahr and his team dropped by a Radio Shack and ordered the salesman to tune the cheapest radio in the store between stations. When they heard hiss, "we were jumping up and down," says Haahr. "The guy thought we were pretty crazy."
Since it launched in 1998, Haahr's online service, Random.org, has distributed nearly 68 billion random numbers. Recipients include an Environmental Protection Agency inspector who uses them to pick which companies to audit, a locksmith who uses the digits to decide where to notch keys, and a music composer who channels randomness into his scores.
Despite these innovations, retro sources of randomness - coins, cards and the dancing pingpong balls of a lottery machine - aren't likely to disappear anytime soon. Which is why Patrick Morton flicks on the Compaq computer in WJZ's lottery vault and punches a few keys.
A window pops up. It reads: "Pick 3: 2-8-3"
It looks like the lottery numbers, but the computer is actually randomly selecting which of the eight identical ball sets to pull from the vault for the midday Pick 3 drawing. The computer also tells him which sets to use for Pick 4 and even which lottery machines to wheel into the TV studio.
An hour before the drawing, Morton and a colleague will place the balls in the machines and conduct several test runs to ensure they spot no biases. Finally Morton calls out, "Game ready!"
At 12:28 p.m., as the lottery's rock-tinged theme music echoes through the cavernous studio, Morton smiles under the hot lights and coaxes seven balls to emerge from the Pick 3 and Pick 4 tumblers as announcer Marty Bass calls out the numbers.
Later, Morton estimates that each 60-second televised drawing requires three hours of preparation. "It can be nerve-racking," he confesses.
Jimmy White, another lottery official standing nearby, adds, "People don't realize the lengths we go to introduce and maintain randomness in the process."