Man Vs. Machine

Curzi, 35, had moved to Oregon in 2012 from San Francisco after selling a software company he'd helped found a decade earlier. He was fascinated with the games—the ubiquitous, flashing terminals found in bars, delis and even pancake houses—and he played occasionally when out drinking with friends.

On this day—Jan. 10, 2014, a Friday—Curzi paused playing video poker while a pal went to get a beer. He used the break to study his hand—a 2, 4, 5, 6 and 7 of different suits. He was close to getting a straight, which would pay $5 on a $1 bet.

The game Curzi was playing, draw poker, allowed him to discard cards and get new ones from the dealer. He knew his best chance was to discard the 2 and hope the machine dealt him a 3 or an 8 to complete a straight.

But the machine suggested he do something Curzi thought strange: It recommended he discard the 7. He would get his straight only if he drew a 3. That would cut Curzi's chances of winning by half—and he thought it was terrible advice.

"Hey, is this right?" Curzi asked his friend when he returned.

Curzi took out his iPhone and snapped photos of the screen and the machine's serial number.

It was the first step to uncovering what he says is a $134 million scam by the Oregon Lottery.

Bad Advice

Here's how the video poker hand Justin Curzi got on Jan. 10, 2014, led him to investigate the Oregon Lottery machines.

Oregon voters approved the state lottery in 1984, and today state-run gambling contributes about $550 million a year to Oregon's budget, behind only personal income taxes.

The lottery encourages dreams of riches. But the games are engineered to take your money. "Everyone should understand that the odds in all our games favor the lottery," says Jack Roberts, Oregon Lottery director.

That's why news reports two months ago that a Portland man was suing the lottery to recoup video poker players' losses struck some as ludicrous. Who would sue over losing money while gambling?

But it's not so simple. Curzi—who friends say is intelligent, analytical and obsessively curious—launched a personal investigation of Oregon video poker machines that led him to conclude the machines were cheating players out of millions of dollars every month. That's why he filed a class-action lawsuit against the Oregon Lottery in Multnomah County Circuit Court, alleging fraud. Lottery officials deny Curzi's allegations.

"Good for him," says Les Bernal, national director of the advocacy group Stop Predatory Gambling, based in Washington, D.C. "What the Oregon Lottery does with these games is create the illusion that you have some control, where in reality you actually have far less."

Curzi is aware some people might assume he's suing to make money. He insists he's not. "The real reason I'm doing this," he says, "is because it's outright wrong."

"Justin is not afraid to jump at things, he's not afraid to question things," says Rob Steele, a friend of Curzi's dating back to high school in New Jersey. "That is just catnip for Justin."

Six days after his curious experience with the Jacks or Better game at Quimby's, Curzi sent a polite and inquisitive email to the Oregon Lottery.

"Hello, my name is Justin," he wrote on Jan. 16, 2014. "I've attached a photo of a hand that was given to me in one of your Oregon Lottery machines."

Curzi explained how he believed the video poker machine should have given him the best advice. "This does not seem to be the case," he wrote.

Draw poker is a game of luck, strategy and second chances. The dealer gives players five cards. Players then get a chance to discard cards in the hopes of being dealt better ones.

When you're playing poker around a table in real life, you're betting against other players in hopes of having the best hand.

But in video poker, you're not betting against anyone. Each hand costs you 25 cents (or more, if you increase your wager), and you win money based on a scale of how strong your hand is. A pair of jacks might win you your 25 cents back. A royal flush—the highest and rarest combination—would win you up to $600.

Unlike slot machines, video poker gives the player a sense that strategy matters. In reality, if you play long enough, the machines are geared to eventually take your money, no matter how many wins you record. Still, the sense that a player can outsmart the game is part of its allure.

What caught Curzi's attention was a feature on the draw poker games called "auto-hold." The feature puts the word "hold" over cards it suggests players should keep. Players can reject the suggestions at any time.

But auto-hold has a second, less obvious function. It allows players to play faster, because they don't have to stop to think about what cards to hold before hitting the button to draw again. That's important because faster play translates to more money for the Oregon Lottery.

Before he got a response from the lottery, Curzi returned to Quimby's, wondering whether the game's bad advice had only been a fluke.

He shoved a $20 bill into a machine to play the same Jacks or Better game. Within 10 minutes, the game was again advising him to hold cards that cut his chances of winning in half. Curzi says he just wanted a simple explanation.

"I certainly didn't think," he says, "I would discover what I know now."

Who Plays Oregon Lottery Games?

Curzi grew up as a sports-focused kid in small-town New Jersey, the son of a prominent lawyer and a stay-at-home mom. Curzi—who played football despite his small size—also wrestled, played baseball and graduated near the top of his high-school class. He played wide receiver at Amherst College, where he majored in economics and history. That led him to New York City after graduation.

"I thought the only two jobs on earth were investment banking and consulting," he says now.

He landed his first job selling investments. Working on commission, he'd target an office building, climb to the top floor, then work his way down, knocking on doors. "I was 21, looking like I was 16, asking people to give me their money," Curzi says. He soon climbed the monthly leader board. His boss told him he was one of the youngest salespeople to reach the top.

He wasn't destined for a traditional job. A sticker on Curzi's apartment door showed a group of people heading one direction, and one person walking the other way. "Routine," it read. "The enemy!"

In 2003, he moved to Brazil and quickly immersed himself in the culture, teaching himself Portuguese within months. "You feel like the guy has been there two or three years," says Ken Barrington, a college friend who visited him.

In Brazil, Curzi met an American computer programmer working on a way to help accountants share QuickBook files. The two teamed up and sold the program, cold-calling potential clients from Rio de Janeiro on an Internet phone line. "We must have sounded like we were speaking through tin cans," Curzi says.

They called the business Emochila—mochila means "backpack" in Spanish and Portuguese—and it blossomed to 30 employees. In 2011, Curzi and his partner sold the company to Thomson Reuters in a private deal; Curzi declines to say for how much. But friends describe him as wealthy. "I'm not Elon Musk," Curzi says of the co-founder of Tesla and PayPal.

Curzi moved to Portland in 2012 with his then-girlfriend (and now wife), who grew up in Tigard, and now lives in a $565,000 Victorian in Northwest Portland. He consults for private clients, provides microloans to entrepreneurs through the website Kiva and drives a 1996 Isuzu Rodeo "whose crowning feature is where a dog chewed the back seats."

Friends say they are not surprised Curzi—who's just as likely to want to discuss North Dakota's fracking economy as the business model for Purringtons Cat Lounge—zeroed in on something as small and seemingly innocuous as a quirk in a video poker game.

"So many times in life, people just overlook the obvious," Barrington says. "Justin has a knack for pointing those things out."

Marlene Meissner, a spokeswoman for the lottery, drafted a response. Auto-hold, she wrote, "is based on optimizing the player's opportunity to win the best (highest prize) rather than simply increasing the odds of winning any prize."

But, as Curzi later discovered, lottery officials guided Meissner to a different answer, so she revised her email before sending it. "In your case, the terminal did advise a strategy — granted not the only strategy — for you to have an opportunity to win with the cards you were dealt," she said in her email to Curzi on Feb. 3.

In other words, the lottery was backing away from telling Curzi auto-hold offered the best option.

Curzi wasn't satisfied. "I know your interpretation of the law is that you only have to suggest 'a' winning combination, but why not the best one?" he wrote in an email the next day.

The lottery's response? "Crickets," Curzi says.

Curzi turned to Jay Zollinger, a lawyer who had helped negotiate the sale of Curzi's business. Zollinger suggested a public records request might turn up some answers.

On Feb. 20, 2014, Curzi and Zollinger formally asked the lottery for documents concerning the Jacks or Better game Curzi played at Quimby's, plus any correspondence, studies and reports about auto-hold. The lottery responded on April 8, saying it would take 30 hours of staff time just to review the records Curzi requested. The lottery wanted a $2,350 deposit to cover its costs.

That fee would have stopped most people. But Curzi's lawyer paid it. The total bill for records eventually came to $3,581.49.

Six months after his request, in August 2014, Curzi received the first of five batches of records.

By September, Curzi had hundreds of pages of emails, memos and spreadsheets. He made a copy of the originals, arranging one set chronologically and the second by topic. He took notes on his laptop in a file that grew to 4,800 words.

Curzi came across a Feb. 2, 2009, email with a spreadsheet attached—"Video Lottery Game Payout Percentage Report." The document had come from Gaming Laboratories International, an independent auditor based in New Jersey that works with many state lotteries to test machines.

The spreadsheet listed all the types of Oregon video poker machines by manufacturer, the millions of games played in one quarter of 2008 and how much money players spent.

In one column, the document showed what various video poker machines, based on calculations of probabilities, were expected to pay out to players over time.

In another column, the document showed what the machines were actually paying out. Curzi thought the payouts should have been very close to what the game's programmers predicted.

Some weren't. Curzi discovered the game he had been playing at Quimby's, the Jacks or Better "Bluebird" terminal produced by WMS Gaming, was off by quite a bit.

The spreadsheet showed Jacks or Better on average should be paying out 90 cents for every $1 players put into the machine. It actually paid out about 87 cents.

That 3-cent difference may seem small, but when multiplied by the huge numbers of video poker games played, it translated to about $1.3 million per year that Jacks or Better wasn't returning to players.

"This," Curzi recalls thinking to himself, "is totally corrupt."

He kept digging and made a second big discovery: Lottery officials knew about the discrepancy, and the auto-hold function on some machines was to blame.

"Due to the vendors' auto-hold strategies, a few other poker games have actual payout percentages that are below theoretical," Carole Hardy, the lottery's then-assistant director for marketing, wrote on April 1, 2009.

Curzi discovered a survey of video poker players the lottery commissioned from Mosak, a marketing research firm.

"Across all player types, the overwhelming majority of players said they prefer the auto-hold feature in video poker games as it makes it more convenient and easier to play," a 2010 Mosak report said. "Players said this feature allows them to hold the correct cards, thus increasing their chances of winning."

Curzi had only hoped to understand how auto-hold worked. He had instead discovered the lottery knew auto-hold sucked millions away from players—and players actually thought auto-hold helped them.

The lottery's rules require "a close approximation of the odds of winning some prize for each game" and say those odds "must be displayed on a Video Lottery game terminal screen.” 

Documents Curzi received show lottery officials debated whether or not they should tell players the actual odds if they relied on auto-hold.

In a memo labeled "confidential" and dated Sept. 15, 2009, lottery officials reported they had been studying their system to find video poker games that might be making payouts that were too high. Instead, they found machines whose payouts were too low.

"This triggered additional investigation regarding the integrity of the games," the memo said. "Further, there was a question whether additional information should be provided to players to ensure they have accurate information regarding how video lottery games pay."

The Sept. 15 memo also contained this nugget about WMS Gaming, maker of the game Curzi played at Quimby's: "WMS has confirmed that the auto-hold strategy for all WMS poker games is set to pay out lower than the other products as a result of the auto-hold strategies WMS has implemented."

Lottery officials, according to a separate 2009 memo, decided to put accurate auto-hold payouts on the Web. But Curzi went looking online, even using the Internet Archive search engine, to see if the lottery had ever made public the lower odds. He found no evidence it had.

Over the next month, Curzi built a spreadsheet to estimate how much money the video poker machines, based on the odds, should have paid out, compared to what they actually did.

What he found startled him. Payouts to video lottery players were as much as 5 percent lower when they used auto-hold than when they didn't. That translated to $134 million.

To Curzi, it was an outrageous discrepancy—especially given that players believed auto-hold helped them, and the lottery knew otherwise. Buried on the lottery's website is one disclaimer: "Auto-hold strategies vary by game, based on the particular features of a game and do not necessarily result in theoretical payouts."

Curzi says that's not enough. The lottery is supposed to be based on chance. "You can't manipulate the game," he says.

In October 2014, he sent the Oregon Lottery a letter detailing his findings and notifying officials he intended to sue unless the lottery reimbursed players within 30 days. On Dec. 4, a claims management consultant in the state's Department of Administrative Services wrote back to say the lottery was still investigating Curzi's claims.

On Dec. 31, Curzi took the Oregon Lottery to court.

Jack Roberts, the lottery director, took over the agency in December 2013, following years of controversy and accusations the agency wasn't doing enough to address problem gambling. He had earlier served as state labor commissioner and ran in the Republican primary for governor in 2002.

Roberts says the lottery is fairly representing players' chances. "Clearly the odds favor us," he says. "That's what gambling establishments are about, but we believe we've been honest in representing what they are."

Roberts wasn't around when the lottery introduced video poker and the auto-hold feature in 1992. "Our assumption has always been that on balance people who play auto-hold do better than people who don't," he says. "We don't tell people that."

He rejects Curzi's allegation the lottery is intentionally misleading players. "I don't think we've ever represented that the auto-hold gives you the optimal result," he says. "The idea was that it gives you a good result."

But records Curzi turned up show the opposite. "The machine recommends the best possible cards to hold in order for the player to win and if the player changes the cards to be held, the possibility of winning will decrease," the Sept. 15, 2009, memo marked "confidential" reads.

Today, the lottery is in the process of replacing all 12,000 video lottery terminals in the state; it's a routine technology update. But one consequence of the upgrade is that Oregon is completely phasing out the WMS Gaming "Bluebird" terminal on which Curzi played Jacks or Better.

Roberts says Jacks or Better is being phased out because it's unpopular with players.

Roberts says the lottery is interested in finding out if more players are concerned about auto-hold.

"It gets complicated in the middle of litigation," he says. "Any actions that we take might be interpreted as an admission that we don't mean to say."

Experts on lotteries and the law say Curzi's odds of winning in court seem low. Rob Carey, an Arizona class-action lawyer, took on several state lotteries over the deceptive practice of selling scratch-off tickets after the top prizes had already been awarded. Carey never succeeded in getting a class established for his lawsuits, but he did win payments for some plaintiffs and forced changes in state lottery practices.

He says Curzi's case hinges on whether the Oregon Lottery's public disclosures were adequate. "It really depends on what they're telling the players," Carey says.

The lottery could be safe even if the disclosures are vague. "You have to show the intent to defraud," says I. Nelson Rose, a law professor at Whittier Law School in Southern California. "I don't think they'll be able to do that."

Rose says it’s the machines’ manufacturers that should be worried.  “If the plaintiff were able to prove this was intentional,” he says, “that supplier could end up paying.”

Nevada-based Scientific Games, owner of WMS Gaming, the maker of the Jacks or Better game Curzi played, declined to answer WW's questions. "It is company policy not to comment on ongoing litigation," Scientific Games spokeswoman Mollie Cole said in an email.

Curzi is undaunted. He wants players to recoup their money. He wants the lottery's auto-hold feature to give good advice, and he wants the agency to give players honest information.

"It goes all the way back to that first photo," he says of the photo he took of the video poker machine's bum recommendation at Quimby's last year. "I look at it and say, 'That's not right.'"

WWeek 2015