We do the research, you get the alpha!
The tightly knit team of a half dozen coders converged in July in a 17th floor hotel suite with a panoramic view of the Las Vegas strip. They had picked the center of gambling in America for a symbolic reason: the team had just spent three years working together to build a new kind of prediction market called Augur, to make it possible for bettors anywhere to place wagers on anything.
Now they were about to unleash their creation on the world by uploading the code to the Ethereum network. Where else could they have gone for this historic moment but to the Strip?
Under a glitzy chandelier in a sky suite strewn with pizza boxes at the Aria Resort, it took the programmers a full day to finish the code and upload all the smart contracts for the decentralized application. They didn’t mind: after three years of working together, partially in a ratty little house on the edge of San Francisco, it was unclear when or if they might see each other again.
That had been the plan all along. They weren’t creating a business where they hoped they all might work. This new kind of enterprise was more like a group of people making an indie movie. They were releasing a protocol—a piece of software that would live forever on the Net. And after that? They would all simply walk away and find something else to work on.
Was it thrilling? “It was not thrilling,” Jack Peterson, Augur’s co-founder recalls.
Oh, but it was. A lot was at stake. In fact, thousands of people had been betting on Augur for three years, including the 2,500 investors who had bought $5.3 million-worth of highly speculative digital "REP" tokens to sponsor its development. Though its creators insist it was a “presale of software licenses,” that event in 2015 was in effect one of the first initial coin offerings.
In the days after the Vegas launch, Augur did not disappoint. Thousands of users had traded upwards of $1.5 million on Augur, and the value of the REP digital tokens grew to the mid $30 range. Hundreds of betting markets proliferated on Augur's interface. Would U.S. President Trump be re-elected? Who would win the France-Belgium semi-final in the 2018 World Cup? Would the price of ether exceed $500 by the year’s end?
But then, within a week, Augur fell to earth. Only a few dozen people were trading it daily. Users complained about the clunky interface, and started to notice the abundance of dud markets (“Does God exist?”). Worse, morally challenged “assassination markets” emerged, which some observers believed might actually encourage bettors to kill celebrities—if the jackpot got high enough. Some news outlets declared Augur a joke. One publication lamented the “hype, the horror, and the letdown of prediction market Augur.”
The truth lies somewhere in between.
If you want to understand what’s happening on the Internet right now and why thousands of developers and billions of dollars are being focused on the promise of web3, Augur is a pretty good place to start. Obviously, it is not a Facebook, Google or Twitter, certainly not at this point in history, and maybe never.
But it represents one of those moments in technology that signals the start of something potentially huge. A better analogy might be to consider one of the early Wright Brothers flights at Kitty Hawk. The Wright Flyer wasn’t much to look at, took you 120 feet and only stayed airborne for seconds. But it was flying, and that was something that might lead somewhere huge, wasn’t it?
Vitalik Buterin--the person most people consider the godfather of Ethereum--certainly thought so. Most people don't know that he was a seed investor, consultant and muse to Augur. Augur, he says, is a success. "Even if it ends up with only 45 users, creating an application of this level of complexity and turning it into an actually working system is still a huge achievement."
Fortune’s Children
Augur’s story starts with Joey Krug, who was born in Knoxville, Illinois in 1995 to an physician's assistant and an ER doctor. He grew up to love betting, business, and bitcoin, and by age 13, he was already winning "thousands" playing the ponies and the stock market, carefully filing the results in an Excel spreadsheet.
He learned about bitcoin in 2011, after reading an article on GPU mining on overclock.net, a hardware site. It ignited his business instincts—by simply hitching a few Radeon GPU units to a gaming computer and letting the whole thing whirr, he was able to earn money, right in the comfort of his childhood bedroom.
It was bitcoin that prompted him to drop out of Pomona College, California, after his freshman year, where he studied computer science. He left to write third-party bitcoin applications, including an app for buying things in bitcoin—which he abandoned once he realised “nobody wanted to buy things in bitcoin.”
Nevertheless, his marginal interests connected him to a Skype group, around 2014, with Buterin, who would soon co-create Ethereum, as well as Peterson, a then 32-year-old engineer and biophysicist working on his own abortive blockchain project, a startup called Dyffy.
Peterson was born in 1982 and grew up in Atlanta, Georgia. Unlike Krug, he had never been much of a gambler. He wasn't even into money that much. He had had a stash of 100 bitcoins, which he accidentally wiped when reformatting his hard drive. Though he narrowly escaped great riches, he has no regrets.
But he thought Intrade, an early prediction market that had been abruptly shut down in 2013, was “incredibly cool.” It was unclear what caused Intrade’s shutdown—the company claimed it had been forced to shut down due to discovering "financial irregularities." Others pointed to a U.S. government lawsuit that prohibited people in the U.S. from using it, which had cut it off from its American market.
Whatever it was, Peterson remembers wishing that “Intrade could be like bitcoin.” By distributing a prediction market’s administration across a global, independent network of computers, he reasoned, it would have no single point of failure. Traders could go on trading, no matter what.
Ethereum Savant
Buterin—whom Krug describes, with mathematical precision, as a “value-added person”—would provide the spark. It was 2014, and his invention, Ethereum, was emerging as a programmable alternative to the bitcoin network. He was mulling over how to resolve a problem with Ethereum's “smart contracts,” digital agreements that fulfill themselves algorithmically, without human intervention. There was just one problem: what would verify that the conditions of a contract had been met, if not humans?
“Blockchain doesn’t know things from the outside world,” Buterin explains. “It doesn’t know what time it is, what the temperature is.” For complex smart contracts to work, “you need to source that info somewhere,” he says. That’s known as the “Oracle Problem.”
During his research, Buterin came across a widely circulated Princeton treatise,“On decentralizing prediction markets and limit order books,” as well as a paper by Paul Sztorc, a statistician at Yale University, detailing a protocol called “Truthcoin.” Both papers, loosely, advocated deferring the smart contracts’ truth-finding duties to a decentralized network of “reporters,” thereby solving the Oracle Problem by establishing a human link with the code. In their vision, a new kind of prediction market could then run on these contracts, which would dispense payouts automatically, without need for a middleman—the algorithmic equivalent of bookies. Thus stimulated, Buterin drafted a blueprint for “Schellingcoin,” which would largely do the same.
Motivated by divisions within the bitcoin community, which Buterin says was then “spiralling into civil war” over technical differences, he published the paper, hoping it might both resolve the Oracle Problem and foster a new kind of “on-chain,” betting-based governance model that could be adopted by the burgeoning Ethereum network. Such a system, he speculated, would encourage his users to put aside their differences and put their money where their mouths were.
The blueprints for Schellingcoin and Truthcoin found their way to Peterson and Krug, as well as to Joe Costello, an angel investor supporting Peterson’s startup Dyffy. Costello, bored on a holiday in the Maldives, read and re-read Truthcoin so many times that he became “obsessed” with the idea of building an advanced version of it; he figured, in time, that Augur would support third-party apps he could profit from. Peterson and Krug were willing to take on the project, which Costello helped kick off with a seed fund of “around $1 million.” (Though Peterson puts it at half a million.)
Costello's seed investment quickly ran dry, however, so Jeremy Gardner—another co-founder who would help author the white paper—forked up a loan to tide them through to the ICO. He bet, in his words, his "entire life savings on Augur." Buterin would later invest, too.
Pennies from Heaven
Thus funded, Krug and Peterson, as well as two advisors, wrote a whitepaper detailing how the protocol would run. They named the project Augur, after the Roman seers, or “augurs,” who predicted the future by observing the flight patterns of birds. The logo was a pyramid, with three points converging upon an all-seeing eye.
To start, they built an alpha version on the bitcoin network. Buterin, however, urged them to change tack and build it on Ethereum, which would be easier; so they did, and it was. Within months, they had a working prototype on the Ethereum “testnet”—the Jornada del Muerto desert of decentralized software—that they could use to sell the project to public investors.
For a 45-day period between August and September, 2015, they sold off 11 million “REP tokens” at 60 cents apiece, saving 20 percent for themselves. The idea was that these tokens would provide minor financial incentives for holders of Augur’s REP token.
These holders, in their role as "reporters," would then collectively act as Augur’s “Oracle” by voting on the outcome of events in exchange for more REP—if they were truthful. If they voted out of line, they would incur a penalty and lose REP. REP holders would also be entitled to 50 percent of all Augur’s trading fees, with the other half going to market makers. (It wasn't until bitcoin's vertiginous rise in 2017 that they realised the REP could run up in price and, conceivably, generate supplementary incomes for its holders.)
The auction went astonishingly well. That same month, the Chinese stock market had tanked, and traditional IPOs were struggling to raise money. Yet Augur, on its 19 pages worth of relatively untested ideas, managed to crowdsource $5.3 million without the support of venture capitalists, banks, or any kind of institutional middleman.
Krug remembers being “pleasantly surprised.”
Pleasantly surprised?
“I was pleasantly shocked,” he says.
Buterin, meanwhile, was pleasantly nothing.
“Hmm,” he reminisces.
Did he at least feel proud? His underlying algorithm churns, searching for the correct variant. “Hmm,” he concludes.
Bingo.
The Price of Immortality
That money, along with the cut of the REP tokens the team had reserved for themselves, would prop up the Forecast Foundation, a not-for-profit.
The Foundation would write the team’s checks, support Augur’s development, and, in Costello’s words, generally “keep it alive and well.” Yet at the same time, it would be functionally powerless.
That’s the rub of the decentralized web. Companies, necessarily, must relinquish control over their products, and willingly withdraw themselves as middlemen. Indeed, the Forecast Foundation had, and has, no central power to either shut Augur down or even forcibly upgrade it. The updates to Augur’s interface, like those on the Bitcoin network, would only be optional downloads for its users. What’s more, to protect the Foundation from regulation and culpability in the event of Augur’s misuse—see “assassination markets”—the Foundation would take no profit from Augur’s markets.
Yet this swings both ways. With no chance of the Foundation generating revenue from the Augur platform itself, these funds, plus the initial seed investments, would have to carry the project through to the end. If those funds were to diminish, Peterson says, the Foundation would be unable to fund its employees and would have to outsource to “volunteer developers.”
Still, the Augur protocol itself would survive any collapse in the Foundation, albeit in an incomplete form. Says Buterin: “If the Forecast Foundation disappears,” he says, “then there’ll be no more future updates.” It’s like if BMW collapsed, but somehow kept rolling out half-built cars.
Costello, however, still thinks it’s possible to turn a buck: Augur could provide valuable speculative information to “any marketplace where people are trying to make a prediction,” he says, citing sports and finance as two lucrative areas to mine.
Third-party apps that scrape and resell reliable predictions to gamblers, and people trading equities, he says, could generate handsome returns; prediction markets, drawing as they do on vast reserves of crowd knowledge, have an uncanny tendency to hit the nail on the head when guessing the future.
The Long Code To Prediction
It would take three years for the core development team, which initially consisted of Krug, Peterson, and two others, Chris Calderon and Scott Leonard, to get the final product off the ground.
Though most of the team collaborated remotely, a handful, including Krug and Gardner, lived and worked in what they called the Bitcoin Basement, a San Francisco crypto hotspot. The Basement, Peterson recalls, was “gross.” Three bedrooms—and one toilet—served six guys, none of whom cleaned. “At least not regularly,” says Peterson. He lived in Oregon at the time and, when visiting, would politely refuse board and sleep in his car. They kept the Crypto Castle, their next (above-ground) home, in "better order," he says. But not much better.
The team programmed. They pushed out a beta in 2016, but it was far from complete. In the summer of 2017, they found a vulnerability that hackers could exploit to lock users’ money in the smart contracts Augur relied on to dispense payouts automatically. “Anyone who had money in one of those smart contracts would have lost it forever,” says Krug. The team had to transfer the code from Serpent to Solidity, a programming language that didn’t contain the vulnerability. It delayed launch by another year. Investors were growing restless. Why was it taking so long?
But slowly, Augur came to life.
Growing Pains
Augur’s launch was widely covered. Thousands of users flocked to the network, numbers not seen since CryptoKitties crashed the Ethereum network a year before. Most of the $1.5 million traded was wagered on World Cup-related markets, and other markets proliferated at breakneck speed. But soon the squib was dampened.
Detractors lined up to blast the software’s clunky, unintuitive interface, the proliferation of dud markets, the high trading fees, and the dismally low value of the REP token. The cryptocurrency, which traded at $32 and was an enticement to Augur’s reporters, has been consistently hovering around $12.
These are all valid points, says Krug. “We said it was going to be expensive, slow to use, and have a terrible UX,” he explains. Peterson adds: “People had unrealistic expectations for what that first iteration would look like.” Augur is a 1.0 product. Augur 2.0, expected to drop soon, will address technical issues, largely drawn from a Reddit “wishlist.” (One upgrade will integrate Augur with the DAI, a dollar-backed stablecoin that will stop traders’ funds from dropping in value.)
There are other more fundamental problems. Paul Sztorc, the Yale statistician who authored the seminal Truthcoin paper that so inspired Buterin, had ripped into the project in a massive take down he published at the end of 2015. The paper lays out every conceivable objection. (The Foundation is also in the midst of an $152 million legal battle with Costello’s former partner, the developer Matt Liston, whom Sztorc supports.)
Will Jennings, media director of PredictIt, a competitor, asserts that Augur never even fully solved the Oracle Problem. People are easily co-opted, he says, and are otherwise unreliable sources of truth. What if, for instance, fundamental religionists hijacked the network and collectively voted that “God is real,” ensuring that payouts went only to the faithful? (Though the $9.83 on that market suggests this won’t happen anytime soon.)
Krug concedes that Augur could indeed be prone to manipulation. But he readily admits the system isn’t perfect. This is a young technology. People thought the first generation iPhone was a “useless piece of tech,” he says. Now look at it.
And anyway, he—and Peterson—have since parted ways from Augur.
Though Krug retains his role as “advisor,” he’s now working full time as a hedge fund manager at Pantera Capital. He doesn’t even use his Augur email address anymore. Peterson, meanwhile, has returned to his old love, biophysics.
And why wouldn’t they leave? They created a business whose explicit end result is no one should run it. So they can jump ship, scot-free, leaving the rest of the development team to subsist on the diminishing winnings of that original, sepia-tinged token sale. And in a way, their departure is a final flourish, a last act of total decentralization; the creators, Augur’s original middlemen, dutifully removing themselves.
And Augur’s muse, Buterin? He frames the whole thing as a massive, cosmic gamble whose payout has yet to be delivered: “We’re staking a claim that if we make blockchain apps more usable, people will use them,” he says, with recursive simplicity.
So will the gamble pay off? Give it some time: It was a decade after Kitty Hawk before the first commercial flight took to the skies. Augur’s small base of users aren’t especially bullish on Augur’s value. But they're still placing bets.