TABLE OF CONTENTS ABSTRACT 129 TABLE OF CONTENTS 130 INTRODUCTION 131 I PREDICTIT, FAKE POLLS, AND ONLINE ELECTORAL MANIPULATION 135 A How PredictIt Operates 135 B The Emergence of Fake Polls 140 II THE AVENUES FOR CRIMINAL LIABILITY 150 A Commodities Fraud 150 1 Development of Commodities Fraud 151 2 Elements of Commodities Fraud 154 3 Application of Commodities Fraud 157 B Wire Fraud 162 1 Defining Wire Fraud 163 2 Applying Wire Fraud 168 C Comparing the Avenues 173 III THE CASE FOR CRIMINAL LIABILITY 174 A The Known (and Possible!) Harms of Fake Polls 174 1 Undermining the Purpose of Political Prediction Markets 175 2 Destroying Trust in Polling Companies and the News Media 177 3 Manipulating the Democratic Process 178 B Weaponizing Fake Polls 182 C Justifying Criminal Liability 186 CONCLUSION 189 INTRODUCTION
Suppose that someone is an investor in one of America's most profitable novelty weathervane manufacturers. The company is doing well, but the investor wants to drive up the value of her shares, so she crafts a clever scheme. Using her business acumen, she generates a devastating, but entirely fraudulent, earnings report for the company's biggest competitor and posts a link to the report on Twitter--from a Twitter account she created that purports to be a legitimate financial news outlet. The news goes viral, the competitor's stock dips, the investor's company sees a slight increase in its price, and she sells her shares for a nice profit. (1)
The conclusion that the investor's actions--a sort-of reverse "pump and dump" (2) or a "distort and short" (3)--would violate federal (and probably state (4) ) law would be uncontroversial. (5) But how would this conclusion differ if, instead of generating a false earnings report in an attempt to drive down a company's stock price, the investor generated a fake poll in an attempt to drive down a political candidate's "stock price" on an online betting market for elections?
PredictIt is a prediction market--officially a "not-for-profit market for event contracts"--operated by the Victory University of Wellington in New Zealand. (6) It allows its users to bet on a number of different political events (7) --like the outcome of elections, the successful passage of legislation in Congress, the possibility of action by federal agencies, the odds of certain political actors being criminally prosecuted, and so on. But unlike virtually every other political prediction market, PredictIt is legal. (8) It received a no-action letter from the U.S. Commodity Futures Trading Commission ("CFTC") in 2014, which authorized its operation with a handful of restrictions. (9) PredictIt is, at least ostensibly, "operated for academic research purposes only," and its "main objective . . . is to determine whether it can aggregate information and predict outcomes of certain events," like elections, "more accurately than through alternative means, such as public opinion polling." (10) Bettors on PredictIt rely on inside information, current events, and opinion polls (both publicly released and internal) in making their investments. (11)
Given the heavy reliance of PredictIt bettors on public opinion polls, it is perhaps unsurprising that some unscrupulous actors decided to publish fake polls, (12) likely in an effort to distort PredictIt's market. (13) This was first reported in July 2017 when Delphi Analytica released a poll of the U.S. Senate election taking place in Michigan in 2018. (14) The poll showed musician Kid Rock, who was tested as the Republican nominee, leading Democratic Senator Debbie Stabenow 30% to 26% and received a great deal of media attention and hand-wringing over Stabenow's election prospects. (15) Notably, the share price of Stabenow "stock" initially dropped on PredictIt from 78 cents to 63 cents, and ended at 70 cents for the day (16)--a fairly significant response to a single event.
However, when Harry Enten--an analyst with Nate Silver's FiveThirtyEight--reviewed the poll, he concluded that "something didn't smell right about it" and that it "may not have been conducted." (17) Enten pointed out that the lack of specifics about the poll's methodology, (18) the total anonymity of the polling firm's leadership, and contemporaneous remarks in chatroom by the person apparently responsible for promulgating the poll all suggested that the poll wasn't actually conducted. (19) If Delphi Analytica didn't actually conduct its Michigan poll--as seems likely--it definitely wasn't alone. Other outfits operating under a similar haze published polls in at least six other elections in five other states. (20)
While the long-term effect of the Michigan poll was virtually undetectable--Kid Rock decided not to run for the Senate and Senator Stabenow easily won re-election over someone who likely never ate a grit sandwich for breakfast (21)--the long-term effects of other fake polls might not be so easily dismissed. Polls are usually seen as reflections of public thought, but they do more than reflect public opinion--they refract it, too. (22) As this Article explains later, this refraction has real-world political effects, like allowing candidates to participate in debates, helping (or hurting) campaigns' fundraising efforts, affecting turnout (positively or negatively), and even impacting election results. (23) Fake polls, therefore, have far more serious harms than manipulating small-dollar political prediction markets--in the wrong hands, they could chip away at the integrity of our democracy.
Therefore, like Wall Street, The Wolf of Wall Street, or any other good movie about securities fraud, the story of fake polls and their impact on PredictIt markets--and elections themselves--deserves a satisfying conclusion, with comeuppance for the fraudsters. This Article argues that, in the denouement of this story, federal criminal liability should come to bear on the creators and purveyors of fake polls. It provides a detailed overview of fake polls and identifies several theories of criminal liability for publication of fake polls. It then argues that, regardless of which theory is utilized, the authors of publicized fake polls should face criminal charges.
Part I begins by providing a more detailed explanation of how PredictIt operates and then addresses the origin of fake polls. It explains--to the greatest extent possible given the shadowy nature of the subject matter--the life cycle of fake polls, from conception to publication to fallout. Next, Part II addresses two different theories of criminal liability for the publishers of fake polls--namely, commodities fraud and wire fraud.
Finally, Part III substantively argues that the publication of fake polls should be considered illegal and prosecuted to the fullest extent of the law. It contends that fake polls could have real-world consequences, like defeating PredictIt's purpose as an organization, affecting the public's trust in polling companies and political journalism, altering election outcomes in close races, and most concerningly, opening the door to even greater foreign interference in American elections.
PREDICTIT, FAKE POLLS, AND ONLINE ELECTORAL MANIPULATION
To understand how the publication of fake polls by rogue traders is likely illegal--and why it should be--it is helpful to understand how PredictIt operates from the user's perspective, from start to finish. Section A begins by providing a thorough, mechanical examination of PredictIt's operations. Section B traces the origin of the first-reported fake polls, details the subsequent release of other fake polls, and puts forward evidence pointing towards the falsity of the polls. It then concludes this Part by explaining how these fraudulent polls affect, or could affect, online betting markets.
How PredictIt Operates
In many ways, PredictIt operates like a miniature version of a stock market but with a much more limited number of "stock" options, all of which relate to the occurrence of a political event. (24) At any given time, PredictIt provides several hundred "markets" for users to purchase shares in. (25) The markets operate by asking a question--like, "Will Donald Trump be the 2020 Republican nominee for President?" (26) or "Who will win the 2020 Iowa Democratic caucuses?" (27)--and then by allowing users to purchase an answer to the question as a "share." (28) Some of the questions present only a binary decision for users--they can buy "Yes" or "No"--while others allow users to buy one (or more) of several different options. (29) The prices are set both by PredictIt's market trends (30) and by users themselves. (31)
To purchase "shares," (32) users must create an account and deposit funds onto the website through PredictIt's designated clearing house. (33) The clearing house holds users' deposited funds in trust in a clearing account, and it maintains a separate ledger for each user's funds. (34) Users purchase "shares" for a particular outcome--for example, predicting that Senate Candidate A will win her election--each of which costs less than one dollar. (35) Users can sell their shares at any time, or they can wait until an event has been completed. (36) If they choose not to sell, their shares will be cashed out with a value that depends on whether their prediction was correct. (37) If it was, their shares will be redeemed for one dollar each; if it wasn't, they are redeemed for nothing at all. (38) Regardless of the user's choice, after a sale has been completed, the user's profit (after subtracting PredictIt's 10% fee on profits) is deposited in trust with the clearing house. (39) When a user wishes to withdraw funds, the clearing house subtracts PredictIt's 5% fee on withdrawals and credits the user for the remainder. (40) As prerequisite conditions for its no-action letter, the CFTC requires PredictIt to "have a limit of 5000 total traders in any particular [election market]," with a "limit on investment by any single participant...