Comment Text:
I would like to protest the shutting down of PredictIt for the following reason
1- PredictIt has inspired, for me, the way that I invest money and donate to candidates. For example, when Roy Moore, a disgraced federal judge (removed from the bench twice) was accused of trying to hit on 16 year olds, I didn’t think it mattered because Alabama would never ever elect a Democrat to the senate. When the betting market, which generally is far more measured than a reporter saying things for clicks, said it was possible for Doug Jones to beat Moore, I donated money to a Democrat for the first time in my life. The predictive model helped inform who I donated to.
In 2020, I based my decision on who to vote for in the Democratic primaries, because PredictIt told me that Joe Biden’s implied odds of beating Donald Trump were higher than Bernie Sanders’s. The predictive model helped me see who I wanted to vote for.
Alternatively, election years create lots of uncertainty and generally you see restaurant spending and other luxury spending go down in election years. I refer to PredictIt to determine how comfortable I feel in my long term security, how I invest my actual money (green vs. coal depending on candidates) and even how I feel about going out to eat twice this week.
2- The allegation that this is luck-based gaming and not skill and knowledge derived, and non-predictive is silly when you consider that VUW, the parent university, isn’t profiting, and only bringing in enough to offset cost. What is, then, this research institution’s game, if not for research, if they aren’t profiting? Sure, the people operating PredictIt are drawing salaries, but VUW sees some reason to continue to stay involved. What is it? Surely, it has to be seeing the value of the academic research. If I can make practical decisions based on PredictIt, surely the professional researchers at VUW can as well.
3- I would also argue that this isn’t a luck-based game, but rather a game of skill and knowledge. Commonwealth v. Dent found that poker wasn’t illegal, because Poker was a game of knowledge and skill. (Yes, I know it was overturned, but only because the appellate court found that luck was integral to poker, NOT because it wouldn’t be illegal if it were a game of skill).
PredictIt has run some markets before that, for every intention, looked like gambling. How many times will Donald Trump tweet this week? Will he use the word “Emails” in the State of the Union address? Even if these somehow had academic merit, they’re still luck-based (though, my understanding was that that was part of the reason for the $850 limit). I can’t pretend that those markets had any real skill component to them.
But the majority of PredictIt would fall into the realm of a Dominant Factor Test
– Participants must have a distinct possibility of exercising skill and must have sufficient data upon which to calculate an informed judgment.
– Participants must have the opportunity to exercise the skill, and the general class of participants must possess the skill.
– Skill or the competitors efforts must sufficiently govern the results
– The standard of skill must be known to the participants, and this standard must govern the results.
– Markets like “Who will win the Democratic Primary” have zillions of potential factors. The value of endorsements, the value of momentum and the order of states, understand the immediate and long term effects of debates and fundraising. Everyone has a model, whether understood or on a spreadsheet, of how it works and why.
But the candidates that win aren’t randomly chosen. You didn’t just get an ace on the river. Joe Sestak was never going to become president. Those factors that are in place matter. And the better your model, the more likely you are to win. Understanding polling models is pivotal. Someone who has a deep understanding of the history of ActiVote vs. RCP vs. 538 vs. Marist College will do better than someone who saw one poll they liked and got cocky.
There’s lots and lots of skill involved, and an almost unlimited amount of data. Barring something shocking like a sudden death, the biggest reason that it might look like luck is because there’s so much information that it’s overwhelming and impossible to comprehend it all. Good historians never say “and then suddenly…” Understanding, culturally, what is happening and what will happen as a result is a skill.
– Of course, participants have the opportunity to use that skill, to predict trends and outcomes. This seems like the least contentious point.
– Yes. This isn’t 100%, of course, things happen. But people who understand statistics, voting patterns, weather patterns, the behavior of swing states, and the effects of actions on fundraising will do substantially better than people who don’t understand that stuff. People who read Q Anon and bet accordingly will do worse than people who regularly bet against Q Anon theories. Sometimes surprises happen, but a good understanding of politics and narrative and media will mean you win more than you lose. It’s pretty easy. Elections aren’t just the random scattering of dice.
– Participants on PI know that there is lots of knowledge out there about elections, politicians, and polling. It’s almost impossible to not know that. I’m not sure to what degree members of PI understand which indicators are more important than others (polling vs. media narrative vs. Q’s predictions) but they absolutely understand that this knowledge dictates the outcome. No sane person on PredictIt can look at the current situation and think that Mitt Romney will win the 2024 presidential election. Why? Because it’s not chance. Whatever of those metrics you use, however you acquire knowledge, whatever model you use, knowledge and skill and insight will tell you that it can’t happen.
5- I believe, absolutely, that PredictIt serves a societal good and has helped inform me. Both through how I understand politics and how I vote/donate/engage with politics. If the site is closed in February of 2025, I’ll be sad but I’ll remember it fondly.
Shutting it down between now and inauguration Day has some extra potential damages. PredictIt does this thing called “negative risk.” It’s actually pretty nifty when you’re not worried about the website’s future.
But imagine you walk up to one of the contestants at the podium on a Jeopardy! game and bet them a dollar that they’ll lose. They take it. You may or may not win a dollar. Then you go to the second contestant and bet a dollar they’ll lose. Now, if one of them wins, you’re looking at even. Now, imagine you walk up to the third person and make the same deal. Suddenly, you’re guaranteed to lose, but… you’re still guaranteed a dollar. Since you’ll win two and lose one.
PredictIt’s system is smart enough to realize that you did it and give you the dollar right there. Certain people on PredictIt really like betting no on everything, with the hope of reaching a risk-free stage like that. I personally had negative-66-dollars put into the Republican vice-presidential market.
There’s only two “fair” ways to liquidate if ordered to on the spot.
One is to cash out at the current market value. But suppose I predicted Kamala Harris will win the presidency at 58 cents. And then she drops to 50. And then it’s ordered to cash out. And she wins. I’m personally standing to lose money on something I’m right about because the website was told to pull the rug out from under my feet. That’s obviously, on the face of it, unfair.
The other alternative is to refund everyone. But remember the risk-free position that led me to having -$66 into a market? If told to liquidate at the time… PredictIt would be in a position where they’d have to tell me to give $66 back. And they’d either take it from the money I’d then have trapped in my account, which I’d consider theft, or send me a bill, which would make no sense.
If PredictIt has to go away, I absolutely believe there’s no way to do that fairly between now and the end of this election cycle.
I understand that PredictIt was offered a NAL and, probably, took massive liberties in regards to the terms involved. Particularly in the form of those Tweet markets (which were dead before Trump was banned from Twitter). But PredictIt is a societal good that lets you understand the climate and what’s happening. And it’s not a game of chance, and I think that there’s a real loss to research and to common sense if it’s shuttered. And I’m asking you to please not do it.
