How Inevitable is Biden’s Nomination, According to Manifold Superforecasters?
A politics piece by Jacob "Conflux" Cohen
Hello and welcome to this experimental first edition of Above The Fold: Manifold Politics! (Well, first since the midterms, anyway.) As of late, Manifold’s politics arm has been building itself into a premier platform for viewing electoral probabilities. In this newsletter, we seek to supplement that with analysis, contextualizing a current news event with markets along with commentary from Manifold’s best traders. Right now, it’s unclear what the future format or frequency will be, so comments are extremely appreciated.
I’m Jacob Cohen, known as Conflux on Manifold, where I’ve had many roles (including site moderator, mana grantmaker, and cohost of the erstwhile Market Manipulation Podcast). A longtime avid political forecasting junkie, I ranked #3 in Manifold’s 2022 midterms prediction tournament. Nevertheless, I wouldn’t consider myself in the very top tier of political intuitions on Manifold: I’m currently ranked #18 in the politics category and #57 overall.
With the intro out of the way, let’s get right into today’s topic!
What Happened?
Just two weeks ago, a highly liquid Manifold market assigned Biden a 96% chance of being renominated in 2024. This makes sense, given that he faces token opposition and won the New Hampshire primary in a landslide despite not appearing on the ballot.
However, on Thursday, February 8th, a recent special counsel report characterized Biden’s memory as “hazy” and “poor,” and said he’d try to come across as “a well-meaning elderly man with a poor memory.” In a press conference given by Biden to dispel this image, he conflated Egypt with Mexico. In general, many voters worry about Biden’s age: a recent NBC poll found that his “age and fitness top the list of voters’ concerns,” with 76% worried about his age, while an ABC poll had 86% saying he was too old.
In the days after the report, the market dropped to 93% and then 90%.
Amid other discourse, on Friday the 16th, Ezra Klein published a New York Times editorial titled “Democrats Have A Better Option Than Biden.” In it, Klein argues that while he thinks Biden has been a “good president,” and is still “sharp in meetings,” he nevertheless is “faltering in his campaign” as a result of his aging. Klein advocates that Biden step down and let delegates at the convention decide who would win — with the reasoning being that Biden is losing, “behind in most polls,” and “there is a ton of talent in the Democratic Party right now.”
Yesterday morning, FiveThirtyEight founder Nate Silver wrote a blog post backing up this view, saying that he felt Biden dropped from a 65-35 favorite to “probably the underdog.” He also states that “if the past couple of weeks are any evidence, [a convention switch] might nevertheless be Democrats’ best option for beating Trump,” though he isn’t definitive on the question of whether a switch would improve the Democrats’ chances.
The market now gives Biden an 89% chance (up from a brief low of 86% after the Klein article). And indeed, bettors don’t think that Biden will lose in the primaries to a declared candidate like former gelato executive Rep. Dean Phillips (his chance is under 1%). Rather, they think that Klein’s Biden-stepping-aside scenario is the most likely, and just maybe Biden will be replaced by a candidate who’s not currently running: perhaps Gavin Newsom, Gretchen Whitmer, Michelle Obama, or Kamala Harris.
What Does Manifold Think?
So why does Manifold assign credence to this theory? Isn’t Ezra Klein just some reporter? Not according to superforecasters.
Joshua (ranked #1 in the politics category on Manifold and #5 overall, he’s the new Manifold Politics Editor-In-Chief) stated, “A brokered convention is usually dismissed as an insane idea, but if I was going to pick one guy to talk the democratic establishment into it I would pick Ezra. So I'm de-risking, because I think if the right people read this article and find it as persuasive as I did, they're going to start talking about it seriously behind the scenes … I still think that’s only about a 10% chance. But we’re at 90%, so I’m selling.” Despite this sell, he currently maintains 87,269 YES shares on the main market.
SemioticRivalry (ranked #2 in politics, #9 overall, 91,500 NO shares) agrees, mentioning that “Ezra is extremely well connected amongst bigwig dems and he just wouldn't be saying this if there wasn't a large faction who agreed,” and noting the unusual writing style: Klein name-drops several Democratic insiders (Barack Obama, Chuck Schumer, Mike Donilon, Ron Klain, Nancy Pelosi, Anita Dunn) who “need to get [Biden] to see this.”
Marcus Abramovitch (the #1 overall leaderboard trader, 147,340 YES shares), meanwhile, said “People be overcorrecting here. I agree Biden is a senile old man who has no business being the nominee. But there is massive status quo bias and large incentives at play for influential people in the democratic party not to change course.” Shump called it “Manifold’s Gamestop moment.”
It’s worthwhile to note here that Manifold isn’t always the best with extreme probabilities. This was noted especially with the superconductor market, which has remained above 2-3% for months despite basically every top leaderboard trader betting NO. The amount of capital investment required to push markets below 5% or 95% can be very large, and so “dumb money” can have an outsized influence.
The market can also be moved by a few wealthy Manifold traders. Joshua quips that “In practice, the market moves down when Katja and Nathan buy No and it moves up when JRay buys Yes,” referring to a few large holders on the market. Still, I consider the downward shift from 96% to 89% to be fairly significant. I think it’s fair to say that the chance of Biden not winning has approximately doubled in the last two weeks.
What About Other Platforms?
Compared to Manifold, real-money markets have always been much more Biden-skeptical. Polymarket gives him a mere 78% chance to be the nominee, PredictIt has him at 75%, and ElectionBettingOdds assigns 73%. Meanwhile, non-market-based forecasting platform Metaculus is a bit higher than Manifold, with a 92% prediction.
There’s been a lot of other discourse on why this is, so I won’t go into too much detail. If you’re curious to read more, there’s a good Marketwise post discussing potential reasons, an Asterisk Magazine article on election market inaccuracy in general, a Manifold market with lots of discourse and a dubious multiple-choice mechanism, and an Astral Codex Ten post with the line “I do think you could probably make a lot of real money in expectation buying Biden [for the general election] on Polymarket - if you were a degenerate gambler with a VPN, of course.”
Other Questions
Even NO bettors on Biden agree that this brokered-convention scenario is a longshot. So are there any related propositions that will provide indications as to how close we are? Well, there is a Manifold market on “Which prominent Democrats will say that Biden should not be the nominee, before August 1st?” Currently it has a mere 13 traders, but currently the only Democrats whose odds are above 20% are those in the “Squad,” with Rep. Ilhan Omar topping the list at 62%. (Rep. Rashida Tlaib has already called for Democrats not to vote for Biden in the primary.) Maybe you all can make that market more accurate! (I’ve just given it a M$1,000 subsidy.) Another question is whether Ezra Klein will change his mind, which is currently priced at 17%.
These will be helpful indicators to watch as we determine how much this brokered-convention scenario is linked to reality.
We can also ask: if Democrats listened to Klein, would they have a better chance of winning the general election? In other words, are these Biden alternatives more “electable”? One could imagine that the answer is yes, due to Biden’s various problems, but one could also imagine that the chaos resulting from a Biden replacement would damage the unity of the Democratic Party. Or perhaps the last-minute nature would leave a candidate who has a devastating scandal or skeleton in their closet, which the normal vetting process didn’t discover.
This is a tough question for prediction markets, because the ~10% likelihood of all the other scenarios combined means that a question like “Would Michelle Obama win a general election against Trump?” is unlikely to resolve definitively to YES or NO — rather, you have to resolve it N/A (which means that all bets are reversed) or PROB (which makes the market turn into a contest of who can spend more mana). Both of these reduce the incentive toward correct predictions. Another option is a massive market showing the probabilities of all the outcomes. It currently states that there’s a 1.3% chance that Gavin Newsom is the nominee and wins, and a 0.8% chance that he’s the nominee and loses, implying a 62% chance that he would win. The problem: for 10 mana, I could change this implied percentage to 18%, so it’s highly non-robust.
This morning, Manifold Politics made a dead-simple market trying to remove the difficulty of factoring in all the other candidates. The question: if anyone but Biden is the nominee, will they win the general? So far, not a ton of traders; you are encouraged to bet!
Switching from market mechanisms to pure opinion, a poll of Manifolders shows Whitmer, Michelle Obama, and (of course) Taylor Swift as particularly electable options.
On the Republican Side
Meanwhile, the markets give Donald Trump a 98% chance of winning the primary in Nikki Haley’s home state of South Carolina, and a 95% chance of winning the nomination overall — even with a 51% chance of a felony conviction by November.
The “veepstakes” are much more uncertain. Rep. Elise Stefanik and Gov. Kristi Noem are the most likely running mates for Trump right now, with Vivek Ramaswamy and Sen. Tim Scott as third- and fourth-likeliest, but the markets aren’t confident at all: there is a “long tail” of additional possibilities.
California Senate
California, home state of Manifold Markets, has a truly contested Senate primary for the first time in years. The state uses a “blanket primary” system in which all candidates run in the same primary, and then the top two finishers (by plurality) advance to the general. Currently, it seems like Democratic Rep. Adam Schiff is almost certain to advance, while it’s a close race between Democratic Rep. Katie Porter (perhaps a slight favorite) and Republican former baseball player Steve Garvey for second place. If Garvey won, he would almost assuredly lose to Schiff due to California’s strong Democratic lean, but Manifold thinks a Schiff-Porter general election would be a tossup (maybe 55-45 or 60-40 favoring Schiff).
Hopefully one of these days, Marcus will finally add “Steve Garvey” as an option to his market, but right now “Other” should basically be interpreted as “Steve Garvey”...
One Last Market
One tradition I want to start in this newsletter is a shoutout to a market that hasn’t received many traders on a fun/lighthearted/interesting topic. Today’s pick investigates the correlation between the 2024 election and the financial situation of the iconically overleveraged Manifold user Tumbles, who currently has around M$700,000 due in loans:
I’ve personally added a M$1,000 subsidy to the above market to encourage accuracy.
Concluding Thoughts
Thanks for reading to the end of this edition of Above The Fold: Manifold Politics! As I said at the top, I’m not sure when the next one will be, but we’d love any feedback on the format: what would you like to see from a Manifold Politics article that would make it more useful? Would you appreciate a posting schedule (perhaps monthly, biweekly, or weekly), or a looser “write when there’s something to write about” attitude? I’ve made a normal market and a bounty market below about the future of this newsletter.
I’ve been Jacob “Conflux” Cohen; blog at tinyurl.com/confluxblog, puzzles at puzzlesforprogress.net. Check out Manifold Politics at manifold.markets/elections, or via its Politics or 2024 Election categories. Thanks to Joshua, nikki, and Gabrielle for proofreading, and to Saul Munn for “sponsoring” this newsletter.
Bye for now!