The public debate for who should be the Democratic Party presidential candidate for the 2020 electoral race hinges on whether the Dems are likelier to win by nominating a progressive or socialist, or by nominating a centrist or moderate. Assuming this is a problem one wanted to approach epistemically (as opposed to just viewing the debate as a power struggle), we should look for facts of the matter about whether, overall, Americans are likelier to elect a progressive/socialist, or moderate/centrist to the Presidency in 2020. Unfortunately there is no apparent consensus about how to make traction on this problem.

  • Having read a variety of sources for months now, left-leaning media outlets give all the usual arguments for why the Dems must nominate Sanders or another progressive to win, and all the more moderate media outlets say the exact opposite. Ideally, some people from both sides would come together for an adversarial collaboration, but it doesn't appear either side can even agree on what would count as good or bad evidence for their positions. In other words, most of the time on both sides of this issue are playing reference class tennis, and neither side is aware this is what they're doing.
  • On social media I started a discussion on this topic among some rationalists and effective altruists to see if it would elicit hope to make traction on this issue. None of them who had a strong opinion on this issue could or would back it up at length, which I'm assuming means they don't have strong, asymmetrical evidence (at least not in the eyes of the opposition) for their position. It's my expectation two rationalists should be aware when they're playing a game of reference class tennis. If on this topic they are aware, it doesn't appear rationalists know how to get out of this failure mode better than anyone else.
  • My search of last resort, at least under the streetlight (read: my attempt to answer this question without doing any work myself), was to check PredictIt. As of this posting, the top 6 contenders have been assigned between a 9% chance and 23% chance of winning the Democratic nomination. In roughly the same rank order for chances of winning the Presidency overall, those same 6 prospective Democratic nominees are currently receiving an assignment of between a 6% and 19% chance of winning. These 6 contenders run the gamut from Beto O'Rourke and Joe Biden,
    commonly regarded as the most moderate/centrist prospective Dem nominees, to Bernie Sanders, whose policy platform is generally furthest to the left among the lot of them. Leading the pack is Kamala Harris who has announced her presidential run and is perceived as somewhere between centrist Dems and democratic socialists like Sanders, with Sanders and Biden tied for 2nd (both are yet to publicly declare if they'll be running in 2020). PredictIt's predictions for which candidate will fare best are strongly correlated with who is regarded as being the strongest contender as a personality. They appear to have little to no correlation with the candidate's perceived ideology. The high level of apparent uncertainty in the predictions is likely an artifact of it being so early in the presidential race, as opposed to the prediction markets pricing in how likely the candidates' policy platforms are to appeal to the median voter.

In other words, if there is solid info out there which could best predict whether, in general, a leftist or centrist Democrat is likelier to win the Democratic nomination and/or the Presidency in 2020, it doesn't look like this has been priced into PredictIt's predictions yet. So to answer this question, we might have to find info prediction markets haven't yet, or that they may have at least overlooked.

Suggestions for how to answer this question:

  • I'm not looking for the kind of evidence the public, for whatever value of "the public," would necessarily agree would resolve this debate. I'm looking to answer this question for my own curiosity. I.e., if you think people are generally too mind-killed on this subject to notice good evidence even when it's staring them in the face, don't worry about that. Please optimize for sending me what you consider the best empirical info on this subject my way without consideration for how anyone else might take it.
  • Uncovering the relevant info could be easier than it appears. Maybe the data I'm looking for isn't a needle in the haystack. For all I know there is some excellent poll that perfectly answers this question, but nobody in the media cited it because the results didn't make for a sexy clickbait headline.
New Answer
New Comment

2 Answers sorted by

paul ince


Perhaps this series by Ben Hunt will help you decide if there is enough centre to support a centrist. I doubt it.



Would you need to do anything super fancy to beat the market? You could in thoery just find 100 median voters who swap parties, tell them both Trump and the newcomer are pedophiles to simulate attack ads and ask who they would vote for after giving them an entire minute to think about it(to simulate the months of them hearing about it). Repeat for each canidate.

Alternatively you could check polling numbers for swing States to filter out the noise of who wins calinforna for free.

That doesn't tell you a lot of important information. It doesn't tell you how much a candidate will motivate people to get to the voting booth on Tuesday vs. staying home.

It doesn't tell you how much grassroots supporters they get that campaign for them.

I agree. I think checking some polling numbers as Monkyyy suggested is a good starting point. It's right it doesn't tell us enough about the dynamics of mobilizing those who typically don't vote; and the dynamic between how much Republicans and Democrats can cause others to flip sides.
This is especially true given that the case for not nominating a centrist is to have stronger motivation of the base.
5 comments, sorted by Click to highlight new comments since:

What you want from a prediction market is not the chance of a given candidate winning the presidency, but the chance of a given candidate winning the presidency if they win the nomination. So, for each of the listed democratic candidates, take Predictit's probability that they win the presidency and divide that by Predictit's probability that they win the nomination: P[presidency | nomination] = P[presidency & nomination] / P[nomination] = P[presidency]/P[nomination], ignoring the chance that someone gets elected president without winning the nomination.

Just looking at the most recently traded prices, I see:

  • Harris: .19/.24 = .79
  • Biden: .12/.16 = .75
  • Warren: .07/.10 = .70
  • Sanders: .10/.15 = .67
  • Brown: .07/.11 = .64
  • O'Rourke: .08/.13 = .62
  • Booker: .06/.10 = .60

That said, the price spreads are ridiculously wide and the trade volume is a trickle, so the error bars on all of those implied probabilities are huge. We'll probably get tighter estimates this time next year.

Thanks for laying that all out. It's more helpful info out of PredictIt. Unfortunately it doesn't appear to point to a trend about whether a more progressive/leftist candidate, or moderate/centrist candidate, would be likelier to win. If I had to guess, I'd say Biden and Harris are more moderate candidates...? But that's not based on much, and like you said the range on all those estimates is still huge.

It may be a wrong question. My model is that specifics outweigh generalities in elections - there's so much noise that it overwhelms any probability difference based on fairly minor ideological differences.

This does not answer the question, but it seems plausible to me that the leftist-centrist axis only has a very small impact on who is likely to win, which would be consistent with PredictIt's estimates.

I'm inclined to agree. I think the relevant axis is likelier to be Anti-Establishment <--> Pro-Establishment, within the Democratic Party. I suspect where someone would fall on that axis is mildly/moderately positively correlated with where they'd fall on a leftist-centrist axis. This doesn't tell us enough though. For example, I'd say Harris is perceived as almost as 'establishment' as Biden or O'Rourke, but she is apparently to the left of both of them on many policies. While we can _roughly_ tell how pro- or anti-Dem establishment a candidate is, this is less amenable to analysis because there aren't quantitative tools for looking at how 'establishment' a candidate is, and definitely not for the current Dem establishment. Such tools exist for evaluating one's place on the ideological spectrum. That political science is better at this I think is why by default most people are more comfortable thinking in terms of left/centrist than anti-establishment/pro-establishment. At least that's why I expect why the debate has defaulted as about the leftist/centrist axis. I imagine this will remain the dominant framing of the issue until much later in the Dem races, when the field has shrunk enough to make more precise evaluations more useful.