🏅 Top traders

The poll was only open for two days, no ping?

@ChurlishGambit I'm sorry you weren't pinged for the poll! I appreciate your feedback, and if (god help me) I ever run a market like this again, I'll try to make sure it feels fairer.

@ChurlishGambit Also, I read your review, and I also want to apologize for saying "casuals" in the blurb above — I revised the language slightly. When I was structuring the poll, I really wanted to reduce bias and increase accuracy. Everyone who bet on this market has a conflict of interest, and therefore is biased. (And I claim that people who are more active on Manifold also tend to me more accurate forecasters.) I'm now realizing from your feedback that I sacrificed a sense of fairness.

I'm curious — how would you have run this market if you were in charge?

@Conflux here’s my list of major changes

1. Trader ping - the entire rationale of trading platforms is harnessing that aggregates of people can predict surprisingly accurately - something that is a true proven fact. Every single attempt to narrow down or reduce the participants makes this accuracy worse. In fact, giving users currency diminishes the accuracy more than single vote aggregates

2. Simplify poll - a yes/no ratio is often better and more clear and also as per the studies the last point was referencing, more accurate than hunting the list of percentages. I do see the appeal in the granularity for something like this, so maybe I would have done every 10 percentage points.

3. Muuuch longer poll open time. That poll should’ve been open a length of time comparable to the market, maybe a quarter as long

4. Market closed during voting. I don’t know if it was, but I saw no market closed notification.

5. Extending poll on 29 voters. It’s absurd to close a poll based market on less than 5% of its traders. Frankly if I couldn’t get above 10% that’s on is own is warranting me considering an N/A resolution. YES holders here dominated NO holders by a ratio of close to 3:1. (69.2% of all holders). This implies a fair poll resolution of 75%, assuming people are betting with their belief relative to the current market value.

There are a few smaller things I would have done differently beyond, but these are the points I actually have issue with.

I personally would have voted 99.9 in the poll. I am quite active and was online through all poll days, just didn’t happen to check a low volatility market.

@Magnify Thanks for the specific actionable feedback! (the best kind) I can’t say I agree with everything, and it’s too late to change things now, but it’s valuable data.

I thought you were one of the commenters making the best case for YES, which is why (iirc) I quoted you in my evidence summary :)

@Conflux I don't think it's too late to change things, markets get re-opened or re-resolved all the time

@Magnify The first point doesn't apply when people are biased towards voting according to how they already bet. Ideally, you want the people who vote and the people who bet to be as different as possible. Otherwise, people just vote how they bet and they bet how they think everyone else will vote, which doesn't necessarily match their beliefs about the underlying question.

Given that the poll was being rounded to the nearest 25%, it doesn't matter that the sample size was small. The selection of who voted was far more important. The idea that we needed a more precise estimate is undermined by your suggestion that this should have been a binary poll. You're also making a common mistake in statistics in thinking that the ratio of the sample size to the size of the population it's drawn from matters very much.

@ChurlishGambit Fair! I guess I meant that there’s a significantly higher bar in my mind for changing things now.

@Conflux Sounds more like stubbornness than rationality, no? What would the harm be in re-running the poll with more people? Heck, you can re-run the poll without even changing the current resolution. If the results are the same, then you don't have to alter anything.

if the results aren't the same, then you need to question your conclusions & methodology...which sounds like the most rational thing, to me :)

@ChurlishGambit It's basically the people who lost complaining until they get the result they want. It biases the final result.

@ChurlishGambit I’m not going to ping traders for an additional irrelevant poll. But feel free to make your own market or your own poll, I’d be curious to see the results :)

@AndrewHebb I came out even, I don't care about the fake money

No ping for the poll is a little wild

@Magnify I'm sorry you weren't pinged for the poll! As I wrote on the other comment above, I appreciate your feedback, and if (god help me) I ever run a market like this again, I'll try to make sure it feels fairer.

My most highly traded market ever, beating The Market ... I often wished I hadn't made it, or had at least written criteria less vague. I really dashed this one off without thinking, never imagining it would get so big. I have a theory that poorly specified markets attract more bettors (I have plenty of other better markets) because people are more likely to think they're obviously mispriced.

Y'all seemed to hate the idea of poll-based resolution with a righteous fury. But I always felt uncomfortable (extremely early, in like the first day of the market; read the description, people) making up some number, because I didn't think I was an expert. Still, I didn't want the poll to be worthless and manipulated.

So I did a major review of the evidence and compiled it into the poll. I took a few steps to reduce manipulation.

• First, I filtered for people who were genuinely interested in forecasting, and had some patience — the title began with "[read description] Crowdsourced forecasting poll", and the description was super long and in a separate Google Doc.

• Second, I started off with an emotional appeal toward honor (which I do believe; I think Manifold can be the Wikipedia of the future, if we all keep our heads)

• Third, I put a catch at the bottom where I promised to throw out 0% or 100% votes, on the grounds that anyone voting that way was probably either crazy or manipulating (though only 3 voters did this so it only changed the final percentage by 2% and didn't affect the final resolution).

I got 29 votes. Most bettors on this market didn't vote (yep, I didn't ping the traders — I didn't hide the market, but I really wanted the input of people who enjoyed forecasting, not just more casual people with a stake in this market).

The average: 41%.

Rounds to 50%, so that's the resolution.

The market was on the wrong side of 50%! Probably because y'all didn't trust me, and thought I would make an unserious poll for unserious people.

But anyway, people absolutely do not agree on this. Look at this crazy bimodal distribution. The most popular single percentage, with 6 out of 29 votes, was 5%, followed by 60% (3/29) and 90% (3/29). No one picked 35%, 40%, 45%, or 50%, the three answers closest to the final average.

Maybe you're still wondering: how did I vote?

I was one of the 3 people who voted for 60% (which, note, would also have rounded to 50%). I wrote this reasoning after my evidence review but before posting the poll:

I'm sympathetic to the superforecasters who think there's no direct evidence even though you'd expect some to emerge, etc etc — the simple fact that BenS and Joshua and Eric bet the way they did updates me a lot — but I just think this could've happened in a lot of ways, and some of them don't involve public evidence or huge scandal (eg when Trump was young, or if it was something that technically counts as "sexual relations" but wasn't fully statutory rape, or if it was Ivanka). it seems like there is a lot of circumstantial evidence, and Trump is a major sex offender who has made comments saying children are hot. in a 57.5 year period, it only needed to happen once.

Ironic, that the market predicted me better than it predicted the poll. I wish you hope and happiness in 2026. May your markets predict what you think they predict.

@Conflux You expected people to be constantly checking the market to see the poll come up? You didn't select for "serious people," you selected for people that check the website (& this particular market) a lot.

Here's the resolution poll, which closes at 10PM PST on Jan 5: https://manifold.markets/Conflux/read-description-crowdsourced-forec

Dear traders: I have not forgotten about this market! While I have been busy with other matters, I have been thinking about how to structure a poll that will be as fair as possible while also producing the most accurate forecast.

I’m thinking about compiling a summary of the most relevant evidence / context / base rates to include in the poll. Let me know if there’s anything you think I should especially include, or if you have any other ideas for making a robust poll.

"when above the age of 20"
that's a pretty broad range, that's like any time after 1967, and he could have had one with a 17 year old when he was 21 ...

I don't think it's possible for one to fully resolve no this way

@AutopoieticErgodicity yes, I think the 21-17 (or similar) thing is underrated as a possibility

@AutopoieticErgodicity 67

@Conflux The age of consent in Pennsylvania is 16 and in New York is 17

@derikk Yes I realized this yesterday. 21-16 would count but not 21-17

@derikk Oh you’re right about mentioning PA! He was at UPenn then maybe?

If this resolves based on a poll, it's basically a Keynesian beauty contest unless you can get enough people who haven't bet to vote.