Hacker News Reader

Comment by Terr_ | Hacker News Reader

> letting people learn what's happening in the world around us

I think that phrasing is a bit too optimistic. Even in cases where the "prediction" cannot influence the outcome, the primary "learning about the world" involves the imputed opinions of bettors.

Betting at horse-races doesn't teach you nearly as much about horses as actually going to the stables. :P

View on HN

Err, checking the line on a horse race is a much more reliable way to tell who has what chance of winning than going to the stables as a random person, and most likely for a professional as well.

Excellent point. I’d go further and say it’s likely to be the single best information, much better than consuming media about horse racing.

My point is that the "learning about what is happening in the world around us" part falls flat when what comes next is "by finding which person to copy test answers from."

I mean, yeah, you technically learn something, but the limitations undercut the sales-pitch.

If there is a better predictor than what is implied by the market then someone could make money on it. Hence the odds tend to be pretty accurate (with enough volume) because obvious edges like that tend to go to zero. A good example of that was the last election where people with access to private polls betted heavily and shifted the market one way.

This comment appears to be downvoted for ideological reasons? To me it seems manifestly true. The comment isn’t saying the market is perfect, just that it tends towards the best estimate because of incentives.

These imputed opinions tend to do a better job than traditional news. On election night, polymarket was ahead of the news channels every step of the way.

Edit: to be clear, I’m referring to polymarket essentially “calling” each swing state well before the networks did. I’m not just referring to the overall outcome.

I am not sure what that has to do with calling elections. The Associated Press calls who wins, not some random person with $5 to throw away.

The AP does not in fact call who wins. They publish their opinion, some people trust the AP to be accurate (or not!). The election is only "called" when all the states have certified their results and the US Congress has accepted those certifications. That happens early in the new year.

Polymarket had each state at above 95% likelihood of trump well ahead of the networks suggesting he was winning each state. It’s not just calling winner, it’s ingesting information faster

Making these statements after the fact is tricky.

Very tricky.

I would be interested in a blog post explaining it.

The classical fallacy is attributing skill to someone who has successfully predicted 10 coin flips (or market trends) in a row, ignoring the fact that there were many other people making different predictions and there was always going to be one of them who was successful.

Of course this is true, and the observation is underpowered. However, on election night it wasn’t just about being right it was about digesting information in real time and promulgating information faster. For example, polymarket had trump at like 99% likelihood of winning when the networks were still playing “horse race”.

DDHQ was also significantly ahead of the large media, and seemed to be what lead some of the moves in the betting markets. Seems like anyone at DDHQ with the fore-knowledge of the race calling could have easily front ran the betting markets. Same with Nate Silver and his daily releases. In the back of my mind I also wonder about that last minute Iowa pollster who released an outlier poll.

Any the folks running the NY Times needle.

Reply on news.ycombinator.com