Your submission is now in Draft mode.

Once it's ready, please submit your draft for review by our team of Community Moderators. Thank you!

Submit Essay

Once you submit your essay, you can no longer edit it.

Pending

This content now needs to be approved by community moderators.

Submitted

This essay was submitted and is waiting for review.

Will Metaculus's predictions significantly improve in the next 6 months?

Question

One of the goals of Metaculus is to produce well-calibrated probabilistic forecasts. According to the Metaculus FAQ: "Like many mental capabilities, prediction is a talent that persists over time and is a skill that can be developed. By giving steady quantitative feedback and assessment, predictors can improve their skill and accuracy, as well as develop a quantified track record." So, as the size and experience of the Metaculus community increases so should the quality of the predictions generated by it.

Recently, Metaculus assembled an interactive display of its track record so far. Several metric are given, including the Brier score, which is essentially the mean square difference between the prediction and the actual outcome:

where are the predicted probabilities and are the observed outcomes, equal to zero if the question resolved negatively and equal to one if it resolved positively. Brier scores range from zero to one and smaller scores are better, with the ideal score being zero and the worst possible being one.

As of question launch, the mean Brier score for the "Metaculus prediction" is 0.150. The Metaculus prediction was the same as the community prediction until June 2017; thereafter it began to use a more sophisticated aggregation and recalibration scheme that promises more accurate predictions.

But how much more accurate? The "Metaculus post-diction" applies the current aggregation method to old data (excluding each question from the data used for "training"), and should give an estimate of future performance if future questions and predictors are comparable to past ones. As of question launch the "postdiction" Brier score is at 0.10.

Over the next 6 months prior to Jan 23, 2018, of order 25-75 binary questions are likely to resolve, about 1/3 to 1/2 as many as have previously resolved. If the community (and the aggregation) is steadily improving, we'd expect the mean Brier score to steadily drop. We'll ask:

Will the mean Brier score of the Metaculus prediction on Jan 23, 2018 be lower than 0.13?

Resolution will be determined via the Metaculus track record page.

Make a Prediction

Prediction

Note: this question resolved before its original close time. All of your predictions came after the resolution, so you did not gain (or lose) any points for it.

Note: this question resolved before its original close time. You earned points up until the question resolution, but not afterwards.

Current points depend on your prediction, the community's prediction, and the result. Your total earned points are averaged over the lifetime of the question, so predict early to get as many points as possible! See the FAQ.