Good Judgement Project Open (GJPO)

Discussion in 'Off-Topic Discussions' started by Dustin, Feb 21, 2021.

  1. Dustin

    Dustin Active Member

    Hi all,

    On the reading thread, I talked about the book Superforecasters. (

    I recently signed up with Good Judgement Open (GJPO), which is an online platform that duplicates what the original forecasters did, except now it's open to the public. You can make forecasts/predictions, assign them probabilities, provide your rationale and then see if you're right.

    Examples of new questions (they usually add new ones on Friday)

    I decided to answer the Ada cryptocurrency and the California recall election. The people who pose the questions also provide "buckets" for the answers. Some will be "yes" or "no", but most will involve a few options. For example, the Ada cryptocurrency provides 4 options:

    1. Less than $0.50
    2. Between $0.50 and $1.00, inclusive
    3. More than $1.00 but less than $2.50
    4. Between $2.50 and $5.00 inclusive
    5. More than $5.00
    While the question on Newsom provides three options:

    1. Yes and Newsom will be recalled
    2. Yes, but Newsom won't be recalled
    3. No [there will be no recall]
    If any of you are interested in participating, let me know and we can follow each other. They say providing feedback on each others answers (especially in areas like counterfactuals - what would need to exist for this prediction NOT to happen, what assumptions am I making, etc.) can be very helpful.
  2. SteveFoerster

    SteveFoerster Resident Gadfly Staff Member

    Is this an idea futures market, where one can bet on the outcome?
  3. Dustin

    Dustin Active Member

    No, because there's no money exchanged - just bragging rights. The book actually mentions the Iowa Electronic Markets (IEM) which is a futures market where you can exchange real money for the outcome of topics, mostly political (e.g. the party control of the US House and Senate or the outcome of elections.) Despite living in Iowa I'd never heard of it. It's run out of the University of Iowa business school:
  4. SteveFoerster

    SteveFoerster Resident Gadfly Staff Member

    This probably belongs in the "So, What Are You Reading" thread, but there's a sci-fi book in which idea futures are an important plot element. It's by Marc Stiegler and called "Earthweb":
    Dustin likes this.
  5. Dustin

    Dustin Active Member

    Because Brier scores can only be compared when looking at the same types of questions (e.g. a meteorologist wouldn't compare their weather forecasting with someone else predicting political events because those things are totally different), GJPO allows you to forecast in a number of different areas and then has leaderboards for each of them. Because questions can take a year or longer to close (and you get a better score for having a longer time window of having a "correct" prediction), I won't have a Brier score for a while.

    They do have an upvotes leaderboard though, for people to upvote each others forecasts. When I joined I was 98,000 or so, exactly at the end of the list. After receiving 6 upvotes, I've catapulted to #2381. The vast majority of the 98,000 haven't actually participated at all, or made a single prediction and then left. To hit the top 1000, you need 15 upvotes. But #1 on the leaderboard has 8013 upvotes. Quite a distribution.
  6. Dustin

    Dustin Active Member

    2 more upvotes and I've moved up to #1802 (can you tell I'm competitive.) I've forecasted on 7 questions that close in the next 12 months or so.
    SteveFoerster likes this.
  7. Maniac Craniac

    Maniac Craniac Moderator Staff Member

    Ah, yes. The old 80/20,000 rule.

Share This Page