Who wants to help Cards Against Humanity f*** with Elon Musk?
www.cardsagainsthumanity.com
We got Musk off our land. Let's build a monument to remind him he's a sad little b**** with no friends.

Today, we’re announcing the Official Elon Musk is a Sad Little Bitch with No Friends Monument Design Committee. For just $10, you can join a movement of like-minded people working together to force a moment of introspection upon the world’s richest, ugliest little bitch.

  • phutatorius@lemmy.zipEnglish
    1·
    1 day ago

    it’s genuinely a better predictive model than what a statistician would be able to come up with

    OK, let’s see the evidence for that assertion.

    Simply by knowing the price of a contract, I can know with decent accuracy how likely something is to happen.

    A retrospective analysis of the contract prices versus actual outcomes would also be needed. Otherwise, all you’re seeing is an opinion poll conducted on a self-selected sample of people unusually prone to being greedheads. And you have no idea of the mechanisms that shape that market.

    • 5318008@lemmy.mlEnglish
      1·
      1 day ago

      OK, let’s see the evidence for that assertion.

      Using prediction markets to estimate the reproducibility of scientific research - PMC

      There’s one.

      But obviously it’s not a guarantee if you don’t know what you’re doing. We also have documentation on where these methods fall short, such as when it’s based on exponential growth, rare events, or something else that humans are bad at understanding. Infectious Disease Forecasting.

      But again, good statisticians are prepared to address those things.

      And you have no idea of the mechanisms that shape that market.

      That’s even better. If I don’t understand the mechanisms shaping the market, I won’t be able to correctly identify predictive variables for my model. In that case, I would be better off trusting the prediction market.

      Obviously it would be better if the person building the model understood the mechanisms… but that takes time and money. If you’re managing 30 models, that might not be worth it.