• ganksy@lemmy.world
    link
    fedilink
    arrow-up
    80
    arrow-down
    1
    ·
    1 month ago

    At least there’s a spoiler at the top of the article. He’s taking Harris and his L of the past 10yrs was Bush v Gore.

    • grue@lemmy.world
      link
      fedilink
      English
      arrow-up
      39
      ·
      1 month ago

      Yet more signs pointing to Trump’s path to victory being to cheat and coup instead of trying to win votes legitimately.

      • ganksy@lemmy.world
        link
        fedilink
        arrow-up
        3
        ·
        1 month ago

        Yes it’s hard to take the cheating, voter suppression, and coup-ing into account statistically. I wasn’t really advocating for the author or prediction. Just wanted those elements of the article when I saw the post.

    • Zaktor@sopuli.xyz
      link
      fedilink
      English
      arrow-up
      12
      ·
      1 month ago

      Lichtman argues he was right in 2000 because his system predicted the popular vote winner, but that means in 2016 he was wrong because Trump didn’t win the popular vote. He then tried to say the keys are now about predicting the electoral college winner, but there wasn’t any change in the keys. He’s just trying to redefine his targets to say he was right after the fact.

  • AbouBenAdhem@lemmy.world
    link
    fedilink
    English
    arrow-up
    42
    arrow-down
    2
    ·
    1 month ago

    If a hundred people try to guess ten coin flips, odds are at least one of them will guess nine out of ten. That doesn’t make them an expert.

    • Zaktor@sopuli.xyz
      link
      fedilink
      English
      arrow-up
      20
      ·
      1 month ago

      And that’s on coin flips. Many of the last 10 elections weren’t hard to predict.

    • PlantJam@lemmy.world
      link
      fedilink
      arrow-up
      6
      ·
      1 month ago

      Coincidentally the odds are that almost exactly one will guess 9 out of 10 correctly, and about four people will guess 8 out of 10 correctly. Odds drop to about 1 in 1000 for guessing all 10 correctly.

  • profdc9@lemmy.world
    link
    fedilink
    arrow-up
    35
    arrow-down
    1
    ·
    1 month ago

    In other news, they interviewed 200 historians and by chance found one that happened to predict 9 of the last 10 elections correctly.

  • eestileib@sh.itjust.works
    link
    fedilink
    arrow-up
    17
    ·
    1 month ago

    If you try to forget who the picture on the right is, and just look at him as a random person, he looks so fucking strange with that makeup and that skin texture.

  • LegoBrickOnFire@lemmy.world
    link
    fedilink
    arrow-up
    10
    ·
    1 month ago

    I don’t want to diminish the talent of the historian in vain but past performance is not always an indicator of being good at predictions, he might just have been lucky in a completely random guess 9 times out of 10. Given the amount of historians making such predictions it is not unlikely that such an historian exists and it would be fairly easy to mistake their success for talent.

    I don’t know where I found this but I found a scheme somewhere to scam some investors: Find a large list of potential victims. Tell one part of it that you predict that market will do A, and the other part that the market will do B. Repeat the process several times, selecting only the investors to whom you’ve always told the correct prediction. Eventually you will have a handfull of people who have “solid proof” that you are a visionnary and you can scam them.

    Again, I absolutely do not mean to say that this particular historian is bad. This story just reminded me of these ideas and I wanted to share.

  • TheFonz@lemmy.world
    link
    fedilink
    arrow-up
    5
    arrow-down
    1
    ·
    1 month ago

    I’m curious if at this point it would be possible to train an LLM on this type of estimation. But I don’t understand Ai really well or if they are even good at predictive work. Im going off of research that involved predicting disease (I think it was diabetes)

    • Leate_Wonceslace@lemmy.dbzer0.com
      link
      fedilink
      English
      arrow-up
      2
      ·
      1 month ago

      LLMs don’t handle booleans, and the 13 keys is an open statement, so the best you could do is train 13 neural networks to determine each of the keys, but you’d need a lot of data for that I suspect we simply don’t have.

      It’d probably be better to train a neural network to just output probabilities of each candidate winning based on specific information, like polling data.

    • willya@lemmyf.ukOP
      link
      fedilink
      English
      arrow-up
      2
      ·
      1 month ago

      Had to run this through ChatGPT and funny enough it sites this article in the first paragraph. It also has no idea about Kennedy dropping out and endorsing Trump.

      As of now, predictions for the 2024 U.S. presidential election suggest a tight race. Kamala Harris, the Democratic nominee, has been forecasted by election expert Allan Lichtman to win, based on his “13 Keys to the White House” model. Lichtman has a strong track record, having correctly predicted most U.S. presidential elections since 1984. He argues that Harris holds more favorable indicators than her main rival, Donald Trump, who is seeking a second non-consecutive term.

      On the other hand, some models, like those from Race to the WH, show a more competitive scenario, with polling and swing state dynamics still evolving. Trump’s ongoing legal issues and the emergence of strong third-party candidates like Robert F. Kennedy Jr. add complexity to the race.

      Ultimately, the final outcome will depend heavily on how these factors unfold in the coming months, as both candidates continue their campaigns

      Sources: US Presidential Election: Nostradamus of US polls predicts a Kamala Harris victory against Trump - India Today

      Predictions for the 2024 Presidential Elections - Live Forecast — Race to the WH

      2024 United States presidential election - Wikipedia

  • Fedizen@lemmy.world
    link
    fedilink
    arrow-up
    3
    ·
    edit-2
    1 month ago

    well given the odds of a know-nothing correct pick being 50% then at best one would expect a coin flip to do on average is 5/10 though there were maybe a couple easy calls in there

  • Media Bias Fact Checker@lemmy.worldB
    link
    fedilink
    arrow-up
    3
    arrow-down
    5
    ·
    1 month ago
    USA Today - News Source Context (Click to view Full Report)

    Information for USA Today:

    MBFC: Left-Center - Credibility: High - Factual Reporting: Mostly Factual - United States of America
    Wikipedia about this source

    Search topics on Ground.News

    https://www.usatoday.com/story/news/politics/elections/2024/09/05/historian-allan-lichtman-2024-election-prediction/75082875007/

    Media Bias Fact Check | bot support