Please consider donating to the Open Medicine Foundation to help people suffering from extremely disabling and underfunded lifelong illnesses with no know treatment.

https://www.omf.ngo

  • 166 Posts
  • 1.02K Comments
Joined 4 months ago
cake
Cake day: July 17th, 2024

help-circle
  • Yes but these are definitely hypotheses. Fivethirtyeight has extensively covered it. In 2016 they covered it a lot, did a post-mortem on the election, and said the industry said they’ll fix it for next time, then in 2020 we got even worse polling errors.

    I miss Nate from the fivethirtyeight days. It’s sad to see his model paywalled.

    Random but if you’re a polling nerd this site is a godsend: https://swingstates.vercel.app/

    its based on the 538 API, they also have a mastodon and bsky bot

    This cycle a lot of polling has oversamples GOP. And they are using a new technique wherein they weight polls based on previous vote. Which is interesting I guess. Anyways, we’ll see what happens in a week.




  • EDIT: I’m stupid and did it with the fine squaring every week instead of doubling and I’m too lazy to recalculate so ignore this

    w1 = (100’000 x 7)2

    w2 = (100’000 x 7 + w1)2

    wn = (100’000 x 7 + wn-1)2

    It looks like it’s impossible to find a true function for wn that depens only on n.

    But we can approximate it with:

    w_n = C^(2^n) 
    

    where C is the money owed at the end of the first week.

    So the approximate formula for how much is owned at the end of the nth week is.

    w_n = (4.9 * 10^11)^(2^(n-1))
    

    where n is the number of weeks since the fine was issued.

    In truth wn will be larger than said number but it’s a decent lower bounds for approximation and it should be accurate to within around a couple percent.

    i did this calculation in rubles but you can just replace 100’000 by 1000 if you wish USD.