7 ms·

Benford's law only applies to numbers drawn from a distribution with a wide spread over multiple orders of magnitude. So, financial transactions in a big compan

Benford's law only applies to numbers drawn from a distribution with a wide spread over multiple orders of magnitude. So, financial transactions in a big company would show this. Votes in US districts, not so much. So there, we've spoken about it. (I think the technical term for the distribution is scale-free but I could be wrong. The law follows reasonably intuitively from thinking about the fractional part of logarithms.)

collyw 6y ago
Wasn't it used in an Iranian election as evidence of shenanigans? >So there, we've spoken about it. You seem to have given your opinion without wanting anyone else to give theirs. Why are you so against discussing it?
- Robin_Message 6y ago
  Yes, from the paper "A first-digit anomaly in the 2009 Iranian presidential election" (https://arxiv.org/pdf/0906.2789.pdf https://arxiv.org/pdf/0906.2789.pdf): "The total numbers of votes per voting area in the MOI’s data vary from about 10^4 to 10^6, i.e. by two orders of magnitude (powers of 10). This suggests that Benford’s Law may be applicable" The issue, as I understand it, is that the US districts that the "law" is being applied to are too small, uniform, and uniformly sized, and therefore the "law" doesn't apply. Additionally, many of the graphs being shared have obvious biases (change of scale between one candidate and the other) which suggests bad faith of those making the graphs (see the skeptics.stackexchange link a sibling shared). > Why are you so against discussing it? I'm not especially against discussing it. I am against presenting it as a fait accompli that the election was rigged, when thirty seconds of Googling and critical thought show it to be a red herring.
  - forest_dweller 6y ago
    Most of the images that involve the graphs are just memeing. I have a Discord with a bunch of users and for want of a better term people are just shit posting them.

collyw 6y ago

Wasn't it used in an Iranian election as evidence of shenanigans? >So there, we've spoken about it. You seem to have given your opinion without wanting anyone else to give theirs. Why are you so against discussing it?

Robin_Message 6y ago
Yes, from the paper "A first-digit anomaly in the 2009 Iranian presidential election" (https://arxiv.org/pdf/0906.2789.pdf https://arxiv.org/pdf/0906.2789.pdf): "The total numbers of votes per voting area in the MOI’s data vary from about 10^4 to 10^6, i.e. by two orders of magnitude (powers of 10). This suggests that Benford’s Law may be applicable" The issue, as I understand it, is that the US districts that the "law" is being applied to are too small, uniform, and uniformly sized, and therefore the "law" doesn't apply. Additionally, many of the graphs being shared have obvious biases (change of scale between one candidate and the other) which suggests bad faith of those making the graphs (see the skeptics.stackexchange link a sibling shared). > Why are you so against discussing it? I'm not especially against discussing it. I am against presenting it as a fait accompli that the election was rigged, when thirty seconds of Googling and critical thought show it to be a red herring.
- forest_dweller 6y ago
  Most of the images that involve the graphs are just memeing. I have a Discord with a bunch of users and for want of a better term people are just shit posting them.