Evan Miller Profile picture
Statistically inclined software developer, occasional blogger about math + stats stuff. Working on evals @AnthropicAI
Nov 10, 2020 9 tweets 2 min read
I guess I'll be live tweeting the #AppleEvent in a few minutes because why not.

First a little background about why I think could be a pivotal moment... The story today will be the long awaited arrival of "ARM on the desktop". If the performance numbers are good, this will be the beginning of a shift to laptops with low-power ARM chips. Consumers can expect longer battery life and potentially built-in cellular comms a la iPad.
Nov 9, 2020 20 tweets 5 min read
All right folks: Another day, another voter fraud thread to debunk. 👇

I call this one The Case of the Limited Precision Data Feed.

(Thread)

The argument comes an "anonymous data scientist" (a.k.a. 4-chan) via @APhilosophae, and has a couple of components.

By plotting batches of votes in the NYT data feed, you can see a nice smooth line emerge from the chaos... Image
Nov 7, 2020 7 tweets 2 min read
Easiest way to explain Benford's Law to someone who knows nothing about stats, but understands compound interest:

Imagine watching a savings account grow from $10 to $100 over 10 years, and writing down the first $ digit every day... After ~3 years, the account will double in value to $20.

After ~6 years, the account doubles in value again, and now has $40.

After ~9 years, the account doubles in value once more, and now has $80.

Finally, in the last year, the account grows from $80 to $100.
Nov 6, 2020 15 tweets 4 min read
There's a meme going around stats Twitter that "Biden's vote tallies violate Benford's Law", implying numerical fabrication of some sort. Gather round folks, it's time for another stats lesson... I'm going to look in particular at this analysis of Milwaukee data by @statsguyphd :

Mar 3, 2020 15 tweets 4 min read
All right folks, quick stats lesson. (Thread)

There's a meme floating because of this article 👇 that "China must be faking their Coronavirus data because you never see R^2 = 0.99 with real data".

barrons.com/articles/china… If you run a regression of cumulative deaths reported in China in the first half of February, and look at the residuals, you get a magic parabola like this: Image