News for hackers. Stuff that matters.
Re:Paper with 350,757 coin flips showing bias published in JASA by fbartos on Monday August 11, 2025 @03:00PM (#44868083)
Paper showing that "Fair Coins Tend to Land on the Same Side They Started: Evidence from 350,757 Flips" circulating here a couple of years back (https://news.ycombinator.com/item?id=42181345) is now published in the Journal of the American Statistical Association.

The main difference from the original version is that we now document a decrease in the same-side bias over time. Meaning the more people flip, the less biased they are. We guess that it's a practice effect -- they might be getting better at coin flipping over time.

(I'm the main author of the paper)

Re:Paper with 350,757 coin flips showing bias published in JASA by pestatije on Monday August 11, 2025 @05:46PM (#44869838)
what's the definition of "fair coin" then? standard coins i assume aren't fair, and neither are biased ones
Re:Paper with 350,757 coin flips showing bias published in JASA by fbartos on Tuesday August 12, 2025 @12:53AM (#44872525)
The coins are as likely to fall on heads as on tails, so they are not biased for any of the sides.

People flipping them however introduce bias due to the wobble of their tosses.