Post #3501697

2026-07-01 00:29 UTC

@RyeNCode@mstdn.ca This is something I’ve encountered. I know that most of the time this won’t happen. But it will be bad when (not if) it does and we are using a platform that has features that allow us to prevent it. Humans are not good at understanding that improbable things are certain across enough instances.

Replies (1)

@RyeNCode@mstdn.ca 2026-07-01 00:45

@abstractcode@eigenmagic.net It's especially egregious when the above exchange occurs during post incident review caused by that exact "rare" event. Depending on the audience, of they are likely to be receptive to new ideas, I like to start with exploring the Birthday Problem / Paradox https://en.wikipedia.org/wiki/Birthday_problem It's scale is small enough for people to have a reasonable (but wrong) intuition about, familiar framing (oh! I share a birthday with Alice and Bob!) and easy to show that the normal intuition is ass-backwards. It's a useful tool to show that no matter how unlikely, the more often it can happen, the more often it will happen. Things like different types of distributions are hard to understand until shown a version that works for you. But ever so useful to understand the world. Regression Towards The Mean for example ¯⁠\⁠⁠(⁠ツ⁠)⁠⁠/⁠¯

Open ##3501694