Hacker News new | past | comments | ask | show | jobs | submit login

You can't just say "survivor bias" and think you have a solid argument. Let's examine your objection more critically.

I mentioned I'd carried every day since 2006. Suppose this isn't quite true, and I've only carried 330 days per year (this is too low, but still). Further, we just started 2017, and I started carrying in August 2006. This means I've carried roughly 3,437 days. If my carrying a Glock on any given day is as high as 99.9% safe (i.e., I have a 0.01% chance of a negligent discharge), then I should have a "survival" probability of only 3.2%. If my daily safety probability drops to only 99.7%, my probability of getting to now with no ND is 0.0%. Decreasing my daily safety probability to only 99% means my probability of no ND is 9.95e-14.

Now think about what this means. Either:

1. Carrying a Glock is very unsafe, and I'm incredibly lucky. In this case your claim of survivor bias would be warranted (assuming there weren't many others like me, which there are).

2. Carrying a Glock is very safe, and the chances of an ND are substantially less than 0.01% per day, in which case the charge of survivor bias is unwarranted.

It strikes me that #2 is the clear winner here.




I think the main thing you've demonstrated with your calculations is that a 0.01% chance of negligent discharge per day is in fact rather unsafe. In particular, a 0.01% chance of discharging per day means that with daily carry for two years, you are more likely to have experienced a negligent discharge than not.


Sadly, this comment of mine is incorrect. I meant 0.1%. My error comes from the same error in the comment I was responding to:

> 99.9% safe (i.e., I have a 0.01% chance of a negligent discharge)


Agreed! Which is why I stated that I think scenario #2 is the correct one, and we don't in fact have a legitimate case of survivor bias. Remember, this discussion began when I said I do it safely. The charge was made that this is survivor bias. By putting numbers to it, I think it becomes clear that the determining factor is not my bias but real, quantifiable safety.




Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: