Hacker News Reader

I think it's probably much truer that big tech companies are optimizing for false negatives because they do have a really deep, talented applicant pool. Some start ups do but it's definitely more of a risk. I completely agree with you that in general you're better off optimizing the F-measure, but I'm not sure it's empirically true that lots of companies thought of as good do this.

View on HN

Thinking about this a little more, your final point may be right. It may be that companies become good by optimizing for F-measure, develop a talented and deep pool, then transition to optimizing for precision to manage the applicant load. Thus, most companies considered good have already transitioned while the ones who haven't aren't widely known.

Reply on news.ycombinator.com