Hacker News Reader

Comment by hyperpape | Hacker News Reader

I don't even disagree with this. It's just also an admission that you don't hire the best.

We can argue about whether there are 10x developers or not, but it doesn't matter for this discussion. If you assume that a developer is going to be with you for a year or more, then it starts to be worth a week or more of your (collective) time to improve the expected productivity of the people you hire by 10%. Not 2x, not 10x, just 10%.

If you aren't spending that time, then you're tacitly either admitting that you're not good at accurately assessing talent, or that you think any warm body will do equally well at your job. Either way, I don't think you're really disagreeing with the core of Dan's post.

View on HN

If you have 100 candidates for a job, then doing a half-hour phone screen for all 100 of them is 50 hours of work -- a bit more than one whole work-week.

And if you're committed to not missing a candidate who might be really good but your heuristics would lead to a false negative on, how much does a half-hour phone screen really tell you? I mean, can you really reject more than half of the candidates based on a half-hour phone screen if you're committed to a very low false-negative rate?

What are you going to do, bring 50 candidates into the office for half-day interviews? You've now brought your sunk cost of interviewing from one person-week to 6 person-weeks. And you've also actually probably spent 3 months doing this process, so you've probably also delayed actually hiring someone for at least a month. 10 person-weeks worth of productivity starts looking a lot less good compared to a 10% productivity bonus.

Given how unreliable interviews can be (both phone and in person), you'd have to solve it with another mechanism.

At my last company, we "screened" ~200-500 candidates per job with a work sample. It was really hard to put together, and worked amazingly well once we had it going. We did a structured interview everyone who passed the work samples and it never felt like a waste of time.

The first round of filtering is one of the jobs HR is supposed to do.

If you use a technique that cuts a lot of people, but cuts more bad people than good people, you'll still end up with plenty of good people. The difference in comparison isn't between the average and the good person who was cut, but between the good person who was cut and the good person who wasn't cut. What is the chance of the heuristic cutting someone 10% better than the best person not cut?

BTW, this is a 'solved' problem: https://en.wikipedia.org/wiki/Secretary_problem

The optimal solution only picks the best candidate 37% of the time.

That was not my reading of it. To me, given a large enough pool (~20+), you can statistically assume you'll get a top 10% candidate with this method, at least that is what i read.

Still, this method at least has some structure to it. As opposed to the anecdotes that the rest of the thread is.

I've never worked at an organization that hired the best. Hiring the best is a myth.

Reply on news.ycombinator.com