Hacker News Reader

Comment by rhexs | Hacker News Reader

I think there's certainly some truth to coding interviews being an effective mechanism in shifting hiring towards H1Bs. It's much easier to say that Americans can't do the job when you create a test around studying for 500 hours to memorize hundreds upon hundreds of coding puzzles in order to build up a mental encyclopedia of how to quickly apply your algorithmic "training" to the puzzle of the day.

I'd bet H1Bs are going to be much more motivated to jump through that hoop. Benefits them and the company doing the hiring that conveniently just can't find enough qualified American engineers to do the job.

View on HN

This is an absurd characterization of coding interviews. I've been on both sides of big tech interviews, and I've been at a few companies. I've seen two extremes. At one end, the company had pretty lax interviews, and I rarely interviewed anyone while I was there. The quality of people at this company was relatively low, from my small sample size of coworkers. At the other end, a company well known for challenging coding interviews, and I regularly interview people. The quality of people at this company was relatively high, again from my small sample size of coworkers.

I think coding interviews are actually working fairly well. They divide people into at least two groups: a first group is people that can pass them without too much trouble, and they mostly seem to get hired, but of course there are mistakes. A second group is people that need to spend 500 hours cramming to appear qualified, and basically are trying to game the system. These people mostly don't get hired, and I suspect also are completely driving the conversation on HN and elsewhere about how coding interviews are a terrible failure.

FAANG nowadays wants you to be able to solve 2 leetcode mediums (or an easy+hard) without errors and with optimal computational complexity in 35 minutes. This is done up to three times so that's six problems and if you miss even one you likely fail the whole thing. To me that's really really hard unless you study to the point of having the solutions memorized. It's not the problems that are difficult per say but the time constraint which means going down the wrong path leaves you with no time to fix things.

I've worked with people that I consider pretty smart, and they definitely study (and enjoy) the kinds of problems that are given in interviews. 500 hours may be an exaggeration, but I think the idea that this stuff comes naturally to successful people is a bit ridiculous.

I would say I've never met anyone who can "invent Quicksort" without too much trouble. However I've met many people who've memorized it for interviews.

I'm not necessarily anti coding interviews, but somehow convincing yourself there is any degree of innate talent is a myth. There is a linear relationship between practice and performance. It's as simple as that. Some people are privileged with better resources to practice more effectively, but at the end of the day, it comes down to who has done it more.

Reply on news.ycombinator.com