Hacker News Reader

Comment by marcinzm | Hacker News Reader

You're missing the quality they're actually testing for: ability to study and dedicate yourself to a task. Or in other words work ethic. The algorithms don't matter but the fact that you can spend months learning them is an important skill. Intelligence is less important for success than work ethic which is why work ethic is tested for so much. Specifically, work ethic and ability to learn in a related domain as, for example, literature study skills may not transfer.

>If there was a perfect correlation between pre and post entrance test performance there wouldn't be PIPs, performance based layoffs etc.

Nothing is perfect, saying it's useless because it's not perfect is a silly argument.

View on HN

In my experience, certainly no interviewers think that they're screening for "ability to study and dedicate yourself to a task." In being on both sides of the table for more than thirty years, I have never heard an interviewer mention that.

Companies encourage you to study for the interview, providing you with links to books, guides and preparation material. They'll delay the interview to give you time to study more. So it seems pretty clear to me that corporate management sees it as a test of study skills even if interviewers don't.

edit: I'm talking about large tech companies, not the startups blindly copying their policies with little thought or consideration.

Never seen that happening. Have interviewed with a few FANG companies. There are hundreds of things they can ask and is not reasonable to study for them all

I've had FAANG tell me to read Cracking the Coding Interview, sent me links to leetcode/hackerank, sent me PDFs with notes on what topics would be covered for domain specific interviews, linked me to their blog posts on architecture design, videos by their engineers on how to prepare, given me the literal behavioral questions I got asked, etc, etc. Facebook sent me well over a dozen links to guides for my interview and half of those had further links in them (including to a list of practice questions on hackerank).

edit: It is also very reasonable to study having done it myself and talked to people who have gotten into these companies. At 15 minutes a pop you can do 100+ leetcode questions a week. Spend a few months and you'll have studied almost every reasonable question. Most of them are repetitive patterns but around 5% have tricks you need to memorize. That covers 2-3 out of your 5 interviews. Behavioral you need to memorize around 5 questions that they'll basically give you ahead of time. System design is harder but there's videos for the 10 topics that will cover 90% of the potential questions.

I really think having people pass these coding interviews is so that the interviewer isn’t responsible in selecting the wrong person.

What kind of team are you with? And how many leetcode-style problems have you solved in your actual work over the past week?

I went through the same thing for my Facebook interview. How do you do this if you have a day job or even worse, a day job and a family?

I feel these companies prefer to promote internally so losing senior talent doesn't bother them as much. Fairness is not a quality I associate with them.

That is definitely not true any more, as they got bigger most of them desperately tried to hire experienced external hires, on the principle that nobody gets for fired for hiring an ex-Googler.

Can you post the source of system design videos ?

> At 15 minutes a pop you can do 100+ leetcode questions a week.

By that logic, most people would have completed all 1500 LC's in 4 months.

Numbers say otherwise. Each medium difficulty problem takes an hour to solve, assuming you can come up with a non-brute-force solution. Most of them force you to lookup, understand and implement other people's ideas.

If you can do 3 problems a day, completely on your own without any lookup, you're doing good pace.

Some companies even go further. I invited for McKinsey as an experienced hire. They not only sent me study materials, but also setup phone calls with McKinsey alums to practice for the interview and gave me the flexibility to speed up or slow down the process based on how prepared I felt for their interview process.

Everyone involved in the process wants you to succeed. If not out of the goodness of our hearts, then because we've invested time. Under-preparedness is a stupid and preventable reason to lose a candidate.

When someone comes in and bombs, that is at least valuable information: we don't want to work with this person. When someone comes in and seems smart and capable, but makes hardly any progress on the problem because it's obviously their first time on a whiteboard, it's just regrettable all around.

We're not selecting for preparedness. In fact, we try to select against it: interviewers take pride in designing twists that separate understanding from recall. If you seem to be recalling, we're supposed to pick a different question, or failing that, report a null signal. But we do need you to be prepared enough that your interview skills don't get in the way of your actual skills.

> The algorithms don't matter but the fact that you can spend months learning them is an important skill.

And you're missing what these tests are actually doing instead: Figuring out which people have already seen the problem at hand, and probably memorized all/some of it.

The problem is that these interviewers will not pass you if you don't solve the problem almost perfectly in an amount of time that is way too small for someone who has never seen it before.

To make matters worse, many of the solutions require some trick or mathematical trivia to get right. Does that really prove you know your Algos & DS?

>And you're missing what these tests are actually doing instead: Figuring out which people have already seen the problem at hand, and probably memorized all/some of it.

That is literally my point. They are very much testing your ability to study for months on end. Study as in memorize. That is by design at this point and not a side-effect. If you aren't willing to show the ability to study successfully for months on end then they're not interested.

edit: Over the course of months of grinding leetcode you will see 99% of the problems they are going to ask. Then it's about memorizing the tricks and patterns.

edit2: And as I said below their recruiters are very open about what you need to do (grind leetcode) to succeed at the interview.

I feel like that selects against innovative people. Memorizing problems that have already been solved is basically the definition of re-inventing the wheel.

I'd argue that memorizing solutions is the exact opposite of reinventing the wheel. You implement an existing tried and true approach rather than trying to come up with a clever solution yourself. Most of the work at large tech companies isn't innovative but rather it's tedious work that has to be done to high standards with verbose documentation. This seems to select the right people for that. Remember that selecting against doesn't mean excluding. Enough innovative people will get in to keep pushing things forward.

You couldn't be more wrong. Memorization is one of the lowest forms of intelligence. Just because you can spit out a DFS algorithm, it doesn't mean you understand why and how it works.

FAANG employers want innovators who think outside the box. You can't do that with sheer memorization. Nothing new would be developed if tech companies hired people who just worked with what they already knew.

There's a difference between software engineers and coders. If tech companies didn't want engineers, they would simply hire coders from a coding boot camp.

I've always thought the SAT is basically gauging the same skill: can you study for a test and do well? The grades you get in college are mostly determined by how well you can study for tests. In the process, hopefully you learn the subject, but that always felt like a side goal.

I think tech interviewing is somewhat similar: are you focused and dedicated enough to learn something arbitrary and produce code for it?

The LSAT too. It's just a giant test-game. By the end of studying for it, I was reminded of how I used to approach StarCraft. Once you figure it out and spend enough time on practice, scoring in the top percentile isn't very remarkable. Nothing you learn will help you in law school but you will have proven your ability to study. (Rare exception: I had a friend who didn't study and just took the test and scored in the 97th percentile. This is very uncommon and I am just taking him at his word that he didn't study.)

> You're missing the quality they're actually testing for: ability to study and dedicate yourself to a task. Or in other words work ethic

TIL I've been "hacking" all their "tests" :D acing them just by curiosity & interest in computers in general, a bit of intelligence and a lot of laziness!

FAANG or other companies? FAANG interviews tend to be timed enough that even if you're intelligent enough to come up with a solution you won't have enough time if you haven't studied. On average you'll get roughly 17 minutes to solve, explain and code a leetcode medium with optimal efficiency and no bugs. Either that or 35 minutes to do a leetcode easy followed by a hard. Again with optimal efficiency and no bugs.

Other companies. Yeah FAANG is probably trickier. Part of the reason I haven’t ever applied is that I cannot be bothered to study. In addition, their salaries aren’t that overwhelming in Europe so why bother...

The stock grants are very, very good, especially in an insane market like we have right now.

Ethics? You are asking people to study and spend hours working on LeetCode problems for something close to completely useless. On the ethical scale, people that accept to spend hours of their limited life doing so are probably not the ones I would think of as "looking to work on something great"

Most of the jobs at these companies are boring grunt work and definitely not working on something great. So they're selecting the right people for that. You do get paid significantly more than other companies so there's a very real incentive for playing the game.

Yeah writing an OS, a machine learning framework, designing deduplication for a photos app with a billion users , working on on board sensors for a self driving car. Such "grunt work". I'd love to know what exciting and hot thing you're working on ? Is CRUD the new sexy ?

You're looking at the best of FAANGs projects compared to the median of overall developer jobs, which isn't a fair comparison.

I like how the original example is "median" and not the "lowest of lows". While my examples are outliers. HN bias again.

Reply on news.ycombinator.com