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

> I still don't get why such questions are even asked

The thesis is not that these exercises are representative of work but rather predictive of performance.

Sales has a similar phenomenon with sports. While there is no athleticism involved in selling, many believe history in competitive sports to be a positive predictor of sales success.

---

You can reasonably argue whether leetcode accomplishes this well or poorly, but...

Always remember that purpose of an interview (for an employer) is to predict performance. So you are looking for the resume screening+interview process that most accurately assesses that.




> Sales has a similar phenomenon with sports. While there is no athleticism involved in selling, many believe history in competitive sports to be a positive predictor of sales success.

This is interesting. I had never heard this before. Is there any research on this? For that matter, is there any actual research showing that leetcode-style interviews actually do predict performance? If so, do they do so any better or differently than an IQ test?


I recall reading that leetcode style questions are basically just a standin for an IQ test, except with plausible deniability for being directly applicable to the job.


I saw this claim. I never saw data.


So you're saying that someone wasting their time studying leetcode to pass a stupid game is a good indicator?

I would almost believe the opposite: if you actually pass those tests with flying colours, it shows me that you believe you needed to do that to be hired, while someone who's actually experienced would never in a million years step down so low.


I believe skill at “leetcode problems” is predictive of general programming skill. Someone who can solve leetcode problems can almost certainly learn css. But, clearly from reading comments here, not the other way around.

Personally I love leetcode style problems. They’re fun. And useful - I use this stuff in my job constantly.


I would be scared you are overengineering and optimising things though. I have seen people implementing complex paradigms and weird optimisations instead of writing simple code just to make sure they are perfectly optimising.

E.g. optimising client side code where N is likely never to be above 300, but instead of few simple lines, write a complicated solution to make sure it can scale to billions and beyond.

I would take any problem solving energy and spend it on side projects instead of doing leetcode. I do like those exercises, but I enjoy building new things more and which gives me practical experience which I think is more important.


Skill gives you the gift of choice. You know how to write it either way around and it’s up to you to decide. Being able to correctly decide when to hack something inefficient together and when to optimise is another skill issue. It’d make a good interview question, that.


Yeah, but leetcode does not necessarily give you that skill or even prove it. And a talented problem solver would be able to find optimal and practical solutions when they are required and are not premature even without doing leetcode.

You might get false positives as well. E.g. you get people who are tunnel visioned on leet code, cracking the coding interview and other common system design books, they know all the answers, but then they completely lack common sense day to day and it can be hard to test for that if you are solely focusing on leetcode.


Of course. I’ve interviewed over 400 people in my career and I’ve never directly asked someone if they’ve done leetcode problems. I don’t care about actual leetcode. Looking at someone’s progress on leetcode as a replacement for an interview would be a terrible idea.

As an interviewer, I care about their skills - technical skills (like debugging ability and CS knowledge), social skills (can we talk about our work together?) and judgement. Your ability to understand data structures and algorithms is signal, but there is a lot more to a good candidate than knowing how to make a bit of code run fast. Knowing when to make code fast is, as you say, just as important.


It's a good indicator for someone will to jump through arbitrary hoops to get promoted at a big corp: write all the paper work as design docs, get all the promos requirements and make sure all the weird business requirements of corpo customers are met. If you are not willing to "step down so low" you a perfect example of the someone who they want to filter out.


I didn't state my personal opinion above, but yes I have seen leetcode aptitude to positively correlate with day-to-day problem solving.


If your company uses leetcode to filter out employees then you are a leetcode team with your internal levels and ranks, not at all representative of the whole population of skilled IT people.


How would you, or anyone, know if your career is representative of the field?

I’m sure plenty of people spend their career never learning or using data structures and algorithms knowledge. But I suspect plenty of people spend their career using this stuff all the time. Eg people who work in databases, compilers, operating systems, video games, ai, crypto currencies, and so on.


I have seen Goodhart's law.


> The thesis is not that these exercises are representative of work but rather predictive of performance.

Well said. Just like in college, calculus (not analysis) and organic chemistry are used as filter courses. Of course, why the two courses, especially organic chemistry, are so hard to American students is another topic. I personally think that it shows the failure of the education system of the US.




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

Search: