Ask HN: Does LeetCode/HackerRank make you a better developer?

I know this is question is very broad, but bear with me.

I consider myself a strong engineer. I have 14 years of experience, I'm mostly autodidact, and I can learn new stacks/technologies pretty easily.

Being an autodidact, I don't possess a degree, however I know about basic algorithms such as binary sort, quick sort, graphs etc.

Recently, I started to interview again, and some interviews require HackerRank test. Some tests are designed to evaluate a specific knowledge, say SQL or multithreaded programming, while other questions are purely algorithmic (in the style of: "find if a string is an anagram").

I don't have a problem with specific domain knowledge questions, such as SQL or multithreading. I do have a problem with purely algorithmic questions. I never used binary search, or needed to reverse a string as fast as possible, in my career.

And, sadly, I start to question my career. Do I suck at those questions because I never worked with something algorithmically hard? Do I suck at them, simply because I'm not used to solving them? Are they even good evaluation tools to understand if the candidate is suitable for the work he/she is going to perform?

So my questions is: do you think that being able to solve LeetCode/HackerRank questions (not brute forcing them, but rather understanding the underlying algorithm) help you become a better developer (rather than better at solving these types of questions), or are they, mostly, useless and it's better to build and ship software?

I share your sentiments. I, too, have never needed to reinvent the wheel and implement my own list class in Python say instead of using the one from the std lib. And countless other similar cases and things.

That being said I think places do Algorthimic tests for two reasons:

Laziness Scale

Laziness. They choose this because they can’t be bothered to come up with a puzzle from the work and make it scoped well enough to fit an interview but also challenging enough to show the interviewer the candidates skills. So instead they have you implement some algorithm you memorized.

Or they’re lazy because they’re incompetent interviewers and use these as a proxy for IQ. If that’s the case just leave. You don’t want to work there.

Then there’s scale. Google likely gets 1000 applicants or a given role. (Maybe more by a factor of 10?) How would they filter out enough candidates before you sit for an hour with an expensive engineer? They need a way to quickly filter out a ton of folks.

So that’s scale. It’s got a bit of laziness in it. But it’s primarily about managing the number of folks looking to try for a single job.

All of that said I’m just a tardigrade in a fast chasm of space trying to take on the Death Star; there’s no way the tech firms will change their hiring ways because of some logical aren’t on my side.

And who knows? Maybe you’ll get hired at Google and then go work on the next generation of kubernetes 2.0 — the point being this is a necessary evil that stands between you and many others to get paid 500k/year.

Not really. You get some exposure to a few concepts that are interesting and maybe a bit understudied in school (like dynamic programming) but challenges like the Advent of Code are better at that in my opinion.

You need to understand where they come from, namely the need for a standard and reproducible way to test candidates in 4 or 5 slots of 45-60 minutes each, in companies with huge and generic hiring pipelines where any developer can be tasked to interview any candidate. Unfortunately a lot of companies have copied the practice from Google and other big companies while they would benefit from a small scale and more personal approach. I feel like today's situation is better than it was 5 years ago though, maybe the prevalence of full or mostly remote interview rounds and/or Copilot plus ChatGPT have helped make companies and teams reconsider.

I think a good developer is composed of many things, and LeetCode/HackerRank type learning can help you become better at only a small subset of those things.

It's like trying to get your body into good shape, you need to work on different parts of the body, not just focus on your left arm (unless your left arm is especially weak and you want to focus on that).

I think leetcode helped me quite a bit as a software developer. I was in a similar boat as you, 12+ years experience doing software. I have an electrical engineering degree (famous for being bad programmers ha!) but no CS background. I could normally hack at problems, figure things out, and get things done in different settings (Linux, embedded, web, etc).

However.. studying leetcode really helped me understand fundamentals much better. I also read some algorithm books. I feel like I have a better understanding of software in general! And it has helped my career tremendously. Just my experience, I totally understand why people don't like the interview but for me I appreciate how coding problems are at least more straightforward than other arbitrary interviews.