Hacker News new | ask | show | jobs
by edhowzerblack 2675 days ago
As a developer who has recently been through the hiring process, I think the problem is not a shortage of talent, but rather that companies are awful at hiring. Here are a few of the main issues:

1. Don't limit your potential talent pool to the miserable and the unemployed. In theory, any developer who is currently employed might be interested in leaving their job to work for you. Perhaps you will offer them more money, or perhaps a better quality of (work) life. However, most developers won't bother to talk to you if they already have a job. Why is that? Simple, if they do talk to you, you're going to offer them a homework assignment. You're going to tell them that it shouldn't take more than an hour, but it will actually take two full days to do right. Giving someone a homework assignment isn't a way to woo them away from their current job. So you are left with a candidate pool that consists mostly of people who are desperate enough that they have to agree to jump through your hoops (i.e. unemployed, miserable in their current job, etc.)

2. Don't discard capable people. The average interview begins with a remote code challenge whereby the candidate, who is probably nervous, has to pair program a completely contrived problem with a complete stranger watching them over a webcam. There are tons of developers who are good at coding solutions to real world problems in real world situations but simply don't perform well in this type of situation. This type of scenario is not really assessing a candidate's skills. It's assessing their familiarity with specific contrived interview problems and their ability to perform under duress.

Of course, the points you raised are fair as well. Recruiters are not technical people, they are sales people. You will have to find a good recruiter. They are out there, you just have to find the right ones. But my point is, once you find the good recruiter(s) don't waste the human capital they can being to you by having a terrible interview process. Here is my advice. Once your recruiter has a handful of resumes you like, carve out half a days worth of 30-minute sessions to meet with them at the recruiter's office. Ask them about projects they've worked on. You'll find out much more about what they know and what they're like to work with than you would with the typical remote code challenge. One you narrow the pool down a bit bring them in for on-sites.

Best of luck!

7 comments

I'd estimate (without bothering to sit down and crunch the numbers) I've had a 30% success rate interviewing for programming jobs. I usually get about 2 rejections before I get an acceptance (and, of course, when I get accepted, I stop interviewing). Based on that, I'd be led to conclude that I was a _dismal_ programmer, probably not fit for the industry at all. If there was as tight a tech shortage as they keep insisting there is, I'd expect it would be pretty near impossible to be rejected for a programming job. Yet I find it happens to me with alarming regularity and based on what I read here, I'm not alone. It's almost as if the "talent shortage" is imaginary...
The talent shortage is absolutely imaginary. The hiring process is run by incompetent internal recruiters and competitive developers with Aspergers whose goal is to reject as many people as possible in order to prove that no one is as smart as they are.
> It's almost as if the "talent shortage" is imaginary...

Talent shortage means that employers can't find workers with the skills the claim they need at the price they want to pay.

> (and, of course, when I get accepted, I stop interviewing)

Then you only ever have a sequence of rejections ending with a job offer. Of course you will feel terrible if this is your strategy.

Do a few more interviews and you'll find that sometimes you get two or three offers in a row.

it is
I got it!

We will hire anyone who sounds good, and fire them within a month if it turns out they're not up to our standards. Then, we will ensure we never get anyone who isn't already employed and anyone just looking for a paycheck or three.

If you heard that even 20% of people were fired within the first month, how likely would you be to take that job?

Note: It's usually called contract to hire. And it just takes the interview stress and pushes it out to 1-6 months.

I don't have sufficient sample data from my personal experience, however, I conduct job interviews the same way I got interviewed and hired for my last job and my last three hires were and are really good teammates.

I don't ask for coding problems and don't have a strict interview structure. Instead, I try to make a conversation about the candidate and tailor interview around their experience. I also try find similar problems they face that we have and talk about solutions that we apply which helps me see how flexible they are or do they outright reject different viewpoints.

Ultimately, the main goal is finding the people that I would like working with because it trumps pure technical skills in importance. It might not work well where some highly specific and rare skills are required, however it works in average enterprise environment.

The interviewer should be a pretty good conversationalist, though.

So what? If you are used to freelancing, it's the same thing.

I would prefer that 80% of people are fired in the first month.

Sure, but, most employees would not prefer that...its a Market for labor. You are buying, Labor is selling. If what you are buying is an employment contract with a penalty-free terminate-for-convenience clause, expect that you are buying that piece of insurance from the employee at a high cost.
To be fair, we already have a penalty-free terminate for convenience in the US in most states, it is really social convention that prevents it.
I'd like to talk more about (1). We do a homework assignment that in fact does takes a few hours, but only for candidates who don't have experience developing in our core stack. If you can code in the languages and frameworks we use, it's a step we bypass. It tells us if they can solve a problem well using the tools and setup of their choice. If we clearly see they did well at this, the on-site can then focus on the more non-coding aspect.

Our on-site interviews are highly contextual and rooted in the real-world where candidates work directly in our codebase. They are reasonable in the things we ask candidates to do (re-factor something, review PRs, implement a small enhancement).

If someone can't code in one of our core languages, it's a tougher assessment since we would have to resort to hypothetical/whiteboard crap (which we hate). How do you assess someone then? We recently had a candidate do a take-home where they didn't do well, and it did save 5-6h of everyone's time and their effort of coming in and being subject to the pressure of the interview.

I'd love to hear counter-opinions on this.

That's nice. If your recruiter called me I would pass unless I happened to be unemployed. And even if I were unemployed, I'd tell your recruiter that I'm interested but I'd prioritize all of the other companies I'm talking to that don't have a homework assignment. If I failed their code challenges, only then would I do your homework assignment. Most developers I know would do the same exact thing. You have to realize that you are NOT the only company the candidate is talking to. In fact, the candidate doesn't even know if they want to work for you. Their recruiter pitched a bunch of companies to them. Your company was one of those. They were like "okay, sure send me over there." Seriously, unless you're a famous company like Google, the candidate never even heard of you. You are one of many places they agreed to let a recruiter send your resume to. Your job is to sell them on why they want to work there, not give them a fucking homework assignment.
I like your response. I think the reason for it is because it's a false indicator of capability and its demeaning. I think some technical discussion is good but this paranoia that somehow you're scamming them with a long con is so stupid and results in terribly non diverse hires.
So in the above reply I completely ignored the fact that he said the homework is only given to people who haven't worked with their tech stack. I also swore and went on a rant. Sorry! It's still something that needs to be said though. Just not to that guy...
Haha thank you for not directing that at me.

Hiring and having a welcoming but selective process is rough :)

Indeed it is my friend. One more thing I'd like to share... A lot of companies start their process by having someone in HR reach out to the candidate. Often these calls provide little of no value to the candidate. They are simply another hurdle. Here are some things I wish the HR folks would take into account:

1. Please be prepared to actually sell your position. Most recruiters simply explain what the business is and then tell you that their engineering team is really tech focused. It's totally generic. It's nothing the candidate hasn't already heard from their recruiter or read on the web site.

What's special about your product or company? What kind of benefits do you offer? What makes your tech team so great? Have you built some amazing open source library? Is the CTO an ex-Google guy?

2. Please be prepared to answer some basic questions that an engineer will have. For example, what is your tech stack. "We use JavaScript on the front end" is not an acceptable answer (unfortunately, It's one I've received several times). Tell me what frameworks you use. I know you're not an engineer, but you could ask one to write it down for you.

3. Please know what the technical assessment process is. "We have a technical assessment" is not an answer. Is it a take home assignment? A hacker rank test? A remote code pairing? If it's a take home assignment, how long does it take? Again, you are likely not the only company the candidate is interviewing with. The candidate needs to know how long the process will take and how much time he has to invest.

> only for candidates who don't have experience developing in our core stack

If you're looking for top talent it wouldn't be measured against any singular tech stack. I tend to take positions on stacks where I'm unfamiliar, and no I wouldn't complete homework. Employers are competing for the good candidates and should have a process of identifying them without hoops.

I'd love to hear your thoughts on the right process. We want to make it easy for the candidate too. Care to share your feedback/opinions?
When I'm interviewing senior candidates or being interviewed, my preferred first interview is a conversation about interests, past work, and areas of future work. Depending on the ability to communicate effectively and deeply about topics of shared knowledge and interests that will decide the next steps. It also gives a good idea of the candidate and the company in the process. The best use of time all around. I have yet to be 'duped' by an imposter by this process. Some may be slightly underwhelming but I've yet to regret a hire or being hired.
I do that as well and that still doesn't tell you much. It's a good first screen filter but isn't always sufficient. For senior candidates though, I agree that a homework assignment is not the best qualification process.
People willing to spend a few hours of their day doing a "homework assignment" will tilt towards the spectrum of people who have a lot of free time which might not intersect that well with the set of people you want to hire.
> If you can code in the languages and frameworks we use, it's a step we bypass.

Curious; can you say more about how you judge this? Isn't that sort of the point of the homework in the first place?

Judge if they can code in our core stack? Well, if they're working with it in their current job, or in side projects. Assessing if they are good, we can do it in the on-site. Like I said, our on-site is based on knowing our stack, so we have to make sure they can work with it before we bring them in.
What is this 'core stack' you speak of, is it special in some way?

I once did an all day on-site pairing interview with Pivotal. The morning was Swift/iOS and afternoon was Java back-end. I didn't have experience in either except some Java desktop GUI from a while back. It was all fine, they wanted to see how I think, what code paths I think of, what tests I choose for coverage. The actual syntax of what's being written wasn't the main point. That translates quickly on the job from experience doing similar tasks in other environments. If on the other hand, you've never written a test, or discussed code with a colleague those are not as easy to pick up from somewhere else.

This seems to align with my experience, particularly your first point.

For context, I’m working as an Associate Product Manager (so not quite a Product Engineer) but I had applied for a job at a Start-up company that I found extremely interesting. Had an initial interview but didn’t hear back from the company for about a month. I had also sent follow up e-mails about this.

When I eventually did hear back from them, they apologised for their delay but then spun me with a homework assignment that would have taken at least 2 days. Safe to say, I kindly declined.

So yeah, really focus on how you’re recruiting for talent.

In the US most of the recruiters I've been getting are more at least technically aware but still struggle because they are external. The only true people I have felt like I have had a useful conversation with are companies who have an in-house talent lead as they can generally describe what is going on and can fit you across teams. In my current role I was cross matched successfully and probably would not have pulled the trigger without the company based talent person involved in the process.

The outside recruiting firms I have tried to work with have just been misery by comparison. They either don't know enough about what's going on or clearly don't even understand the position they are offering and it takes like 3 or 4 back and forths to get the real job description and it doesn't even make sense. From the other side once my current role started resourcing from an outside firm the matches were worse and worse and I wonder if it just was because the rep on the other side couldn't talk about our actual company in any meaningful way.

> There are tons of developers who are good at coding solutions to real world problems in real world situations but simply don't perform well in this type of situation.

What do you propose instead to see whether one can do real world problems in real world situations?

I think he's saying that normally a person isn't coding while someone is watching you. I have no problem implementing a data structure on my own but if you ask me to do it in 15 minutes while you're watching every typo I'll probably freeze up.
Has anyone ever flipped the script in an interview, asking the interviewer if they can watch them write code at their workstation for an hour so they can get a sense of what it is like to work at the company?
You can learn a lot about what a developer knows and is capable of by talking to them about the projects they worked on. Ask them detailed questions about the particular problems they dealt with on projects and how they were solved. Ask them to explain core concepts that are intrinsic to the technologies they've worked with. You can also get a feel for their personality this way.

Remember, we're talking about a screening process. Once they come onsite for the real deal you can throw all sorts of stuff at them.

If (1) coding homework sucks, and (2) coding live sucks, how do you propose assessing a candidate's coding ability? Take their word for it?

Absent a large open-source repository or other public corpus of work, you generally have to rely on one or the other.