Hacker News Reader

Comment by ingvul | Hacker News Reader

> We have a really high bar for hiring at Repl.it, Dalia and the other youngsters from Palestine performed better than at least half the experienced engineers I've interviewed in the past.

I really don't want to sound negative, but I find this difficult to believe. Sorry if I sound too harsh, but experienced engineers who don't perform better than recent graduates sure it's a thing, but 50% of the ones you have interviewed don't perform better than a recent graduate? Perhaps you were exaggerating? Or perhaps your interview process is really focused on what recent graduates know best (popular algorithms) and not in what experienced engineers know best (how to deal with real world codebases). Again, I don't mean to sound harsh, I think perhaps that "...at least half..." was just a way of saying "Dalia was really good" (which sounds more credible).

View on HN

Experienced engineers who perform better than a recent graduate are usually not interviewing. They have jobs that pay them extremely well.

There's an adverse selection problem with interviewing, in that people who are good tend to disappear from the labor market and when they do appear on the labor market they get snapped up quickly. New grads don't have this adverse selection effect: there is a very good reason why they don't already have a job. This is why companies invest so much in internships: this will often be the only time to snap up a promising young developer before they start building a career at your competitor.

https://www.joelonsoftware.com/2006/09/06/finding-great-deve...

Exactly. People who are good are always going to be referrals, once they've worked at a few places, their networks are just not going to let them get away. The only times I've found where you are interviewing someone experienced who turns out good is

1) someone coming from a big corporate job where they were undervalued/underutilized

2) someone coming from a different region, different country, where they have no local network

Exactly. People who are good are always going to be referrals

Only inside a community, not coming from outside.

Also, referrals still have to go through interviews, so it's not really unusual to interview someone who is senior and skilled.

Right, and that's why we still source and recruit, to pull in people from outside of our community. It just has a much lower hit rate.

That’s not true. You can have people who are in multiple communities refer in someone from outside.

I’ve had it happen to me and done it for others.

Its the old boys club. You do realize there are tons of programmers in just the US that have no contacts with the coasts? This is what makes the Ivy League so valuable which is very good at discriminating over large groups of people.

My favorite is businesses and colleges that don't take applicants from PO Boxes.

> Experienced engineers who perform better than a recent graduate are usually not interviewing. They have jobs that pay them extremely well.

Doubtful. There is constant churn in tech, as the best way to get a raise is to switch jobs. It's well documented that junior/senior engineers (and beyond) switch jobs, on average, every 2-3 years. Longer tenures are generally favored due to vesting schedules (though I've had several friends ditch Amazon before the [iirc] 4-year cliff).

Is the constant churn just a Silicon Valley thing? I get a raise every year of about 8% which keeps me well compensated even after working at the same job for 8 years. It’s possible the company I work for is an outlier? We have lots of engineers who have been working here for 15+ years.

As a result, my interviewing skills have deteriorated significantly. I wouldn’t be surprised if a talented programmer straight out of college could out-interview me even though I have much more experience.

Looking at my friend group, it seems like a mixed bag.

We're in our early 40s. About half have been at their current place of employment for close to 10 years now. The other half seem to change jobs every couple of years, although some of those had a long tenure before the recent bought of job swapping (they haven't found a new place they like).

A steady raise rate of 8% when not being promoted would be phenomenal. After 15 years you'd be triple your original pay, which, for non-promo compensation, is unheard of.

I agree that it is pretty great. I think triple my starting pay is very doable at my company. I know that the people at the top of the engineering org chart are doing quite well for themselves.

You're an exception. In developed countries for sure most companies don't give out raises higher than a few percentage points, and definitely not constantly 8%. For reference, at your current rate in 10 years you're doubling your salary.

In developing countries your story is a bit more common but even there you can accelerate things by switching companies once every few years (somewhere between 3-5). After a while you probably want to stay put to get promotions.

It also depends where in your career are you. When I started as a junior and improved year by year, my salary grew 10% yearly in the same company, for 6 years.

It definitely depends on industry location. In the big tech centers, 10% avg increase over time would be, well average (Seattle, Austin, NY, Boston, bay area). I was shocked when I learned my last company was paying new college grads more than 100k. Then it got to be 120, 130 (total comp, salary, bonus, stock). I think the starting salary could have gone up 10% a year. This was a company that wasn't a faang but competed with the faangs for hiring.

I think you need to define "interview". I recently just got a new job and in the interview process I definitely did worse on the algorithms than I would've done right out of college, but everything else went great. "Here's how I've done it for 8 years that has been to the satisfaction of people who write paychecks" goes a long way for many companies.

Yes, that's true. It depends on how the interview is structured. I would not excel in an interview that relies on whiteboard coding, algorithm problems, and CS trivia, but I can talk for hours about the projects that I have worked on and real problems that I have solved.

I doubled my salary each time I changed jobs, from internship -> first job out of high school -> first job out of college -> Google. When I got to Google I said, "I guess that's the end of the doubling." Nope, my compensation doubled again while I was an employee there, and then doubled again. Left to do a startup and then went back - at double the compensation.

The "problem" that I have now is that there are few companies that can offer me double my salary, and none in my area. I'm not willing to relocate so my only hope is that FAANG-type companies start to hire remote developers.

I put problem in quotations because it is very hard to complain about being paid too much :(

Just need to double your job about 25 times and then you'll make as much money as Jeff Bezos

Interview skills do degrade. You have to practice before you interview if you haven't been doing it recently. Interviewing is a performance, it's different than normal programming, but the world mostly expects you to handle that difference.

1. You have to solve a problem immediately (whiteboard or now in a webbrowser)

2. It's often frowned upon to research via google search something, although in real life most people do it all the time.

3. you have only an hour to solve a problem, and what if you don't see the 'right and easy way' to do it? In real life things aren't so simple.

4. Don't forget it's a performance! You have to be on your best, sharpest most intelligent and witty behavior, think through things, it can be exhausting.

These are just a few of the ways that interviewing is artificial.

Interviewing is a performance, it's different than normal programming, but the world mostly expects you to handle that difference.

I was looking last year, I flunked my few interviews. Then I paused and spent a few weeks working through CtCI. After that I was easily getting offers. Was I a better programmer than before? No, of course not, I had just learned how to put on a show, and I forgot it all the moment I had an offer I was happy with.

What is CtCI?

that's only a 6% increase (accounting for us inflation). I am sure you can get more if you change jobs.

Yes, but even in that case, they tend to get picked up quickly. They're not staying in the pool for months and months and months, racking up dozens of interviews.

Yep. Last time I interviewed I did one interview at a random place I didn't care about just to get rid of jitters, one at a mid-tier place I'd actually accept if I didn't get another offer and then two at FANG companies. I easily passed all of them except one of the FANGs and then I was off the market. The time before that I did a single interview. The time before that though I wasn't as experience and probably interviewed at 20-30 places before I got a job.

> I did one interview at a random place I didn't care about just to get rid of jitters,

What a shitty thing to do.

Most of the time when people do this, they're open to accepting an offer if it turned out to be really great. They just know that such an offer from a particular company is really unlikely.

I don't see a big problem with it. While it may 'waste' some time on the company's part, it also sends a signal of, "hey, you could get some of these really high quality engineers if you were willing to [pay more/offer more vacation/offer remote work/etc.]".

Now, if you're unwilling to accept an offer from a particular company even if their terms blow you away, then yeah, that's a dick move.

This is such a weird comment to make: these are companies that can quite literally (and quite often) let you go for any reason at all (if you live in an at-will-employment state). I've done this exact thing before, and I've also resigned without 2 weeks notice to chase greener pastures. Business decisions should always be purely transactional and borne out of your own interest. Anything else is self-sabotage.

Companies sure as shit don't care about you, and you caring about them gives them a leg up, not you.

How so? If they turned out to be amazing then I would have considered an offer, I just didn't go into it expecting to accept anything.

This is exactly what I've done the last few interview cycles I went through. For me, my screening interview was with a company that had a tone of public problems with treatment of female engineers and fired the ceo soon after. But they had some good engineers too. I interviewed with them and it was a mess, but I did get good experience. I thought I failed, they never told me, then 6 months later tried to get me to come do a full loop.

I think it's a smart strategy to pick 1 or 2 companies you'd never want to work out, including for low pay, and do your practice there. You should tell them if you pass why you don't want to come there, including for the low pay.

Eventually you hit a ceiling, and there aren't companies that pay better than your current employer. You're getting raises each time you switch jobs because your experience & job performance qualifies you for jobs at progressively more economically successful businesses, which both can and will pay you better. This can't last forever: eventually you get to the "center" of your industry, the set of companies with more money than everyone else, and you're better off performing better within them than finding another job.

(Or alternatively, you find a fast-growing startup that's growing faster than the wealthiest companies in your industry, hop on the ground floor for stock options, and ride the stock up. Once that happens you don't need money, though.)

This has been my personal experience as well. Stay at the company and get a 6-10% raise or jump ship and get a 50% raise.

Try doing the numbers in your head.

You switch jobs every two years, applying to four companies, getting two offers and taking the one that most closely doubles your salary, then you’re gone from the market again.

Meanwhile the other 1000 guys who applied for that gig you took are still out there looking. They'll apply for 50 jobs this week, 50 next week, and 50 more every week for the next two years. They’ll be your competition again next time you’re on the market.

Now think about how many of “you” it will take doing your four interviews every 2 years before you are anything but noise in the process from a hiring perspective.

It’s the reason guys like you get offered nearly every job you apply for. Because the person in charge of hiring is amazed to have found somebody capable of programming computers at all.

A few years back when I was not so good at being a leveraged employee delivering value, I remember doing ~30 interviews in a single week because I needed to talk to that many people to find a job in a timely manner.

Sounds nice and awfully convenient for those supporting the current interviewing practices. But does this apply to all experienced engineers, 50%, 5% ? US only, Europe, Russia, India?

We need better proof and better data than a 2006 blog post.

It doesn't have to apply to all experienced engineers to make a big difference. Say that only 50% of engineers on the job market are bad, but the bad ones apply for 100 jobs and the good ones apply to 2. Then of the interviews conducted, 98% of them will be with bad engineers.

You’re simply taking away from an incredible achievement here.

She cleared interviews at both Repl.it and Google. She implemented an assignment based on operational transformations and having studied this myself it’s far from trivial. If you folks ever retire this question Id love to have a crack at it.

She also cleared the Google interview which goes deep into algorithmic aspects and system design. Which means that she’s brilliant at both abstract design and execution.

To be able to do both while graduating does place someone in the upper brackets of engineering skill. Some don’t fulfill it because of other reasons but that’s another matter.

It’s not like the bar is being lowered. They’re held to the same bar as Stanford and MIT grads who apply and they come from a third world country with only a bit of remedial coaching. It’s what top school grads already know from their campus coaching and tips from their seniors.

People who say this is akin to gaming the process you too can easily get the same by dropping a trivial amount of money on CTCI, EPI and Pramp for mock interviews.

I wrote this a little bit too much in frustration but I’m tired of these assumptions that people from the third world cannot show incredible potential sometimes exceeding their first world peers and are only held back by bad systems and politics.

There are untold depths of genius all over the planet. We haven’t even come CLOSE to most people on earth realizing even a fraction of their potential.

“ I am, somehow, less interested in the weight and convolutions of Einstein’s brain than in the near certainty that people of equal talent have lived and died in cotton fields and sweatshops.” - Stephen Jay Gould

But this is what people are objecting to, isn't it?

That anyone reasonably clever which spends months studying CTCI, EPI and Pramp and whatever else and doing mock interviews will pass the interview, while an experienced programmer will not just by virtue of their skills and knowledge.

But OP also passed a different kind of interview so it’s a moot point in this case.

> anyone reasonably clever which spends months studying CTCI, EPI and Pramp and whatever else and doing mock interviews will pass the interview

Spoken like someone who has never interviewed at Google.

I have 6 years experience, a M.S. in CS, have worked at SV unicorns, and studied for 2 months for my Google interview and still failed. Am I an idiot? Possibly, but more likely that the interviews are hard and there is a lot of randomness.

Until you actually study and try yourself, don't talk about how easy it is for anyone "reasonably clever".

People with MS and 6 years of experience are judged differently from fresh graduates.

Randomness is definitely a factor, there are so many things that can go wrong, from having a bad day, an interviewer having a bad day, just bad luck on the set of questions you are asked, the list goes on.

I have performed a lot of coding interviews (probably 400+), I'm painfully aware of how limited the signal is that I can reliably read from 45 minutes with a candidate

I deliberately ask a question that has no algorithmic or data structure component to it (and tell candidates that) it's just a simple problem solving coding question which allows some insight into general coding and engineering chops

I still see experienced engineers struggle. It is hard to pinpoint exactly why, but lack of preparation/practice definitely seems to be a problem

Covid appears (at least for me) to have killed off the whiteboard

> I still see experienced engineers struggle

I've seen experienced engineers who I know for a fact can code and solve problems decently completely freeze and blank out during easy live coding challenges.

I think interviewing is a stressful situation. It's hard for reasons outside an applicant's knowledge or intelligence. Interviewing seems to be a skill in itself. I know I hate it... :(

I give out a certain coding question to my candidates. I recently interviewed and was given the exact question - and I blanked! Like, this was a coding kata style problem that I’ve done in several languages and seen done in others. I just absolutely blanked. Brains are strange.

Steve Yegge famously talked about Google's "interview anti-loop", where you get two interviewers in your process who wouldn't have hired each other, so the things one of them values are actually negatives for the other, and viceversa.

I don't know if that's still a problem at Google, but it could explain why some people don't pass the interview process even though they are reasonably clever.

From my more than 5 years at google I saw a lot of random stupid interview questions. I was astounding. 10 years ago they didn't seem to do any interview question checking, you could ask anything.

Sorry, "I was astounding" was meant to be "It was astounding". I felt only average at google ;-)

I was referring to tech interviews in general, not a specific company. Interviewing's like dating: if you do all the right things you may still not charm the person you want, but the odds are very good that you will be able to find a good partner.

I feel like you’re projecting unrelated frustrations onto the questions being asked:

> taking away from an incredible achievement here

> held to the same bar as Stanford and MIT grad

> you too can easily get the same

> assumptions that people from the third world cannot

These have nothing to do with the discussion, which is that it’s disingenuous to call a new grad better than an experienced engineer in that the interview process is obviously biased towards new grads.

No one is complaining that she got the job. We all know how to “game the system.” Whether from Stanford or community college, anyone with Leetcode and a few weeks can easily pass these interviews. So I am not sure why you are implying that people are bitter about some perceived inability to get such a position.

People are just pointing out how disingenuous of a statement it is for the OP to say “better than an experienced engineer” when they have literally no metric to judge this. And no, system design interviews don’t really count. A few days with the System Design Primer will solve that.

> Whether from Stanford or community college, anyone with Leetcode and a few weeks can easily pass these interviews.

That is, of course, if your resume doesn't get tossed out for not being diverse enough [0][1].

[0] https://www.theverge.com/2018/3/2/17070624/google-youtube-wi...

[1] https://www.wired.com/story/new-lawsuit-exposes-googles-desp...

Rubbish.

>These have nothing to do with the discussion, which is that it’s disingenuous to call a new grad better than an experienced engineer in that the interview process is obviously biased towards new grads.

From root comment: “Dalia and the other youngsters from Palestine performed better than at least half the experienced engineers I've interviewed in the past.”. There’s nothing disingenuous about the interviewer’s belief.

From Dalia’s article: “repl.it’s interviews were really different. First they gave me an operational transformation homework assignment.”, “For my second interview, about two weeks later, I had to prepare a presentation with ideas to improve the product.”.

You are making stuff up about Leetcode and whatever - the article itself has facts that show your assumptions to be wildly incorrect.

The Repl.it OT question is easily accessible by just applying to one of their postings.

I did it a little while ago, this is the question: https://otcatchup.util.repl.co/

I believe all of the Palestinians Repl.it has interviewed have come from Manara. Manara has a pretty strong vetting system and in order for applicants to be accepted into the cohorts, they have to pass a coding assessment, video interviews, etc. Manara takes strong Palestinian talent and tries to make them exceptional. This is the reason why they might be performing better than more experienced engineers.

Our interview has almost zero traditional algorithms. And no, he’s not exaggerating. I’ve been interviewing people for a decade now and neither experience or pedigree is a good predictor of success.

Read the post if you haven’t because it touched on our process. And watch this video to learn more: https://youtu.be/kABh44IVWMo

College kids often just took an algorithms course. Experienced devs have to brush up on this stuff to jump through the hoops while also managing their day job and any other commitments.

I do a lot of hiring at a "big n" company and I'd agree with this

We don't really do an "Algorithms and Data structures" interview, our interview questions are modeled after real world problems we face(d).

I'll wager doing well in interviews isn't a good predicator of success either.

In our interview it is because it’s modeled after the real world as opposed to LARPing computer science.

It sounds like you simply don't have experience interviewing engineers, especially since you don't cite any such experience to back up your beliefs. I interviewed exactly 151 engineers over the last year for a mix of entry-level and senior roles, using the same format for all the interviews. The interview is a mix of scenario-based questions and actual coding. There is no discussion of data structures or algorithms, the coding exercise requires no special knowledge and can be done in any language, and there are no trick questions. If you can build useable software you can ace the interview.

My experience lines up exactly with what the GP said. The overwhelming majority of experienced engineers that are interviewing simply can't write useable code. I understand it's hard to believe, but it is the reality whether you believe it or not.

Can you elaborate? I believe that a fair amount of people coming out of college cannot merge 2 sorted linked lists, they can't code even fizzbuzz. But for people that write code every day, I think all of them could do those things. PMs and managers who used to code 10 years ago (like me, ahem) should be able to do that but they might be rusty.

https://www.joelonsoftware.com/2006/09/06/finding-great-deve...

Interview processes are significantly biased towards seeing large numbers of non-hire-able people over hire-able people. It's not surprising that someone with few connections and just starting their career could perform well against the biased view of an unfiltered interview pipeline.

oh, nostrademons beat me to the punch.

That’s because the interview is basically memorizing algorithms.

A real on-site interview should be:

1) give vague instructions on a fixing/modifying a moderately complex piece of software

2) have them ask good questions until they get to the heart of what they are supposed to do.

3) work with someone to accomplish this task. Use google or whatever else you need to finish it.

This interview should be 5 hours long including lunch. This is your best indicator of success.

When did 5 hour interviews become the new normal? That seems sadistic to me.

Microsoft had all day interviews back in the 90s. It’s more efficient for both parties and has been for decades.

Well in academia, and generally in research, interviews are about 1.5 days of solid interviews, plus an hour presentation. In normal times, add a good dose of jetlag.

I once interviewed at a place that was almost two solid days of 30 minute interviews, one after another. By the end of the second day I completely didn't care about anything they wanted to ask about or what they thought about my responses. I think this stamina crushing test was actually part of the evaluation. Not kidding about that.

Right, I can understand that's normal in academia, but I thought the context here is software development interviews.

Yeah was just a comment about how widespread sadism is.

Seriously.

My wife just interviewed and got a new job in the past week, for a proposal writing job at a decent-sized tech consulting firm (300 employees) where she'll be making six figures.

To get this job, she spent about two hours of preparation learning about the company and hunting down samples of past proposals (this is her standard process for preparing for an interview, by the way), then had one 30 minute interview with her would-be boss, and one 30 minute interview with three would-be coworkers (at the same time). She had a job offer a few hours after the second interview.

Now she's had several years of prior experience for some pretty large companies and worked on very large proposals in the past, but the difference seems to be in her industry they trust past experience, whereas in tech it has almost zero value, they just care about whether you can past their coding exercises.

The programmer/software engineer interview process is just so broken. I have to grind toy algorithm coding problems and rewatch algorithm lectures for weeks just to psyche myself up to go through the interview gauntlet again.

I've even neglected getting back to pings from recruiters just because I wasn't feeling up for going through the whole process at that point in my life and/or I knew I wouldn't have enough time to prepare myself for the interview to even have a chance to make it through it and I'd be wasting my time.

I get that employers are getting inundated with people that they at least feel they couldn't code (I bet they'd think that about me as well if they brought me in to interview today, even though I've been basically a one man dev team for startups before and currently developing and supporting software that services millions of customers) so they feel the need to verify the skills.

I just find verifying skills in the midst of the interview very difficult, especially if it's testing knowledge I haven't used very recently, since my brain is constantly context switching out technical details and platforms and apis based on my current work needs.

There needs to be a good way I can prove "Hey, I really can code" outside of an interview, once, that's somehow trusted. I thought that's what a degree in Computer Science was supposed to prove, but apparently that was a waste of money.

> the difference seems to be in her industry they trust past experience, whereas in tech it has almost zero value, they just care about whether you can past their coding exercises.

I'm of two thoughts.

One, the hazing ritual (reversion to mean) is because we forgot how to interview properly and can't think of any other strategies. For this, I mostly blame corporate HR wankery and failings.

Two, some Mensa style geeks do a weird bully flex, probably out of insecurity. And per our common negative attribution bias, these few "bad apples" are the ones we remember.

There needs to be a good way I can prove "Hey, I really can code" outside of an interview, once, that's somehow trusted.

Some people say this is what fizzbuzz is supposed to do. Others, particularly people outside the USA, say the solution is engineering licensure. A lot of people in hiring think that's not enough. But what is definitely lacking is institutional trust between companies outside the FAANG bubble, and perhaps blue chip companies like IBM.

The problem of low institutional trust means that working for Company A as a programmer for years means nothing to Company B, and you have to prove yourself all over again if you have to interview for Company B even if your GitHub is loaded with open source side projects. So the problem is not proprietary code or not being able to show your work to a new employer, it's trust.

Anyway, the developer interview culture is severely broken, and it's been discussed a lot elsewhere on HN, yet nobody has been able to solve the problem. We have smart thermostats, can order books with our voice from the couch, and 3D print a house, but we can't solve hiring and just have to accept the status quo if we work in tech.

They didn't say all recent graduates performed better than 50%, just the ones that they interviewed from Palestine. It sounds plausible that they only interviewed a handful and they were better than that 50%.

Then the numbers are too low to draw conclusions.

Yea, the numbers would be if that was the case. I don't think the person that said that statistic was saying it like it passed the 5 sigma mark. It was given as an offhand anecdote and I didn't see a reason for the overly negative post that I responded to originally.

where did he draw any conclusion?? It seem like you're the one who jump to conclusion here.

It was just merely stating a fact that Dalia and her acquaintance perform better in the hiring process than at least half the experienced engineers he has ever interview.

If anyone is jumping to conclusion its you guys

"It seem" you're the one jumping for a fight. My comment still stands regardless of anyone coming to any conclusion.

I don't know who "you guys" is.

take it as whatever it is you want.

The interview process is being gamed. Just read the article. I am not trying to single out her. It is an industry wide practice. You mostly get people who are good at passing the interview versus who know how to program. But that is ok, because the promotion process is being gamed too. You have people who know how to get promoted vs who know how to provide value. Don’t worry, our educational system is being gamed as well.

I agree with this. Here's some truths:

* Experienced engineers will perform better than graduates in general

* The distribution of experienced engineers who are poor performers won't over represent application to one particular company unless they are an outlier (Repl.it is not Google)

This makes me wonder what Repl.it's hiring process is and why it is not doing well at attracting good people.

> Experienced engineers will perform better than graduates in general

In my experience there are a lot of experienced engineers that are very low quality, and it isn’t uncommon to have experienced engineers that have negative productivity (working with them costs more than they produce). See https://thedailywtf.com/ . Starting with recent grads (filtered for potential) can easily be more productive over time, because you can teach them good habits.

> The distribution of experienced engineers who are poor performers won't over represent application

Hiring being inundated with poor quality candidates is a common problem due to adverse selection: https://news.ycombinator.com/item?id=26254114

That Joel article is enlightening. Thanks.

Not sure who owns pedantic.com or why it doesn't redirect to HN. It certainly should.

Reply on news.ycombinator.com