|
blah_blah posted:Sure, if you don't care whether your SWEs can do basic math or not. The time (and space) complexity of bubble sort is incredibly basic and most reasonably strong candidates should be able to figure it out in a few minutes on a whiteboard. The second thing you said may be true, but the correlation of being a productive software engineer and the ability to whiteboard fun math puzzles like this - especially in an interview setting - is very close to zero.
|
#
?
Mar 29, 2016 22:45
|
|
|
|
| # ? May 17, 2024 07:01 |
|
|
blah_blah posted:Sure, if you don't care whether your SWEs can do basic math or not. The time (and space) complexity of bubble sort is incredibly basic and most reasonably strong candidates should be able to figure it out in a few minutes on a whiteboard. This is backed up by the research that says that technical interviews are no better than chance at predicting a successful hire.
|
|
#
?
Mar 29, 2016 22:47
|
|
|
Larry Schwimmer was (perhaps still is, I dunno) famous for asking the best questions during TGIF, hence the outbreak of laughter.
|
|
#
?
Mar 29, 2016 22:49
|
|
|
bacon! posted:The second thing you said may be true, but the correlation of being a productive software engineer and the ability to whiteboard fun math puzzles like this - especially in an interview setting - is very close to zero. Arsenic Lupin posted:Somebody fresh out of college or grad school, yes. But calculus is also basic math, and I think we can both agree that expecting somebody 15 years out of college to solve a basic differential equation, when s/he hasn't used them since college, is beside the point. You can do a lot of good computer work while having forgotten, say, computability theory, or compiler design, or any other class that is specific to a particular kind of problem solving. In my experience -- and yours may vary -- that sort of whiteboard challenge can often be more about cock-measuring than about technical competence. There's a definite cargo-cult vibe of "I did this to get my job, and therefore you must, too." We're not talking about quicksort or mergesort or something with actual difficulty here. A candidate should be able to realize that the worst situation here is that the list is reverse sorted, and they should be able to realize that this results in (n-1) + (n-2) + ... + 1 swaps. And they should understand that that has complexity on the order of n^2 (you don't even need to remember the formula for the sum to realize that). This isn't a fun math puzzle but literally asking a candidate whether they can walk through a very simple snippet of code and describe what it actually does. Arsenic Lupin posted:This is backed up by the research that says that technical interviews are no better than chance at predicting a successful hire. Citation please, because as phrased above this is a massive overstatement.
|
|
#
?
Mar 29, 2016 22:56
|
|
|
blah_blah posted:Citation please, because as phrased above this is a massive overstatement. quote:Years ago, we did a study to determine whether anyone at Google is particularly good at hiring. We looked at tens of thousands of interviews, and everyone who had done the interviews and what they scored the candidate, and how that person ultimately performed in their job. We found zero relationship. It’s a complete random mess, except for one guy who was highly predictive because he only interviewed people for a very specialized area, where he happened to be the world’s leading expert.
|
|
#
?
Mar 29, 2016 23:10
|
|
|
blah_blah posted:Citation please, because as phrased above this is a massive overstatement. This is a pretty good one: http://www.nytimes.com/2013/06/20/b...nZG+IlY9sYUxzBg quote:We looked at tens of thousands of interviews, and everyone who had done the interviews and what they scored the candidate, and how that person ultimately performed in their job. We found zero relationship. It’s a complete random mess edit: beaten by a microsecond. I will add my own experience as a software hiring manager: I used to use whiteboard brain teaser type puzzles in hiring selection, and in two separate cases hired "smart" people that dominated this type of challenge that later performed horribly in a team engineering setting. bacon! fucked around with this message at 23:13 on Mar 29, 2016 |
|
#
?
Mar 29, 2016 23:10
|
|
|
There's probably a bit of a ceiling effect among the people Google green-lit, though. Still in general I would advise people to have their own projects on github etc. over prepping for technical interviews per-se, unless you're rusty on something you know they care about and can really brush up with just a little bit of work.
|
|
#
?
Mar 29, 2016 23:24
|
|
|
bacon! posted:edit: beaten by a microsecond. I will add my own experience as a software hiring manager: I used to use whiteboard brain teaser type puzzles in hiring selection, and in two separate cases hired "smart" people that dominated this type of challenge that later performed horribly in a team engineering setting. That's the thing. It turns out actual software is a lot less like America's Got Talent and a lot more like sitting in the corner for a long time thinking, *then* testing out your ideas on some combination of coworkers, a whiteboard, and a duck.
|
|
#
?
Mar 29, 2016 23:26
|
|
|
Arsenic Lupin posted:Laszlo Bock, talking about Google's research into interviews. bacon! posted:This is a pretty good one: http://www.nytimes.com/2013/06/20/b...nZG+IlY9sYUxzBg Sure, I'm familiar with that. But it's not at all equivalent to the earlier quote, mainly because there is no counterfactual. Google accepts well under 1% of people who apply to jobs there, and probably under 2% of people who even make it to a screen in the first place. They certainly don't 'A/B test' their hiring process. So looking for insights only among the pool of candidates that are accepted creates a massive bias. Sure, there's a lot of noise/random variation, and the candidates are tested on isn't in perfect alignment with the skills required to do the job well -- but you really don't have to worry too much about false negatives from candidates who walk up to a whiteboard and can't write a single coherent thing in 30 minutes. And as someone who has done about 100 technical interviews in the last 2 years, and is a senior IC, I can say confidently that there are a lot of people meeting that description (although I'm in data science and not SWE).
|
|
#
?
Mar 29, 2016 23:40
|
|
|
Arsenic Lupin posted:Laszlo Bock, talking about Google's research into interviews. Did he ever explain how they tracked the outcomes of people who failed the tech interviews and weren't hired?
|
|
#
?
Mar 29, 2016 23:40
|
|
|
bacon! posted:The second thing you said may be true, but the correlation of being a productive software engineer and the ability to whiteboard fun math puzzles like this - especially in an interview setting - is very close to zero. If they're an actual engineer and they hold a PE, you can assume that they're going to be thoroughly steeped in math to begin with. If it's just a code jockey calling himself an "engineer" without any qualifications then yes, it may be an issue.
|
|
#
?
Mar 29, 2016 23:46
|
|
|
Arsenic Lupin posted:That's the thing. It turns out actual software is a lot less like America's Got Talent and a lot more like sitting in the corner for a long time thinking, *then* testing out your ideas on some combination of coworkers, a whiteboard, and a duck. well computer science is one of the liberal arts so this tracks
|
|
#
?
Mar 29, 2016 23:58
|
|
|
bacon! posted:The second thing you said may be true, but the correlation of being a productive software engineer and the ability to whiteboard fun math puzzles like this - especially in an interview setting - is very close to zero. However fizzbuzz still remains a great weedout question. And the correlation of being a productive software engineer and the inability to psudocode up fizzbuzz is also very close to zero.
|
|
#
?
Mar 29, 2016 23:59
|
|
|
You don't have to prove "we're rejecting good candidates" to prove "our interviews don't correlate with the quality of our hires". The point being, there were some data that was predictive (I think), but interviews weren't one of them.) Yes, data is plural. Fight me.
|
|
#
?
Mar 29, 2016 23:59
|
|
|
blah_blah posted:but you really don't have to worry too much about false negatives from candidates who walk up to a whiteboard and can't write a single coherent thing in 30 minutes. Do you mean, "you don't have to worry about the outcome of false negatives" or "there were not many false negatives"? The former is a reason people posit there are problems with this typical hiring process and homogenous hiring outcomes: https://modelviewculture.com/pieces/technical-interviews-are-bullshit
|
|
#
?
Mar 30, 2016 00:02
|
|
|
A first pass for a technical interview should always involve some form of reading existing code and bolting on an enhancement or fixing a bug. gently caress this fizzbuzz and white boarding from scratch poo poo I'm tired of developers who insist on rewriting everything because they can't decipher anything but their own code.
|
|
#
?
Mar 30, 2016 00:11
|
|
|
karthun posted:However fizzbuzz still remains a great weedout question. And the correlation of being a productive software engineer and the inability to psudocode up fizzbuzz is also very close to zero. When I first learned about fizzbuzz, I said "I don't get it, no one will ever fail this question!". It hurt to be so wrong
|
|
#
?
Mar 30, 2016 00:15
|
|
|
nachos posted:A first pass for a technical interview should always involve some form of reading existing code and bolting on an enhancement or fixing a bug. gently caress this fizzbuzz and white boarding from scratch poo poo I'm tired of developers who insist on rewriting everything because they can't decipher anything but their own code. That's a very sensible idea.
|
|
#
?
Mar 30, 2016 00:17
|
|
|
bacon! posted:When I first learned about fizzbuzz, I said "I don't get it, no one will ever fail this question!". It hurt to be so wrong We mainly use it to make sure they didn't lie about knowing C since so many people think C is the same as C++, C# or even Java.
|
|
#
?
Mar 30, 2016 00:20
|
|
|
Kinda like how literally every resume for data entry has 'Microsoft Office 2012 Expert" on it but they need help changing fonts
|
|
#
?
Mar 30, 2016 00:25
|
|
|
bacon! posted:Do you mean, "you don't have to worry about the outcome of false negatives" or "there were not many false negatives"? The former is a reason people posit there are problems with this typical hiring process and homogenous hiring outcomes: https://modelviewculture.com/pieces/technical-interviews-are-bullshit No, I mean that I am confident that there are very few false negatives among candidates who perform especially poorly in particular ways on technical interviews. As mentioned, I don't do SWE interviews, but when I ask a candidate a question about expected values, and they dither for 15 minutes without ever at any point writing anything that resembles an expected value, I'm not worried that I'm passing on some misunderstood statistical genius. Believe me, I don't think that the technical interview process is anywhere near perfect, it definitely produces lots of false negatives (and somewhat fewer false positives by design), and fails to test a lot of non-technical skills that are necessary for success, especially at a more senior level. But saying that 'technical interviews are no better than chance at predicting a successful hire' is absolutely false.
|
|
#
?
Mar 30, 2016 00:27
|
|
|
blah_blah posted:No, I mean that I am confident that there are very few false negatives among candidates who perform especially poorly in particular ways on technical interviews. As mentioned, I don't do SWE interviews, but when I ask a candidate a question about expected values, and they dither for 15 minutes without ever at any point writing anything that resembles an expected value, I'm not worried that I'm passing on some misunderstood statistical genius. The difference is that every data scientist will use expected value in their day to day work. The same is not true of software whiteboarding exercises.
|
|
#
?
Mar 30, 2016 00:32
|
|
|
nachos posted:The difference is that every data scientist will use expected value in their day to day work. The same is not true of software whiteboarding exercises. as opposed to all of those other scientists
|
|
#
?
Mar 30, 2016 00:33
|
|
|
shrike82 posted:Is this actually a thing IRL? I've never met a person in Unicornland or outside that subscribes to Singularity BS. A few incredibly smart/successful people like Ray Kurzweil and Peter Theil seem to believe in it, but they might have gone a bit crazy from their success/intelligence. This article covers a lot of people who Thiel has funded who seem to believe in the singularity future. Whether they really are true believers or are just taking Thiel for a ride in exchange for his funding is unknown, though. http://www.newyorker.com/magazine/2011/11/28/no-death-no-taxes
|
|
#
?
Mar 30, 2016 02:37
|
|
|
While this awesome thread is on the topic of interviews. Does anyone have any tips of working out how on the ASD someone is? One guy slipped through my interview process who is good enough but has certain autistic tendencies that really affect his performance and cultural fit. One way I kind of dealt with this was by bringing in a female staff member to the interview and seeing how they interacted with her. Many very technical guys simply wouldn't look at her even looking at me when answering her questions. Another give away seems to be entirely too long resumes or resumes with irrelevant detail. This seems to be a huge problem with hiring tech people, but I don't seem to hear much about it. EDIT: This is all shorthand for technical guys who don't play well with others. Not about actual diagnosis etc. Everyone has important skill sets but some people, just.. don't fit. I jokingly blame autism. It doesn't have to be. But goddamn does it feel like it. It's really hard to work out in a couple of one hour interviews. Gail Wynand posted:This is why you take people to lunch..  I only do that with people I am going to hire already but just want to double check they are on the same page. That's a brilliant idea as a last interview step.Wank fucked around with this message at 03:26 on Mar 30, 2016 |
|
#
?
Mar 30, 2016 02:47
|
|
|
Wank posted:While this awesome thread is on the topic of interviews. Does anyone have any tips of working out how on the ASD someone is? One guy slipped through my interview process who is good enough but has certain autistic tendencies that really affect his performance and cultural fit. Are you trying to get us to help you skirt the ADA? 
|
|
#
?
Mar 30, 2016 02:48
|
|
|
Wank posted:While this awesome thread is on the topic of interviews. Does anyone have any tips of working out how on the ASD someone is? One guy slipped through my interview process who is good enough but has certain autistic tendencies that really affect his performance and cultural fit. It matters whether someone can interact properly and is a good fit but ASD isn't automatically a disqualifier there. A close friend of mine is clearly on the spectrum but it doesn't really matter because he's a top notch developer who works well with people.
|
|
#
?
Mar 30, 2016 02:53
|
|
|
blah_blah posted:Sure, I'm familiar with that. But it's not at all equivalent to the earlier quote, mainly because there is no counterfactual. Google accepts well under 1% of people who apply to jobs there, and probably under 2% of people who even make it to a screen in the first place. They certainly don't 'A/B test' their hiring process.
|
|
#
?
Mar 30, 2016 03:03
|
|
|
I think the real complaint with a lot of technical interviewing is just how pervasive algorithm puzzles have become, and lots of nerds who are involved in the interviewing process being terrible interviewers, using the puzzles as a crutch for good judgement. Being a good interviewer takes more practice and experience than being the one interviewed really, and plenty of organizations dont really acknowledge this. The questions are either highly trainable or "ah ha" questions that might involve knowledge of some mathematical or other trivia to solve as well as the interviewer would like. Some interviewers can't help but be clever. Sometimes its just trivia about languages that someone will be rusty on, even if it's something you "should" know.
|
|
#
?
Mar 30, 2016 03:03
|
|
|
Wank posted:While this awesome thread is on the topic of interviews. Does anyone have any tips of working out how on the ASD someone is? One guy slipped through my interview process who is good enough but has certain autistic tendencies that really affect his performance and cultural fit.
|
|
#
?
Mar 30, 2016 03:05
|
|
|
lancemantis posted:I think the real complaint with a lot of technical interviewing is just how pervasive algorithm puzzles have become, and lots of nerds who are involved in the interviewing process being terrible interviewers, using the puzzles as a crutch for good judgement. Being a good interviewer takes more practice and experience than being the one interviewed really, and plenty of organizations dont really acknowledge this. edit: BTW, does Larry Page still personally approve every hire at Google? Soy Division fucked around with this message at 03:13 on Mar 30, 2016 |
|
#
?
Mar 30, 2016 03:07
|
|
|
Gail Wynand posted:Citation needed, I've heard wildly varying stories of technical interviews at Google, from white boarding to take home code challenges to easter egg code games triggered by searching for certain Python related keywords. AFAIK all their interview process always results in an onsite, but that wasn't what I was referring to. The only truly interesting A/B test you can do here is to give offers to candidates who didn't pass your hiring bar (bonus points if you have some sort of continuous measure of how far they were from passing your hiring bar) and then see how they fare as employees (preferably as a function of how far they were from getting hired). For obvious reasons, no one does this.
|
|
#
?
Mar 30, 2016 03:45
|
|
|
The problem with the little algorithm puzzle challenges is that once one becomes popular enough you'll have incompetent people just memorizing the answers to as many of them as possible. Aside from that some of the time limits are just ridiculous. Just because I can't remember a little bit of trivia that would make doing that puzzle trivial in 20 minutes doesn't mean that I'm a lovely developer. I can see using little code tests to make sure a person understands basic coding things like, you know, the syntax of the languages, for loops, recursion, functions, etc. but some of the questions I've seen asked on code tests are absurd. Granted the other side of it is that just because somebody can figure out the little algorithm puzzles doesn't mean they can engineer gigantic programs properly or write readable, maintainable code.
|
|
#
?
Mar 30, 2016 03:45
|
|
|
Why are manholes round?
|
|
#
?
Mar 30, 2016 03:54
|
|
|
Jumpingmanjim posted:Why are manholes round? Antlers. The reason is antlers. IT JUST IS, OK?
|
|
#
?
Mar 30, 2016 03:55
|
|
|
ToxicSlurpee posted:Antlers. You're hired.
|
|
#
?
Mar 30, 2016 03:57
|
|
|
I also would have accepted 'because manhole covers are round'
|
|
#
?
Mar 30, 2016 04:08
|
|
|
Jumpingmanjim posted:Why are manholes round? so the dick will fit snugly inside it
|
|
#
?
Mar 30, 2016 04:09
|
|
|
Jumpingmanjim posted:I also would have accepted 'because manhole covers are round' Well my answer was almost going to be not speaking while spiking a ham sandwich on the table hard as I could.
|
|
#
?
Mar 30, 2016 04:18
|
|
|
|
| # ? May 17, 2024 07:01 |
|
|
To contribute - Surfstitch an Australian startup darling (not a unicorn) which specialises in online retailing of surf gear (the kids tell me it is pretty popular). In February the company announced a profit warning, the shares dropped 40% and then the CEO resigned to buy the company. This was nearly a month ago and still no news on the legalities on this. Or is this a good move to make a public company private again? http://www.smh.com.au/business/retail/surfstitch-chief-has-gone-rogue-20160310-gnfawf.html
|
|
#
?
Mar 30, 2016 04:23
|
|