Employers that do this are ridiculous. As are educators, too. What is this, preparation for when coders are kidnapped by terrorists and forced to recite how to tar/untar stuff on command? (relevant xkcd, of course)
In my experience, they're usually startups or smaller companies that are trying to "copy google" in their hiring practices, but don't fully think out how they're doing it. It's not a good sign.
:-D you never know!
Unless they are trying to hire a mentat.
Personally, I think it's as absurd as writing shell scripts on a computer with no `man`, or without being able to refresh your memory with `<command> --help`. Good luck with that!
I was in an interview where they asked me to configure specific firewall rules on a notepad with no access to a terminal, so I couldn't check man iptables for stuff like how to match certain sources/destinations hosts/IPs. And I was not too familiar with the opts (was it --dest, --dst, or -d? and how do calculate the network mask for this range of IPs?). It was also for a junior position so I thought they were taking a piss at me and simply told them so.
The candidate was kind enough to avoid posting the name of the company (to hopefully hide the walk-through from Google queries for "crack me hiring test company name", but an HN discussion might then arise around the walk-through, in which the company would be named, and well, there goes your test.
It's hard to accurately run "take home" tests when hiring because of just this.
You're asking a potential candidate to spend hours cracking your program but you can't be bothered to change it for each batch?
I expect any company applying this interview technique to at least put some effort into it and even then it pays off since a batch could contain hundreds of candidates and you only have to build the program once.
I strongly suspected that sometimes goals of the person responsible for hiring are not quite aligned with the company they work for. Example:
Company's goal: Hire an exceptional candidate.
Lazy or overworked hiring manager's goal: Reduce the size of the stack of 1000's of resume's I have to go through using any means possible that will make it appear like he/she actually did the work to find the candidate.
"Lemme see, use keyword search to ditch the ones without 'Github' on their resume'. Now ditch the ones who don't have a college degree. Still some resume's left? OK, how about tossing the ones without a CS degree?"
Run a filter with enough conditions and you can get that pile down to nothing pretty fast. If that is still not enough filtering, ask the candidates to do something very time consuming. This a great way to weed out people who are very busy or exceptional at what they do and don't need to jump through your hoops.
I'm surprised they didn't make him sign an NDA for it...
Make it slightly wrong, or obviously fingerprinted.
>>I've had a "real-life" test where I wasn't allowed to verify or look up information, or where I'm giving a very short time to execute, and I've always thought those were absurd.
You certainly haven't worked at major IT services firms here in India. I spent early part of working years (around 2006) coding by purely reading books and discussing things with folks on internal mailing list. Because only managers and levels above were allowed internet access.
>>The idea that you would ever have to do something like this in a bomb-diffusing type of scenario with no internet and a ticking clock, would only test if you have already solved the problem before, not whether you could investigate a new problem.
This is unfortunately true even for coding interviews. Most algorithm questions are impossible to answer unless you have rote memorized the algorithms and their implementations.
By complex, I mean the question should be difficult enough that you can't easily hold every parameter, case and step in your head: I want to see how you deal with situations that you have to mentally walk through and issues that you might lose track of.
By potentially elegant, I mean both brute-force and optimal solutions should be able to fit on a small whiteboard, without requiring more than 2-4 variable declarations and no more than a nested for loop (or some sort of recursion). The optimal solution shouldn't require putting together pieces that they aren't really expected to know/use, i.e. no more complex than arrays, maps, strings and ints, and any other "data structures" the candidate ought to know, given their background. For instance, if they claim to be a crypto expert, they should know how to use a black box that gives them private and public keys.
By straightforward I mean, you don't want the candidate to have to have an "aha" moment. If you're asking for them to write a max-flow variant, you'd better be explicit that you don't mind a really brute-force solution, or most of the interview will be spent trying to remember complex algorithms. Is the algorithm one that you can understand by walking through test cases/figuring out what steps you took in your mind to go from input to output? If it takes more than that, you're going to get a few false positives, i.e. "Wow she was great! No one's answered this so fast!" (because she just spent 20 hours doing this on a homework), and many false negatives.
Of course, there's tests like user:B5Geek's, where the candidate gets a PC with a slightly-unplugged network cable and has to realise this fact.
In a lot of ways, it's a matter of identity crisis for software developers. We're all, industry included, not quite sure whether we're mechanics, carpenters, architects or scientists.
I know I am far less tolerant of bullshit interviews than I once was, and would have far fewer qualms vocalizing what I'm thinking - "I don't know. If I saw this in real life I'd check Google, because it reads exactly like the sort of contrived problem used in interviews and thus is well documented, and likely solved quickly by those who have memorized it. I am not one of those. I'll take a stab at it now if you just want to see how I reason through it, but if coming up with an efficient, correct implementation is part of your criteria we can probably save both of us the trouble" - is that considered acting like an ass?
Surely there's more efficient ways of evaluating that.
I confess I don't know how I'd apply this to developers. They pass fizzbuzz, OK - what do you do next?
I know what I would do, just have a 10 minute conversation with them about technology. I can easily figure out just where a person is in regards to their development as a programmer just by listening to them talk about what they're paying attention to in the tech landscape. If they don't blink when I mention Hacker News, I know they're on the right track. Do they know what Go is, or Haskell? Do they have an opinion about Node.js?
These criteria are intended to gauge motivation and enthusiasm in the absence of an impressive track record. If they're capable but not highly motivated, then that's fine so long as you're only hiring them to do stuff inside their comfort zone. We have a Microsoft guy just like that. If they're motivated but not capable, then with coaching they'll be able to do anything. The best way to discover someone's motivation is to get a gander at what they're thinking about when they aren't working.
If you're looking for someone who is motivated and capable, you can eventually find someone like that, but you'll have a hell of a time keeping them. Developers have so many options these days that you need to have at least 3 of: competitive salary, decent perks, fun work environment, a glamorous field; in order to hold on to them for longer than a year.
Most companies are lucky to manage one, if they're very lucky two. Since salary is often the easiest thing to manipulate and honestly the most important to the employee, err on the side of paying too much. It's expensive but not having competent developers is far more expensive.
Programming is a wide field, there's much more to it than the front page discussions on HN can cover.
Of course, if you're looking for a Web developer to make "cool new stuff" for the latest fadgets..
The ability to solve common problems acts as a good proxy for experience, If they can't google how to fix a retain cycle or a broken constraint, we know they probably don't know what they're talking about.
The ability to build a good, small system on top of the broken app acts as a good proxy for development talent. If they can't easily set up a dead-simple delegate to do an asynchronous web call, they can't hack as a mobile developer.
Most importantly, if they don't care enough about the job to take two hours to whip up a super-simple app, then we don't have to waste our time interviewing them.
For example, an IBEW apprentice will do a lot of gruntwork on the job site. Pulling wire, for instance. But they'll get exposed to how things are done, to trade industry expectations, and what a project looks like from start to end. Part of the week, they spend in a classroom learning the specifics of their trade. Not just the high level theory, but the nitty gritty, based on real world applications and problems. This apprenticeship lasts from 2-5 years depending.
I don't see why this model wouldn't work really well with programming.
You don't get constant mentoring at the company, but can ask one of the developers or fellow apprentices when you're stuck with one of the tasks you get assigned. Or for example, any time a developer had a small task that he felt, I couldn't do by myself, or it was too time critical he just asked me to sit beside him and watch, listen and take notes while he did the work. How it works is different in any company but the outcome should be the same: real world experience on real projects with real clients and their time/budget constraints.
That means, once your finished and a professional yourself, it isn't a question of interest but a completely normal thing to help out the apprentices. Because that's how you got to the point where you are.
We may laugh, but starting in technical support gives you an idea of how your users are using your product, if your documentation is accurate and helpful, what the pain points for people not in the department you are destined for are, and gives you an idea of how the product works.
It also quickly weeds out the people who can't effectively communicate before they get to the programming team.
I have only seen it (co-op) used in the last few years.
Algorithm tests only measure the number of man hours the candidates spend on career cup daily. Or how good they are rote memorizing things.
There are many ways to check how good a candidate is at doing their actual work. For instance you can take a candidate to a code review and see what kind of inputs they contribute to the discussion. This alone will give you good deal of idea on how well they understand code and quality. There are other ways, like say asking them to write unit test cases for a class- That would give a lot of idea on how they think with regards to breaking and fixing things.
One more way I have after doing basic checks is pair programming, or picking a totally new problem and working with the candidate to see how they think, how they work and how good their communication skills are.
There are many other ways too. Recently I've been given coding assignments for very practical applications, though those turn out to be a little difficult and time consuming. And at times the expectations are very high. Like asking the candidate to complete a whole application in very short times. It can be hectic with your current job, and a family. But generally those are a good indication of how good the candidate will do at their actual work.
I agree, but only if you're allowed to use references/google/etc and given a reasonable amount of time to accomplish it. I've had a "real-life" test where I wasn't allowed to verify or look up information, or where I'm giving a very short time to execute, and I've always thought those were absurd.
The idea that you would ever have to do something like this in a bomb-diffusing type of scenario with no internet and a ticking clock, would only test if you have already solved the problem before, not whether you could investigate a new problem.