Well, just tell em you didn't realise. Maybe you'll get lucky and they'll give you another chance.

Couldn't hurt, right?

Never sorted strings before? Wow, big blind spot there.

Oh well, at least you've got some experience with interviews etc now.

Well at my work, one of the things we want from our new employees is for them to ask questions when they get stuck, and not just sit there and waste time. A bit different in a job interview context of course (and seems like they simply didn't explain this very well?), but there's a good chance that at this point they will say "this guy doesn't get it" and it could make them think less of him.

sort on their ascii values and as second key their length?

It's okay Cecil, we're rooting for you.

This seems to be a common thing humans do.

You'd be surprised.

I mean I've noticed it's common for people to NOT ask questions, yeah.

@Oh_Man Well it wouldn't hurt. I explained via closing email that my solution took too long because my solution was too complex. But in the end it is my fault for not understanding what kind of communication was expected. In my mind email is not like instant messaging, but the idea was to have a more open discussion about the problem. Communication skills are one of the job reqs and it seems I need to work on that. Live and learn.

@Lanth Blind spot eh? Imo writing a string sorting algorithm from scratch without STL and no C++11 features is non trivial unless you use strcmp (or write strcmp yourself), or have implemented such an algorithm before (which few have since in real life std::sort with Sand Wraith's suggestion is a good solution). Auxiliary data also needs to be sorted, not just the keys. This is on top of from scratch integer sorting, text file parsing and data structure implementation.

Why couldn't I have used strcmp? I misunderstood something the proctor said over email. Again, it's my poor communication, not tech deficiency.

@Sand Wraith Yeah! Exactly. The sorting I implemented was radix sort for strings. This took too long for me. Turns out the proctor said using an n log n sort with strcmp as predicate (like you suggested) would have been a good solution. I should have just asked if using strcmp was okay and gone with this.

Considering I only program in C which doesn't have STL or C++11...

Yeah, I would struggle trying to write radix sort (did it in uni once but not since then). Bummer.

Jesus christ I didn't have to pass any sort of this bullshit to get my job, nor do I think a person should have to... are these tests things that they expect you to pass or just exercises so they can see how you work through a problem? At my job the correct answer to the sorting problem is "import a library with efficient sorting algorithms" because you're wasting time if you write that shit yourself. Of course the idea is that you should be -able- to do that stuff if necessary, but if someone's sitting down to optimize some code I'd expect them to understand things like amdahl's law and to be thorough about it, not implement a breakthrough algorithm in a single 6 hour session.

Why must someone know Amdahl's Law? I hate that speedup-related shit. ;_;

Because there's no point in optimizing something that isn't going to help. You could easily work for a whole week on an optimization that gives you a 2% runtime speedup - but as a smart engineer you need to know that it isn't worth your time to do that work (or, in specific rare cases, when it is). Other times you might be able to speed up your code by orders of magnitude with just a little tweaking, but my point is I want my engineers to be smart and know the general concepts. They need to know the details too - what if we actually needed to implement a custom sort mechanism for some reason? But in general there's so much open source code already out there that it's more important (at least in my job, obviously there are different cases out there) to know when, why, and how to use a particular code than to be able to implement it necessarily. Amdahl's law to me isn't about calculating speedups (though obviously there's an equation for that) - it's about knowing when and when not to bother with something.

Got feedback, didn't pass. Alright, search continues elsewhere (again)!

@Vrael Yeah well, not everyone interviews like you or I would. So, I have to play their game.

Yeah, the six-hour test was weird. I typically got a series of 4-6 45-minute interviews; communication problems are a lot more obvious in that situation.

If it's any consolation, my gut instinct that any place with a recruiting procedure as inefficient and confusion-prone as the one you describe is likely not a great place to work. There's a correlation between how companies recruit people, and how people work there.

It's a computer scientist's job to solve problems, so I'd imagine something like "select the best algorithm to do this within the time constraint" is a very useful test to rate someone's ability.


You also don't write software that shoots people into space.


If you tried to implement an over-engineered solution that's a technical problem, not a communicative one.


You've never asked teachers in school for help on explaining a question? Reading instructions, and carefully, is extremely important.


It's not a game because there are things called deadlines. Being able to manage your time most effectively is a important ability when approaching any problem. This is especially true for a place like SpaceX, which is in a business where deadlines aren't things that can be flexed.


It's highly efficient, because it effectively grades people on their merit. They even gave Cecil a second chance to prove his technical ability. I think it does a disservice to be in denial about yourself.

Perhaps you should apply to work at Valve!

We need someone on the inside to leak us info about Half Life 3.

Lol jokes it's never coming out.

#hopelost

People, satellites, explosives, same shit different cargo. But my point is that these 'tests' are a little ridiculous. It's one thing if companies are just using them as an entry point to observe you work, but if you're expected to like, get a 34/34 or something to get the job that seems a little stupid to me. If they're just using these tests to see how you work through problems, interact with people, how you might fit in, etc, sure that's fine. Even if you had someone fail the test you might still really want them for your company, and if you have someone get a perfect score you might not want that person for some reason. The stupid thing is that I expect what cecil described to be commonplace - the HR department is too dumb to understand the tests and they don't collaborate enough with the actual engineers that the new hire will be working with, so you end up with some dumb process like "Hire the top 3 scores" or some crap like that.

@Sacrieur Valid points, but you don't really know the full context of the situation and are making a lot of assumptions that aren't necessarily accurate. I can't really go into the details since it's not appropriate, but I can assure you this was a communication problem. During the test I didn't know wtf was expected of me (and didn't ask) and as a result performed poorly.

@Vrael Yeah getting 34/34 or whatever is a really stupid culling method. There are a lot of reasons why something could go wrong and a variety of competent applicants aren't going to get passed this sort of method due random chance. I can only imagine it works well if someone goes through the process 2-3 times to iron out the eccentricities and understand "the ropes". Because clearly this test isn't only judging my abilities to perform the job, it's testing for a lot of interview-specific skills/knowledge along the way.

@Sand Wraith
They only hire people with proven experience

Minor update: Currently focusing on 5th Cell. Also applying to other places, but 5th Cell looks most promising. Since Thanksgiving is hitting right now can't really make any more progress until maybe next week or the week after.

What's 5th cell?

They're most notable for developing Scribblenauts.