Hacker News Reader

I had assumed it to be a clear reference to this series, although it isn't listed in the references.

https://aphyr.com/posts/340-reversing-the-technical-intervie...

Although TFA at least comes with half a page of explanations! That should be plenty, yeah? ;)

> although it isn't listed in the references.

Third item under "Further reading". (Though maybe that was added between when you wrote the above and now.)

> There will be other meetings, but you need not participate. Send an eagle in your place.

Thank you, I missed that.

Sometimes I feel like everyone who comments on HN submissions has forgotten what fun looks like.

It's correct to underline the heterogeneous nature of programming, but you don't go far enough. What you speak of is only one singular manifestation of a programming role. In this sense, what you're hiring for is expertise with a specific ecosystem, usually because you want to leverage some pre-existing piece of software that exists in that ecosystem, often times due to legacy code. This is not universal.

Someone that could whiteboard correct programs written in SKI is someone I'd be delighted to hire. Likewise someone that can write correct forth by hand. It's holding a very large compute graph in your head comfortably enough that serialising it with a pen doesn't really slow you down.

I can't do either of those things though. I'll have to stay in the high level world instead. Sadness.

> Someone that could whiteboard correct programs written in SKI is someone I'd be delighted to hire. Likewise someone that can write correct forth by hand.

This is potentially true.

Nevertheless don't confuse "the kind of person who has very nerdy interests, thus I would like to sit near to him in the office" with "candidate who will likely be the best one to do the work that has to be done in the best possible way". While I do agree that cognitive capabilities are one factor that often correlates well with the subclause "in the best possible way", there exist other factors, too. I don't claim that you mingled these two very different personal profiles, but in my opinion quite some programmers with nerdy interests do.

But if you work in an environment where being able to whiteboard correct programs written in SKI or being able to write correct Forth by hand is indeed (plausibly) highly correlated with "likely [to] be the best one to do the work that has to be done in the best possible way" I congratulate you on having a work environment that in my opinion very few programmers can relish.

I think I like teams with specialisms. I can think of a few instances where things went better because people were doing things they loved, where other people would have really struggled with the same task.

I'm on ten years of trying and failing to make any sense of cmake. As a nominally C++ developer, I owe being able to contribute to the projects substantially to other people managing to make cmake do things, and periodically baby-stepping me through trivial changes to it. I'm eternally grateful to the many people who have helped me with that, though I'd also like cmake to die and go away forever.

I remember James being the right guy for release checklists. Every release he'd read the checklist (not try to remember it) so saw when it had changed. Did all the things on it, even when some sounded redundant, or couldn't possibly be worth checking. Took hours, showed every sign of enjoying the process.

I also remember Richard for wizardry with procedural arithmetic. For loops nested six deep with exciting integer arithmetic scattered across the levels implementing some AI thing, written with no indication that he was losing track of the details. Would cheerfully review code of unreasonable complexity and pick out errors in it from the browser diff.

I'm in compiler dev. Lots of applied graph theory really. Hypothetical SKI/forth chap would have better spacial reasoning than me and sometimes that's the limiting factor in the work. Thank you - it is fortunate that our current environment is supportive of language implementation work, long may that remain the case.

Having written a small amount of SKI, long ago, I think you can avoid having to visualise the entire graph by reasonably modular approach: naming functions and then using those as building blocks.

Names are very helpful. You get to design your own high-level world, but then you get to use it. Since you can replace any name by its definition in SKI, it's easy to expand, just very tedious.

>> For 2: In particular since programming is separated into so many "cultures" of which many of them take deep effort to get more than a superficial knowledge (that's why you talk of "Java shops", "Microsoft/C# shops", ...), in a programming interview, you also want to see whether the candidate will likely be a fit for the prevalent programming culture in the respective company. If you advertise a JavaScript position, it is highly unlikely that if you are that deeply into combinatory logic, you will be a good fit for this position - and thus he would very likely (for good reasons) be rejected if he came up with such a solution.

> Below x IQ only

Pay enough money and I'll dance like a circus bear in your favorite OOP/FP/FRP flavor.

Unfortunely that isn't what happens in most cases.

Instead we get a bunch of frustated people that think they are equally valuable as any other FAANG founder, or they are going to be the next SV wonder even though they aren't located in SV, and try to impose their views of the worlds on the candidates.

Also in the end, the actual work has nothing to do with the coding interview.

> Also in the end, the actual work has nothing to do with the coding interview.

Of course, for very obvious reasons a programming interview is different from the actual work, but if the test designer did not make a serious attempt that success in the questions that are tested are is much correlated as possible to the actual work, it is in my opinion rather likely that the test should be redesigned.

What test designer?

I have seen people requested on the spot to join a meeting interview where the candidate is already on the room.

> I have seen people requested on the spot to join a meeting interview where the candidate is already on the room.

Such kinds of interviews of course exist, but they often rather serve the purpose of chatting so that one can see whether the candidate gets on well with the future colleagues.

"Improvised" programming interviews are, of course, often a bad idea, but nevertheless these may also serve a purpose: do some more open-ended discussions about programming topics the candidate likes so that the colleagues/company can see whether they would profit from this kind of knowledge.