Slashdot Mirror
Programmers Are Confessing Their Coding Sins To Protest a Broken Job Interview Process (theoutline.com)
A number of programmers have taken it to Twitter to bring it to everyone's, but particularly recruiter's, attention about the grueling interview process in their field that relies heavily on technical questions. David Heinemeier Hansson, a well-known programmer and the creator of the popular Ruby on Rails coding framework, started it when he tweeted, "Hello, my name is David. I would fail to write bubble sort on a whiteboard. I look code up on the internet all the time. I don't do riddles." Another coder added, "Hello, my name is Tim. I'm a lead at Google with over 30 years coding experience and I need to look up how to get length of a python string." Another coder chimed in, "Hello my name is Mike, I'm a GDE and lead at NY Times, I don't know what np complete means. Should I?" A feature story on The Outline adds: This interview style, widely used by major tech companies including Google and Amazon, typically pits candidates against a whiteboard without access to reference material -- a scenario working programmers say is demoralizing and an unrealistic test of actual ability. People spend weeks preparing for this process, afraid that the interviewer will quiz them on the one obscure algorithm they haven't studied. "A cottage industry has emerged that reminds us uncomfortably of SAT prep," Karla Monterroso, VP of programs for Code2040, an organization for black and Latino techies, wrote in a critique of the whiteboard interview. [...] This means companies tend to favor recent computer science grads from top-tier schools who have had time to cram; in other words, it doesn't help diversify the field with women, older people, and people of color.
Would be a trial (as in free trial). Throw a fairly standard problem at them, but not one with a simple, common place implementation. Drop them at a computer with internet access, give them a couple hours, and see what they have at the end of it. It's not perfect, but it's probably a better way to evaluate skills.
I'm an employer. I've interviewed nearly everybody we employ at my company. And treating a hiring interview like a rote memory exam is a terrible way to qualify a potential developer hire!
What do programmers actually do? Try testing that!
We do "whiteboard style" for part of our interviews, but only to cover basic comprehension of algorithms. More than anything, we look for basic familiarity with logic structure, and the demonstrated ability to solve problems. Our coding section of our interview process is in the subject's language of choice, including pseudo code, and is "open book" - we want to see what happens when the dev runs into a problem they don't already know! (Critical test: can they come up with a working, supportable algorithm for a problem they don't yet already know an answer for?)
After 20 years of programming experience, I STILL routinely look up the order of arguments for function calls via Google. Who cares to remember when Google has the answer in 0.10 seconds?
Test what the devs will actually DO in an anticipated normal work day and make your decisions based on that.
I have no problem with your religion until you decide it's reason to deprive others of the truth.
Nothing of course. But this is what passes for journalism in the 21st century.
It's infected just about everything including science fiction. Pointless to argue with them since they are full of ideological fervor for their activism. My relatives emigrated from the former Soviet Union and told of similar stories where you repeated the Communist Party line no matter how ridiculous it sounded.
What I've always found funny about this interview process is that the assumption is always that the interviewer knows the correct answer(s) to the question. It's painful when they don't.
Years ago I interviewed at Google and was asked a question about bit counting (some variation on "given a bit vector, wat's the fastest way to count the number of 1s?"). I quickly answered, "well, if your processor has a population count instruction, stream your vector through that and accumulate the result in a register". Having just evaluated bit counting methods as part of my Ph.D. dissertation, I knew this was the fastest way to do it, assuming the instruction was available (it's not on x86, but is on Power/VMX and most DSPs support it as well).
After I got a blank stare back from the interviewee, I said, "Oh, you were looking for the lookup table answer". We could have left it at that, but he went on to explain using some very convoluted logic how the lookup table would actually be faster than using a dedicated instruction and that my answer was clearly wrong. I mentioned a little bit about the number of cycles required for each approach, but he had none of it. In his mind, I had the wrong answer, even though my second answer was what he was looking for.
It was at that moment that I realized Google was not going to be a good place to work.
-Chris
Hello, my name is Anonymous Coward. I am a software developer.
I was talking to you about a database kernel position which is a field I have spent much of my career in and in one case was a key member of the design/coding team developing the initial (and several subsequent) releases of a revolutionary enterprise class scalable DBMS which is still being sold over thirty years later and in production at many of the biggest companies. You, of course, must have known that because you contacted me and, unusually, I decided to stop by and chat with you because I have a soft spot for startup companies, database, and engineering managers who look around seeking talent themselves rather than relying solely on recruiters. You asserted that your startup was developing an enterprise class scalable DBMS system so it seemed like it might be interesting.
You then asked me to write, on the whiteboard, a 'debit/credit' function that took three arguments and returned, IIRC, the amount transferred (admittedly that was a little odd). The first two arguments were account objects (passed by reference if I recall) representing the source and destination accounts respectively and the third argument was an amount (presumably in the same currency as both accounts were maintained in) to transfer (a float if I recall -- that, of course, is likely stupid in itself when dealing with currency).
I, understandably, was perplexed -- you must want something more than this simple function (I'm not fresh out of high school looking for a summer job) and I tried to get more requirements out of you and got none except when I asked if this function was to provide its own concurrency control or if a higher level would have insured both accounts were sufficiently locked you said something like "Okay, you need to provide the locking" and agreed that I could assume existence of "Acquire/ReleaseWriteLockAccount(AccountObject)" functions.
Okay -- so I'm thinking this is a concurrency problem, no problem (still lame, but okay...). Obviously I need to deal with deadlock potential so I ask if I there is some unique attribute of each account (you didn't give me a description of the account object's public members) that had a well defined order that I (and other developers) agreed to use to order locks when locking multiple accounts to eliminate or at least reduce deadlocks. You weren't prepared for the question so I asked if I could assume (I think) that the account number was a public field, unique among all accounts, and supported the LessThan operator with traditional semantics. You said, sure.
I wrote the code making sure to lock the accounts in order of account number, transfer the funds, unlock correctly and return the required value.
You were not happy -- what you really wanted me to check, it turns out, was that the source account would not end up with a negative balance and fail to do the transfer and return 0 as the result.
What you didn't seem to realize is that in the real world cash balances on individual accounts go negative sometimes. In fact, at that instant, I had one (of multiple) accounts at one brokerage firm which had had a small negative cash balance due to an inter-account transfer out for over six months and no one cared because I had multiple accounts at that firm with a net value of well over a million dollars and cash balances (well, "cash equivalents"), overall, of a few hundred thousand dollars.
Both of us made a decision in that discussion. You decided you didn't want to hire me. I decided I would never work for you or your company as I didn't want to risk working on/at a team/company which was so poor at evaluating people and would pass on good developers who know how the real world works and would likely hire unqualified people who answered your kindergarten questions the particular way you wanted.
Of course, I was glad that we both made those decisions because your company seems to have never got past a very modest A round and seems to have disappeared off the face of the Earth quietly.
Why is there an "insightful" mod and why isn't it "-1"? If I wanted insight, I wouldn't be reading
But bubble sort is a horrendous sorting algorithm with no practical applications.
I heard this one once. You have a bunch of straight lines that go from x=0 to x=1 and start and finish at various points on the y axis. Naturally some lines cross. Start with the lines ordered by their y coordinates at x=0. Then replace the y coordinates with the value at x=1. If you then bubble sort them to get them in the correct order, every exchange corresponds to an intersection of the lines.
So, if you want to find the intersections, I don't believe there is a more efficient way of doing it (proof needed!). Bubble sort has a particularly large number of exchanges even for an O(N^2) sorting algorithm, but you can't escape that cost in this case since that many intersections exist.
Fun fact.
So, if you're doing something with line intersections then there is a use for bubble sort.
So, there is like ONE practical use for bubble sort in some obscure use cases. Taa-daa!
SJW n. One who posts facts.