Comment by seattle_spring | Hacker News Reader

Comment by seattle_spring | Hacker News Reader

Google, Front-end interview:

Interviewer: "Prove to me that a random number is truly random."

Me: "uhhh, I'm not really sure where to start with that one."

Interviewer: "Prove. To me. That a random number. Is. Truly. Random"

Literally just me floundering because the interviewer wouldn't give me a clue as to what he was looking for.

------------------------------

Docusign, front-end interview, after phone screen:

Interviewer: "Write a program using Coffeescript, Jade, Stylus."

Me: "I have a ton of experience with FE engineering, but 0 experience with those specific technologies. It would take me a week+ to ramp up sufficiently to write a program I'd be proud to submit."

Them: "That's ok, some of our candidates spend 2 or more weeks on this problem."

Me: "Ok, well... uh, I'm going to go take this other job and not spend 2 weeks on song-and-dance for a B-tier startup."

Regarding the first question, fundamentally you can’t “prove” a number is truly random. Even if a number or sequence is completely not random looking (e.g. a valid copy of Windows 98), there is a chance it really is random and coincidental. You could have immediately said that. Did you?

But, given that this was a front end job, a common way to visually test the quality of a pseudo-random number generator is to generate a series of numbers and plot the results with pixels visually[1]. If the PRNG is high quality you’ll get a nice even static field. If there is a bias it will show up as a gradient or pattern in the pixels. I suspect this is actually what the interviewer was after.

[1] http://judeokelly.com/simple-visual-random-number-test/

That was literally my first guess, and he flatly said it was wrong and not reliable, and to try again. I'm not joking here.

Wow, that first question would leave me totally stumped and going into a rant about determinism while I watch the interviewer slowly mark my application with "do not ever hire this guy"...

If you can prove a number is random then it can't be random... therefore since no numbers can be proven random then they are all random numbers.

I think a better question could've been: "Is [ 1, 1, 1, 1 ] a random sequence?". Just as a filter question to see if they understand randomness, then if it mattered to the position, dive further into PRNG, etc

Reply on news.ycombinator.com