Unable to load image

[Not Drama] Just felt this comment was very poignant about the state of technical discussion.

https://lobste.rs/s/ih3cwj/where_do_you_discuss_computer_related#c_4rjfse

All of my friends at uni would just stare blank-eyed whenever I spoke of anything programming related. My coworkers (and this is at a FAANG) don't give a f.ck and are just here for the easy paycheck. In general, very few care about programming. Sad.

Even when I speak with my father, he always has the "whatever gets the job done" boomer attitude and shuts down discussion.

Jesus Christ.

"I talk to ChatGPT sometimes"

https://lobste.rs/s/ih3cwj/where_do_you_discuss_computer_related#c_nsf4sc

Reminds of that boomer webm about videogames & groomercord:

https://lobste.rs/s/ih3cwj/where_do_you_discuss_computer_related#c_hr4maw

53
Jump in the discussion.

No email address required.

Yet they obsess over leetcode as a benchmark for competency.

Jump in the discussion.

No email address required.

lol it's just cause they legally can't give iq tests.

Jump in the discussion.

No email address required.

leetcode is better than iq tests because it directly tests coding ability?? there exist high iq people who are shit at coding and can't improve, just like there are high iq people who are shit writers or musicians and can't improve. literal r-slur.

Jump in the discussion.

No email address required.

Leetcode is a horrible screener if you care about technical ability. It screens for people who spend all their time to memorize answers to canned algorithmic problems unimpeded by drains on their attention such as building working software or having any clue at all how the tooling works. It specifically gets you the kind of people who are shit at building software and can't improve. The correct answer to an algorithmic puzzle at a company interview is "what the frick does this have to do with the problem domain" or "quicksort is in the standard library, dumbass".

Jump in the discussion.

No email address required.

But IQ tests are worse in every way by that standard. If 'balance a binary tree' is irrelevant, how can 'which of these images is most like the other three images' relevant? If companies used IQ tests people would memorize all the solutions to the image puzzle problems. It's still better than IQ tests.

Jump in the discussion.

No email address required.

If companies used IQ tests people would memorize all the solutions to the image puzzle problems.

No, you can trivially randomise or autogenerate IQ tests.

They're very much relevant. If they weren't relevant people wouldn't spend so much time pooping on them hilighting how irrelevant they are.

The only problem with them is that they hilight stuff that goes against the current orthodoxy, namely that intelligence is heritable, which has obvious implications.

Jump in the discussion.

No email address required.

No, you can trivially randomise or autogenerate IQ tests.

you can practice for autogenerated iq tests because the auto-generated tests will all be similar. the computer isn't a supergenius that can come up with entirely novel questions. there'll be a certain library of types of tests, and you can simply practice all of them. there would be iq test practice sites just like leetcode. of course.

They're very much relevant

fifth grade reading level. I agree IQ tests are relevant. they are just less relevant than subject-matter tests. Like leetcode. Leetcode involves programming and is narrowly scoped. There are also a wide variety of leetcode-style problems of varying difficulty, you don't need to use the same 50 easy ones - see any competitive programming ite.

The only problem with them is that they hilight stuff that goes against the current orthodoxy, namely that intelligence is heritable, which has obvious implications.

How do IQ tests do this any more than subjectmatter tests?

Jump in the discussion.

No email address required.

I mean, if I was hiring, I'd do both.

I also don't think that IQ tests are illegal, are they, for hiring in the US? That sounds like one of those reddity rumours.

Jump in the discussion.

No email address required.

If I were doing small-scale hiring, I wouldn't do IQ tests because I'd just have a scaling hierarchy of subject-related interview questions, with the expectation that everyone would fail the hardest question. I don't think I'd learn much if my close friends all took Iq tests and told me their scores - I've interacted with them enough to know how good they are at various tasks.

If I were managing a large-scale hiring operation I might include an IQ test as a source of data though.

I also don't think that IQ tests are illegal, are they, for hiring in the US?

https://en.wikipedia.org/wiki/Griggs_v._Duke_Power_Co.

I am not sure how true that is in practice though, you just get a test that's a "job performance test" that's actually an IQ test

Jump in the discussion.

No email address required.

There were posts on Blind and Stackexchange about strategies to not expose during an interview that you memorized the leetcode question.

It's polluted the ops side of tech now and filtering out actual ops guys in favor of the same cookie cutter coders who pull ops boilerplate.

Jump in the discussion.

No email address required.

I feel like internally at FAANG nobody cares about leetcode. You care about it to get in, then never think about it till your next job hunt

Jump in the discussion.

No email address required.

Because low-level algorithm design is not relevant to 99% of FAANG work which is simply gluing parts together.

The actual hard code part is done by some little weird cadre of library designers who you never see and who never have to interview because there's probably only about 5000 of them in the world and they all know each other from some obscure mailing list.

Jump in the discussion.

No email address required.

I think its cope that libraries are hard too. It's all just CRUD in different forms :marseyshrug:

Jump in the discussion.

No email address required.

It's more that libraries are where the heavy lifting is done, the most purely mathematical and theoretical stuff, algorithms, math, you know...

Like, most people who dabble in 'machine learning' are just gluing PyTorch features together. All the hard work is in the library. Most of what the dabblers are doing is just implementation.

Jump in the discussion.

No email address required.

It’s better than any other industry. At least you can study for it.

Jump in the discussion.

No email address required.

Link copied to clipboard
Action successful!
Error, please refresh the page and try again.