7 ms·

Not to put too fine a point on it, but...out of a thousand applicants--who presumably had college degrees/and or years of experience programming for other compa

by rhelz 3y ago

Not to put too fine a point on it, but...out of a thousand applicants--who presumably had college degrees/and or years of experience programming for other companies--there just has to be some who would have been successful at your company. I mean, if you lined up 600 random people grabbed from off the street, you'd find 6 to 10 of them with a genius level IQ. In fact, you'd have a good chance of finding somebody smarter than anybody in your company. You are obviously filtering too much. And it would be one thing if the system merely filtered out too many good people, but the ones the system let's pass are not the ones you are looking for! Its bad in every possible way a filter can be bad--filtering out what you want, and letting through what you don't want. > disqualified for not meeting the basic requirements What is a basic requirement? If a basic requirement is something like "5 years of Java experience" its filtering out topflight c++ and c# programmers. Good programmers can make anything work; bad programmers will remain bad programmers no matter what programming language is used. > dozens interviewed Over my 40-year career, programming interviews haven't changed at all, basically testing how much of corman, leiserson, and rivest you have memorized, and how much oof the book "Cracking the Coding interview" you can regurgitate. That was ok 40 years ago, when programming was largely creating new systems from scratch. But these days, with all the frameworks, not so much. There's 40 more years of infrastructure which has been created, and to be a successful programmer, you need to be able to leverage other people's code. But who interviews to test that skill? Again, the fact that the one person the system let through didn't thrive in your company just screams that it is not testing for what you really need. Such a "stringent" process shouldn't be a point of pride.

ThrustVectoring 3y ago
The applicant pool is not randomly selected, though. The 6-10 randomly selected geniuses are likely all happily and productively employed elsewhere, there's a competitive filter against high-level talent that you have to also get through.
bb88 3y ago
> Over my 40-year career, programming interviews haven't changed at all I think they've gotten worse actually. A lot of times they're led by 20-somethings that are looking for only the skills that are limited to their own shallow experience. "Oh so you wrote compilers in C? Too bad, we're coding node.js here. Have any experience with npm?"
whstl 3y ago
> there just has to be some who would have been successful at your company Interviewing 600 candidates is incredibly expensive. And no, you can't really filter by the resume alone. The candidates you don't want have become masters at forging credentials.
hayst4ck 3y ago
https://en.wikipedia.org/wiki/The_Market_for_Lemons https://en.wikipedia.org/wiki/The_Market_for_Lemons > Akerlof examines how the quality of goods traded in a market can degrade in the presence of information asymmetry between buyers and sellers, which ultimately leaves goods that are found to be defective after purchase in the market, noted by the term 'lemon' in the title of the paper. > Akerlof's theory of the "Market for Lemons" paper applies to markets with information asymmetry, focusing on the used car market. Information asymmetry within the market relates to the seller having more information about the quality of the car as opposed to the buyer, creating adverse selection.[1] Adverse selection is a phenomenon where, buyers result in buying lower quality goods due to sellers not willing to sell high quality goods at the lower prices buyers are willing to pay. This can lead to a market collapse due to the lower equilibrium price and quantity of goods traded in the market than a market with perfect information.
dullcrisp 3y ago
New hiring process: sample 600 people from the population at random, administer IQ tests, hire highest scorer. Voila, only geniuses!
- rhelz 3y ago
  Is there any evidence that would be worse than the way we're doing it now??
  - dullcrisp 3y ago
    None whatsoever!!
  - mrguyorama 3y ago
    Literal random hiring would probably be better than what the industry does now.
- sneed_chucker 3y ago
  A lot of technical interviewing is thinly veiled IQ testing
h02 3y ago
Yes, but we can't possibly interview 1000+ people, so we have to filter it somewhat based on what they give us: their resume. If you can build me a tool that can find in a stack of 1000 resumes a genius then I'd like to start a company with you and we can make a lot of money together. Basic requirement is at least 5 years of professional experience. Most of the people who apply don't have that. Then after that, a large number of people who make it to the coding test can't pass the first basic test, or struggle through it: generate a random string using only the standard libraries. If you can't do that then you are not a senior developer. Or they appear to be using ChatGPT and cheat through the test. etc. etc. And memorizing algorithms wouldn't help you pass our assessment, it's actually easier than that in my opinion.
- rhelz 3y ago
  > If you can build me a tool ... I'd like to start a company with you and we can make a lot of money together. If you can't tell how good a programmer is by his resume, then....why the hell does everybody use resumes to try to do it??? The kicker is you can't sort good programmers from bad programmers by keyword filtering. Asking for specific languages, or specific frameworks, is a particularly egregious sin. If one of the magic keywords is "nodejs", you'll filter out that guy who has been using javascript + PHP for Facebook for 10 years. That guy is not going to have any trouble picking up nodejs. If "5 years experience" is a magic keyword, you'll filter out all geniuses who decided that 4 years at Amazon is enough for anybody and wants to make a move. So, if you are a genius, and you are one of the 1,000 people who applied, what are you to do? If your resume is pre-filtered by keyword, there is a HUGE incentive to put those keywords in--at least their resume would have a chance to be seen by a person, and its not like he'll face any repercussions for it. Its a classic vicious circle: companies are too picky on the requirements, which prompts applicants to fudge on their resume--which prompts companies to be EVEN MORE picky in their requirements, in desperate hope that if they just put enough keywords in, they will only get the resumes they want. The industry has got to break out of this cycle. > wouldn't help you pass our assessment, I hate to put it in such stark terms, but if your assessment is ruling out good programmers and letting through bad programmer(s), its worthless. You need replace it with something that actually works.
mr-ron 3y ago
My take is the reason why there is such an emphesis on specific experience (and not just being smart) is modern web apps today are very sensitive with a slew of libraries and frameworks built on top of each other, in a multiservice environment each with different flavors of technologies. Someone coming in fresh without any experience with all of the libraries and tech are just not going to success.
- rhelz 3y ago
  No doubt you have carefully curated and assembled the stack you have because it gives you a competitive advantage. But everybody else is also trying to get a competitive advantage, so nobody is going to have exactly the same stack, used in exactly the same way, as any other company. Its true, programming has changed: we are building on top of frameworks and incorporating services today, and not doing so much development from the bare metal. But that doesn't mean that you can't hire anybody who isn't a clone of you: it means that the we have to train ourselves to have the ABIITY to just jump into a new stack, at a new company, and quickly familiarize ourselves with it. THAT is the key skill going forward. The crucial skills are not can you remember how red-black trees work, or how to use dynamic programming to implement this algorithm. And its not whether they can put the right keywords on their resume. It's can somebody dump 100,000 lines of code in your lap and have confidence that you can handle it. And also, if somebody can't join your team and be productively "plugged into the matrix" in a few weeks, that means your stack has some architectural problems: lets face it, even code you wrote 6 months ago might as well be brand-new code to you. The code base has to be such that it's easy for people to quickly read it and figure out what is wrong and how to fix it. Otherwise, even people who have been there for 10 years using that exact same stack are not going to have any success.

by rhelz 3y ago

ThrustVectoring 3y ago

The applicant pool is not randomly selected, though. The 6-10 randomly selected geniuses are likely all happily and productively employed elsewhere, there's a competitive filter against high-level talent that you have to also get through.

bb88 3y ago

> Over my 40-year career, programming interviews haven't changed at all I think they've gotten worse actually. A lot of times they're led by 20-somethings that are looking for only the skills that are limited to their own shallow experience. "Oh so you wrote compilers in C? Too bad, we're coding node.js here. Have any experience with npm?"

whstl 3y ago

> there just has to be some who would have been successful at your company Interviewing 600 candidates is incredibly expensive. And no, you can't really filter by the resume alone. The candidates you don't want have become masters at forging credentials.

hayst4ck 3y ago

https://en.wikipedia.org/wiki/The_Market_for_Lemons https://en.wikipedia.org/wiki/The_Market_for_Lemons > Akerlof examines how the quality of goods traded in a market can degrade in the presence of information asymmetry between buyers and sellers, which ultimately leaves goods that are found to be defective after purchase in the market, noted by the term 'lemon' in the title of the paper. > Akerlof's theory of the "Market for Lemons" paper applies to markets with information asymmetry, focusing on the used car market. Information asymmetry within the market relates to the seller having more information about the quality of the car as opposed to the buyer, creating adverse selection.[1] Adverse selection is a phenomenon where, buyers result in buying lower quality goods due to sellers not willing to sell high quality goods at the lower prices buyers are willing to pay. This can lead to a market collapse due to the lower equilibrium price and quantity of goods traded in the market than a market with perfect information.

dullcrisp 3y ago

New hiring process: sample 600 people from the population at random, administer IQ tests, hire highest scorer. Voila, only geniuses!

rhelz 3y ago
Is there any evidence that would be worse than the way we're doing it now??
- dullcrisp 3y ago
  None whatsoever!!
- mrguyorama 3y ago
  Literal random hiring would probably be better than what the industry does now.
sneed_chucker 3y ago
A lot of technical interviewing is thinly veiled IQ testing

h02 3y ago

Yes, but we can't possibly interview 1000+ people, so we have to filter it somewhat based on what they give us: their resume. If you can build me a tool that can find in a stack of 1000 resumes a genius then I'd like to start a company with you and we can make a lot of money together. Basic requirement is at least 5 years of professional experience. Most of the people who apply don't have that. Then after that, a large number of people who make it to the coding test can't pass the first basic test, or struggle through it: generate a random string using only the standard libraries. If you can't do that then you are not a senior developer. Or they appear to be using ChatGPT and cheat through the test. etc. etc. And memorizing algorithms wouldn't help you pass our assessment, it's actually easier than that in my opinion.

rhelz 3y ago
> If you can build me a tool ... I'd like to start a company with you and we can make a lot of money together. If you can't tell how good a programmer is by his resume, then....why the hell does everybody use resumes to try to do it??? The kicker is you can't sort good programmers from bad programmers by keyword filtering. Asking for specific languages, or specific frameworks, is a particularly egregious sin. If one of the magic keywords is "nodejs", you'll filter out that guy who has been using javascript + PHP for Facebook for 10 years. That guy is not going to have any trouble picking up nodejs. If "5 years experience" is a magic keyword, you'll filter out all geniuses who decided that 4 years at Amazon is enough for anybody and wants to make a move. So, if you are a genius, and you are one of the 1,000 people who applied, what are you to do? If your resume is pre-filtered by keyword, there is a HUGE incentive to put those keywords in--at least their resume would have a chance to be seen by a person, and its not like he'll face any repercussions for it. Its a classic vicious circle: companies are too picky on the requirements, which prompts applicants to fudge on their resume--which prompts companies to be EVEN MORE picky in their requirements, in desperate hope that if they just put enough keywords in, they will only get the resumes they want. The industry has got to break out of this cycle. > wouldn't help you pass our assessment, I hate to put it in such stark terms, but if your assessment is ruling out good programmers and letting through bad programmer(s), its worthless. You need replace it with something that actually works.

mr-ron 3y ago

My take is the reason why there is such an emphesis on specific experience (and not just being smart) is modern web apps today are very sensitive with a slew of libraries and frameworks built on top of each other, in a multiservice environment each with different flavors of technologies. Someone coming in fresh without any experience with all of the libraries and tech are just not going to success.

rhelz 3y ago
No doubt you have carefully curated and assembled the stack you have because it gives you a competitive advantage. But everybody else is also trying to get a competitive advantage, so nobody is going to have exactly the same stack, used in exactly the same way, as any other company. Its true, programming has changed: we are building on top of frameworks and incorporating services today, and not doing so much development from the bare metal. But that doesn't mean that you can't hire anybody who isn't a clone of you: it means that the we have to train ourselves to have the ABIITY to just jump into a new stack, at a new company, and quickly familiarize ourselves with it. THAT is the key skill going forward. The crucial skills are not can you remember how red-black trees work, or how to use dynamic programming to implement this algorithm. And its not whether they can put the right keywords on their resume. It's can somebody dump 100,000 lines of code in your lap and have confidence that you can handle it. And also, if somebody can't join your team and be productively "plugged into the matrix" in a few weeks, that means your stack has some architectural problems: lets face it, even code you wrote 6 months ago might as well be brand-new code to you. The code base has to be such that it's easy for people to quickly read it and figure out what is wrong and how to fix it. Otherwise, even people who have been there for 10 years using that exact same stack are not going to have any success.