I think this used to make more sense when it wasn't so obvious you had to prepare. Now that top tech companies are literally telling people you to prepare ahead of interviews, sending preparation guides and flexibly scheduling your interviews as to give you time to prepare, this doesn't really hold as true.

Two things - there's no way out of the problem that anything that requires time and effort is going to filter for people with time and that to some extent any test of any kind that is hard to pass is going to be a filter for having prepared at some point in your life. And you cannot avoid testing for things that take time and effort - time and effort are what make anyone actually good at anything. It takes a lot more time and effort for you to get to a point where passing these interviews requires a couple of months of preparation than the couple of months of preparation itself. That filters even more people out. If you're trying to measure people exactly as they are, that's more unfair, because you're effectively saying, I'm going to try to hire people who already had the opportunity to be exposed to the kinds of things I want in engineers. And there's no way that opportunity isn't skewed towards the more privileged in the first place. Preparation that is to available to everyone makes things substantially more fair if you care about that sort of thing.

Second, in terms "older, female, poorer" - 1) virtually no one who's qualified at that level is "poor" in any real sense and my experience is that prestigious startups with a less rigorous technical test skew substantially wealthier in terms of their employees, mostly because paper credentials are much more correlated with wealth, 2) my general experience is that a higher percentage of men are resistant to having to study for things like this and requiring preparation and have preparation guides widely available help women on average - my experience as an interviwer is that people who show up obviously unprepared are disproportionately male, 3) age is a factor but not directly so - it's a factor in part because it's correlated with parenthood (and other caring obligations), complacency and entitlement some people develop later in life and possibly the opportunity cost (maybe you're already doing well in your career). Parenthood is probably the only thing that you may want to legitimately control for, especially for primary caretakers, but you could also make an argument that being a primary caretaker does have work implications.

This isn't a problem for colleges because their goal isn't to be fair - their goal is to be unfair. The goal is to to accept people that are already going to be successful. And obviously people in a dysfunctional family situation where they have to work 25 hours/week on a job that cannot be showcased in a college application, are going to have a harder time being successful later in life. If they truly are the type that can succeed, they probably could figure out a way to sell the unusual experience (lots of high school students work, but almost always to fund their own spending, not to feed their family) in a compelling way to admissions officers.

In short, everything is unfair and requiring a couple of months of preparation is probably more fair than anything else that goes on in life - I mean, consider in comparison what it takes to become a doctor or a lawyer and what kinds of people can afford to go through these processes. Unfairness in an individual sense is a fact of life and I don't know what you can do about the fact that acquiring skills takes natural ability, time and grit and these things are not equally distributed, not even close.

Unfairness as it applies to underrepresented minorities and protected classes is worth bearing in mind and trying to correct for but I'm not really qualified to opine on what can be done here as I don't fully understand the labor laws surrounding affirmative action. Other than that if you're an industry with a pipeline that has a skewed ratio, there's a very fine line you have to walk. My understanding is that for large companies, it's both 1) illegal not have any kind of affirmative action and end up with ratios that are completely out of whack as with respect to protected classes and 2) illegal to have an explicit form of affirmative action in the filtering process that favors underrepresented minorities to correct the ratios.

I feel like you've been talking past my point. It shouldn't be controversial that an ideal interview process would be one in which people that are currently doing the job well would pass without any preparation*. And thus, one that doesn't have that property and instead requires an extra step of X hours of preparation, is clearly flawed.

Slightly less obvious, but still non-controversial imo, an interview process should consider the totality of a candidate's experience / skills / abilities relevant to performing the job more valuable than a single narrow trait (like hard-working-ness).

Again, maybe what we have is the best practical solution or not, but it's still flawed and can be legitimately criticized as such.


* Note: This doesn't need to be the only way to pass said interview (as you said - a transparent interview process that you can prepare for is a beneficial feature).

You're simply wishing away the complexity of human life. It's like saying an ideal sorting algorithm would has the time complexity of O(1). Sure that would be nice, but that doesn't mean the merge sort's totally not ideal time complexity is a red flag. Nothing about life works that way. People don't show up to the first date exactly as they show up to a dinner with a partner of 10 years. Product advertisements don't have footage of the product being used normally. In any competitive process, in any situation where you have to prove something to a total stranger, you are going to try hard, all of your competitors are going to try hard, and the bar is going to be set higher accordingly. And that's okay.

Also switching jobs is different from staying productive in the same job. It's possible that some of your complacent employees are productive now, but wouldn't be productive in a new, analogous situation. A new employee certainly has to learn more new things and work harder to become as productive, assuming the same level of ability and skills. The bar needs to be higher simply as a result of that, let alone all kinds of information asymmetry.

This is a meaningless statement. Everything is flawed.

A typical interview loop incorporates all kinds of other things, no one is trying to value a single narrow trait. I'm talking about this one trait because 1) it's extremely important and often overlooked and 2) it happens to be indirectly measured by an element of the process that is specifically being criticized. I'm not saying, let's try to create a process where the level of preparation is the only thing that matters, but that the level of preparation that's implicitly required is a feature, not a bug. And that a more perfect coding interview that measures how good people are at coding in a workplace situation minus preparation may actually lead to worse employees, because grit matters even more on a longer time scale.

Also one of the reasons I'm going pretty hard to defend the process and tying it to grit isn't so much that I think the process is awesome. There are literally borderline bitter posts from me in this thread where I struggled to accept this reality a few years back, having flunked some of those interviews due to a complete lack of preparation. Rather, it's because most people here and elsewhere are arguing against the process not from a genuine concern that it's bad for the companies involved, but out of a sense of entitlement - they feel they (and perhaps others they can relate to) should be able to get a job without trying that hard. And the argumentation that follows is just rationalization of that sentiment without regard for the realities involved.

So I'm tying this directly back to grit because honestly it's something all of us had at some point. All of became good at coding because we kept trying hard and that's how we're here and that's why we're even being considered for these types of high-paying jobs. We shouldn't forget that. And when we stop being able to go back that mentality where we're not entitled to anything and have to continuously get better, we're no longer at our best. If simple preparation for an interview feels like it's well above your normal level of effort, it's likely that 1) you're unchallenged and complacent at work or 2) you're also overlooking obvious ways you can be far better at your job.

This is a ridiculous analogy.

And, I guess, just for the record, I haven't done an interview for a job since my first job out of school.

Because, when it's possible, it's way better to hire people you know are good at a job (or good candidates for growing into a job) without making them do some arbitrary amount of work.

I think it's because you haven't thought this through. I'm pretty sure I've seen computational models for things like this where you can effectively demonstrate that some level of information asymmetry is computationally inevitable, that is eliminating this type of inefficiency effectively requires us to transcend the limits of computation.

I don't believe this is better, let alone way better. For effective collaborative decision-making in general, which hiring is just one form of, you need a culture where similar decisions are made based on a consistent framework that allows decisions to be made and justified based on rigorous evidence that adheres to certain principles. Otherwise, you're just making whatever decision that makes you feel good, then rationalizing it. It's like a banana republic without the rule of law.

For your hiring process to be considered part of a consistent framework, you have to have a system where you extend an offer to every single person that somebody at the company just knows is good at a job and obviously that doesn't work. Maybe this is justified at some super high level, but that can only affect the hiring process for a few people at most and higher-ups being able to bypass the hiring process for people they know often will have disastrous consequences. My experience is that some of the worst people I've worked with were hired because somebody higher up knew them from somewhere else.

It's clearly way better. Hell, it's basically a truism that it's better. And it's the reason that few internal promotions/transfers involve making people learn a bunch of stuff on their own time and then doing the same standard interviewing process you give to external candidates.

If you actually believe all the stuff you're saying here - you'd be advocating for a system where no promotion is made without the "candidate" going off to show how gritty they are and doing a standard external candidate interview process.


Of course this isn't true. There are any number of "consistent frameworks" that one could come up with that would still prioritize hiring people that are good.

Now, that's not to say there aren't "red flags" and problems with the "hire someone you know is good". It's not scalable - but it does work extremely well when you've got a core group of very talented people that are highly motivated to succeed. It can also be combined with other hiring practices to take advantage of a meaningful signal while mitigating some of the downsides that this practice can create.

Me instigating the candybar vs. jj outlast battle that breaks the internet.

Btw from the startups I've been around and interviewed with - at least in the late 2010s there seems to be a big stereotype: FAANG devs >>> startup devs >>>>>>>>>>>>> corporate non-tech company devs. (And lol now I'm a public sector dev - the lowest of the low.)

But holy cow could I put together an insanely good Dirty Dozen-esque team of devs from my 20+ years of mostly corporate jobs. If someone had a great idea and enough funding to lure these people away from their current gigs - we could absolutely conquer the world. I know they're great and I know I can trust them to own their piece but keep the big picture in mind. You can never tell that stuff from an interview. You have to go through battles together.

Most of them would probably do well enough to get hired at a FAANG - assuming no age bias. But the company who hired them wouldn't have any idea what they had until the devs actually got to work.

One - the best DBA/DB programmer I've ever worked with - is so nervous in social situations I don't think she'd ever get hired. She has that thing where she laughs at everything when she's nervous, even stuff that's not at all funny. She has to come by word of mouth. But I'd walk through fire to get her on the team if we needed a database programmer.

It's not better and the fact that you think it's a "truism" is quite troubling.

This is a bad analogy. A better analogy would be that you'd be advocating for a promotion system where an arbitrary higher-up decides, hey X did great work this year and when X did some work for me, it was great, let's promote him. As opposed to one where all kinds of evidence is compiled with respect to this person's achievements and compared against others in similar situations.

It's almost like the concept of rigor is unfamiliar to you and you missed me emphasizing this while you're holding on this whole grit thing, which is more important when you don't have direct evidence of performance. X working with Y and coming up with the impression that Y did good work (and having no evidence of this but trying to use this to justify hiring Y at another company) isn't equivalent to an objective process that compiles evidence from multiple sources and making a decision rigorously.

The concept of comparability is important - in order to make decisions of this nature, you have to put both people in the same situation. For promotion, people will naturally have been in similar situations and achievements at the same company are more easily compared. For hiring, this is close to impossible, not to mention there not being any objective evidence of performance. How does A's impression of B working at C compare to D's impression of E working at F? Are A and D even telling the truth? Are they even in position to evaluate the totality of their colleagues' performance? How do you even make that comparison? This is why you need a standardized process for interviews.

Companies that don't need a scalable process usually don't have a filtering problem in the first place. They almost always have a recruiting problem and no one good would want to work there unless they know you. This is why "people you know" is where you start - it's not so much that you know they were good, but that they know you're not a huckster, whereas most high quality candidates you don't know personally would start with a lot of skepticism. Generally speaking, companies that don't need a scalable filtering system usually aren't able to attract high-quality candidates other than people the founders and early employees know. And you're not really talking about filtering, it's more about convincing.

There's tremendous variation both within "startups" and "corporate non-tech" that it's hard to make generalizations like this - some startups are top-notch, while others have absolutely no one competent. Same with corporate - some non-tech companies take technology serious enough to invest deeply in top talent, at others it's just a cost center. And at those companies, the general talent level they target for other functions likely spills over to technology as well.

The main differences to me are that 1) the bar is consistently high enough that you can generally assume almost anyone you work with is somewhat competent across a fairly wide range of dimensions, which helps with trust and 2) the culture is demanding in a way that allows people to grow faster, 3) there's are clear growth paths and role models, which keeps the level of motivation high, 4) likely related to 3 but the willingness to do things that are seemingly outside of one's comfort zone (especially non-technical stuff) are generally high.

It basically works something like this:
https://en.wikipedia.org/wiki/O-ring...ic_development

CB, I didn’t read that essay because it seems like you just don’t know what ‘know’ means.

Are you trying to play a gotcha where you're now saying by "people you know are good at a job" you meant people that you can predict with 100% accuracy as to whether will be good? If not, I don't know where you're going with this, it seems irrelevant. If so, it's simply irrelevant to the actual reality where no one knows any such thing about anyone else. When most people are saying they know something about someone's future performance, they are just sharing some limited extrapolation based on limited evidence.

Let me help you out.


or

I think most people would agree these aren't cases where the higher up knows that X or Y would do the job well.


I think most people would agree this is a case where many different people know the person would do the job well.


In case it's not clear to you, "know" in this context means with fairly high certainty but of course people can't 100% predict the future off of only past results


Lol, nope.


Anyway, this started as an intellectually interesting discussion but like most discussions with you it quickly becomes exhausting reading walls of text and finding all the poor assumptions / twisted words / misconstrued positions to point out.

#unsubscribe

This thread is getting that slow-load thing that happens to huge threads.

Does this mean you would hire anyone who spent a few years at a reputable place? Lots of people must have looked at their total body of work they did at those places and thought they were good for them not to get fired. I mean it's weird we've kind of come full circle here, but you could probably at least say this about any senior engineer that spent 2+ years at a FAANG type company. I'm sure they are pretty good in some context. I wonder what could go wrong if you had a blanket open door policy to engineers who meet this criteria.

So, in case it's not clear, knowing with fairly high certainty is not enough, because you don't get to hire this person, you only get to hire this person if they also pick you. This is an adverse selection problem and they may not be who they once were.

Small companies that are primarily hiring people they know without putting them through any sort of interview process are almost always going to be not objectively desirable places to work. This is true statistically in the actual startup ecosystem. Some of them may turn out later to have been great places but virtually no such company is objectively desirable at this stage. Companies that have since become desirable (and are run well) aren't going to just hire "people they know are good" and not compare them objectively against thousands of other candidates that are.

I suppose there could be exceptional cases like a super successful entrepreneur starting a new company and seeding the company with extraordinary engineers they've worked in the past, but that's super rare, simply because the odds that the very best people in your network are available exactly when you need them is fairly low. The vast majority of times most people are doing this for desirable engineering positions, it's just cronyism leading to mediocre to bad hires.

.

lmao not having candybar on ignore

How I came to love the coding interview...

There's more or less three things that determine if an employee is going to be a good hire.
1. Raw intelligence - a really smart person will figure out how to solve a problem better than a less intelligent person
2. Experience or domain knowledge - if you've encountered a problem before, you have a good chance at coming up with the solution for your new employer. This solution could even be better than the one a smart person figures out since you have less risk at running into a new pitfall.
3. Diligence - a hard worker will find a way to grind it out.

In a coding interview which is usually only a few hours long, the easiest thing to test for is #1, intelligence. You can give a classic leetcode problem and immediately filter out quite a few people who simply won't be able to get the job done. Of course, the candidate can game the system a little by practicing these types of problems. There are only so many questions that are short enough to be accomplished in the time given while being hard enough to act as a reasonable filter for intelligence. But if someone puts in the work and trains themselves well on the top 100 leetcode problems, they aren't really gaming the system. They just simply allowed the interviewer to judge #3 diligence instead of #1 intelligence whether the interviewer knows it or not.
Judging a candidate for #2 experience is probably most difficult to determine in a reliable way in a short interview process. I think this is what most industries outside of coding interviews attempt to interview for. Someone with a natural ability to sell their ideas will excel in these interviews without actually having the required experience. You can imagine an introvert with a ton of experience getting passed over for an extrovert in a lot of interviews I bet.

But that's why we should be glad we have the coding/whiteboard interviews. We're in the rare industry that has some of the most objective interviewing. You don't even have to be that smart to get one of the top jobs. The interview questions are all completely solved on the internet. All you have to do is grind it out and learn the top 100 questions from leetcode and you can make a top 1% salary in the world.

As a meta comment:

This is a good example of a strong personality anchoring a conversation. There’s way more than this. How you work with others. How you adapt to changing situations/technology. Leadership abilities. How you handle stressful situations. Communication skills. How good you are at finding/adapting solutions to your problem. Logic skills. Etc. Etc.

Grit shouldn’t beat many of those (even if you group them into higher level generalizations) out in a top 3 of what makes a good hire.