robertbieber

Probably never been a better time in history to be in the market for a new hand plane, either

Pretty much any custom furniture maker is going to have and use hand tools when appropriate though. Sure the really mass produced stuff is all coming off power feeders and CNCs, but I'd wager most one off pieces are seeing at least some chisel/plane/handsaw usage even if they're mostly built with power tools

I dunno, hopefully one day we'll get a good opportunity to disambiguate. But in that 2021-2024-ish timespan all the big tech companies were ramping up their hiring like mad and it was definitely significantly easier to get in. I just can't imagine that not having an impact on the way recruiters and hiring managers see those companies on a resume in the future

It kind of seems like this is becoming less true nowadays. I spent the first ~decade of my career at FB/Snap/Stripe, and for a long time I could pretty much guarantee an interview from just dropping my resume on any SWE listing I found. Nowadays, very much not the case. Not sure how much of that is just the overall market being down, but I suspect at least part of it is that the big companies hired like crazy during the pandemic and now recruiters don't assign as much value to just having gotten a job at one of them

Every tech company I've worked for in my entire career has made me sign an NDA

This is an extremely weird amount of psycho analyzing to be doing to a stranger based on one Reddit post. It's completely normal to feel out of sorts three days after having a daily routine you've done for years abruptly disappear

> That's how you do it. If you expect to write a full spec and let AI run by itself for 10 minutes implementing it with prestine code, in am existing code base, you'll be disappointed. It'll produce thousands of lines of code with bugs and bad design - especially if the AI doesn't have any coding guidelines etc. tbh this is much less the case today than it was like 6 months ago. I've been gradually pushing the scope of how big a task I can pass off to Claude code and getting surprisingly good results out of multi-hundred-line changes. It still makes dumb mistakes, but increasingly the dumb mistakes are things like "hey this is suboptimal you need to batch these awaits" rather than "this library method you used does not exist."

I really think this is just a question of company culture. The first ~decade of my career was all big tech, and not to overly self-promote but I think I was fairly successful in some demanding environments. First job after that was at a startup that had some really bizarre/clunky architecture, a ton of tech debt and mostly more junior engineers. I was very careful about not doing the whole "I have arrived from Big Tech to enlighten you" trope, but I expected that at some point the team would come to appreciate the experience behind my suggestions. That point never came. I got shuffled around a bunch of chaotic and mostly unimportant projects, never got the impression anyone from management particularly cared about the experience I had, even when I could directly explain how technical choices I was seeing made had gone poorly in the past. Current job is another, smaller startup, but night and day difference in the culture. More experienced engineers, more experienced management, and just dramatically more cultural respect for earned experience. And probably not coincidentally, *much* better architecture and general technical situation

idk man, it's pretty easy to just diffuse responsibility to the Nth degree and come to the conclusion that simply existing in society forces you to fund the government with your taxes so at the end of the day it doesn't really matter if you're personally working on the Murderbot project. I don't buy into that kind of myopic worldview. The fact that you can't preclude any possible Nth degree downstream negative impacts of your actions down the line doesn't absolve you of the moral responsibility to have some hard lines on things that you simply will not do

This feels like a slightly less egregious to the idea of "I would google it" as a master answer to every question. If you're doing an interview and you can clearly tell what the spirit of the question is, the smart thing to do is always to align yourself with that spirit. Any response that's basically presenting a reason you shouldn't have to answer the actual question is likely to come across as smug, too-clever, or mildly annoying at best

> I do think you'll still need people who understand what the code is doing. Someone will need to verify that the output is actually good, and what code to ask the LLM to create You've said two things here that aren't actually intrinsically connected. If LLMs actually achieve the same fidelity of translation from human language to high level languase that compilers have going from high level languages to machine code, no one will need to understand what the code is doing to be able to verify that the output is good. Not any more than I have to scour my compiler's output for mistakes. Now personally I don't think we're ever going to see AI coding tools anywhere near that effective (well, not from LLMs, who knows what the future holds). But if that's the direction you think things are moving, you better start brushing up on other marketable skills because I promise you just having some basic product sense and ability to interpret the output of software is not nearly as rare or difficult to teach as you think it is

Technical ability is the *distinguishing* job of a software engineer. We do other things, but the other things we do can all be taught to any literate, halfway intelligent person pretty easily. The reason we make hundreds of thousands of dollars instead of 50k and maybe health insurance if we're lucky is because if you want to build software, you *need* someone who has the specialized skills of writing code, deploying software, debugging, configuring web services, and so on. If we find ourselves in a world where there's a machine anyone can use to do those things, our unique economic position poofs right out of existence. Do you have any idea how many people there are out there with master's degrees barely making ends meet? Just being an intelligent, even highly educated person is not a rare or valuable trait. Being good at writing software is, and if that ever changes we're all toast.

> 1. High level languages made programming easier. AI makes it even easier. That's the whole point You're just completely ignoring the point you're replying to here. Making it easier for skilled programmers to write software is a completely, categorically different thing from removing the need for skilled programmers to build software at all. If the latter happens, programming will pay at best about as much as any relatively unskilled office job, because any literate, halfway intelligent person will be able to do the job

Something I've learned after getting screwed by this kind of thing a couple of times is that no explanation will ever be as good as not having to explain yourself. It's a lot easier to just play the game and make sure whatever numbers upper management cares about look good for you

> Some people will never listen Let's also be clear about the fact that being disagreed with is not the same as not being listened to. I think a lot of attitudes like the OP here come from an underlying belief that no one could possibly disagree with you if they *really* understood what you're saying, which is of course not how anything works

lmao yeah that's really aiming for the stratosphere. Even in a relatively low stress part of the company, E6 is pretty demanding. E7s are the kind of people whose output I find difficult to comprehend. E8 or E9, I'm genuinely unsure what kind of accomplishments you would be expected to deliver every quarter but it's certainly not a bar I'd want to have to meet every PSC cycle. Pay must be spectacular though

This kind of thing seemed like a unicorn five years ago, in today's market it seems like...something even rarer than a unicorn, I guess.

I've been working as a SWE since 2012, mostly at big tech companies, fairly successfully. I can count on one hand the number of times I've used a debugger since graduating college. I'm fully aware of how they work and how to operate them, but in almost every case I just find them more hassle than they're worth and end up burning more time stepping through code and getting distracted by irrelevant details. I know some people love them and use them all the time, but for me personally the only time I'd consider it indispensable is if I don't have the source code

Also, lol, definitely no one is doing brain surgery for free either

> So when someone needs to be bumped to a different rating you can either pick randomly or use an "objective" metric You're really big on the tertium non datur fallacy, aren't you? There are, in fact, a wide range of options for bubbling up subjective evaluations and leveling them across orgs. Certainly a hell of a lot more options than "use LOC or pick at random." Generally the way it's done is you get all the managers from one level into a room and you have them discuss among themselves to calibrate their ratings, then you do the same with the next level of managers, and so on and so forth until you've got the entire company's ratings set. Obviously that's a messy, deeply flawed process, but at least it maps in some general way to real world performance, as opposed to meaningless metrics that you may as well have plucked out of a random number generator

Then someone has to make a hard judgment call. That sucks and it's obviously imperfect, but dramatically less so than just cutting whoever committed the fewest lines. At that point you may as well just use a dice roll as your metric, since it's also objective but at least the engineers can't game it by committing tons of junk volume

> Subjective judgments also aren't helpful to the people being judged. "I think you're not as good as Mary" isn't actionable. Also, like, this just straightforwardly isn't true? You've obviously just put forward a worthless subjective evaluation as a strawman here, but in the real world actual subjective evaluations can be very detailed and actionable. For something closer to a real example: * Jake committed 10k lines this half, almost entirely cleaning up old feature flags. He was responsible for one small project which didn't ship. He keeps saying it will be ready soon, but consistently fails to update stake holders when projected dates slip. * Mary committed 3k lines this half. She delivered two major infrastructure improvement projects that will save the company about 50k a year. Her estimates were mostly correct, and when projects risked missing deadlines she notified everyone and explained what was going on immediately. She also helped onboard two new grad team members who are now making quality contributions to the team. If in this scenario you pick Jake because his higher LOC count is an "objective" measurement despite being inversely correlated to his value to the team...I just don't even know how to respond to that kind of silliness. It's completely baby-brained nonsense

You realize calibrating employee ratings across broader orgs is something pretty much every tech company big enough to have an HR department does multiple times per year, right? And those calibrations don't have to resort to nonsense like ranking by line counts

A lot of people don't really account for the fact that contributing to a healthy team environment in ways that aren't just technical is also an important form of work

It's fascinating that given a range of available criteria, you would pick one that's "objective" even if it's quality is known to be worthless over subjective judgments that are at least in some way nominally tethered to reality

I mean sure it's always possible that everyone at the top of the industry just sucks at interpreting metrics, but I've never seen any indication that's the case

No one ever really seems to account for the fact that at bigger companies they do, in fact, track the correlation between interview performance and job performance. Early in my career at FB, for insurance, they stopped giving system design interviews to new grads because the data wasn't showing that they really predicted job performance. When you're hiring thousands of engineers, you can actually get some pretty good numbers on these things

If there are any external investors the equity has *a* value, it's just a much more speculative value than at a public company

Most jobs are more precarious, significantly lower paid, and easier to evaluate than software engineering. If you lie your way through an interview to get a line cook job and you can't actually cook, you're just gonna get fired your first day. No benefit/401k enrollment to deal with, no severance, it just ends up costing whoever hired you like half a day's pay and some frustration. In a SWE job you've got a pretty significant onboarding and ramping up period during which you're collecting a salary and benefits. Even the worst engineers I've seen get hired at big tech companies make it at least a few months before getting fired. You probably wouldn't make it as long at a small startup, but it'll still be a lot more money at the end of the day than most jobs

I dunno man, no one likes coding interviews but I have yet to hear the mythical ideal solution to the basic problem of taking a zillion resume submissions and deciding which ones to take to fill a limited number of positions. Usually the answer is some variant of they should go by what's on my resume/what I tell them about myself, but looking good on paper is a lot easier than actually being good, and of course everyone talks themselves up in the interview. The comparison to actual engineers always comes up, but you know what else actual engineers have? Education requirements and professional licensing that comes with rigorous testing, and legal liability for the quality of their work. In software "engineering," I'm literally just some guy walking in off the street as far as a prospective employer knows. Everything on my resume including references would be pretty easy to fake if I were so inclined, even easier to exaggerate. There's no licensing body certifying I can actually do what I claim to do, so ultimately I just don't get that worked up about someone wanting me to demonstrate that I can code before offering me a high paying job. There could conceivably be a future where software engineering becomes a real engineering discipline with licensing and certifications and all that associated overhead, but I have a feeling that most of us wouldn't be thrilled with that either

What cost? This isn't a trap door decision, if AI coding turns out to be the bee's knees down the line you can just start using it then, it's not like you're going to be incapable of typing code requests into a chat box if you don't get started *right now*

This is 1000% dependent on the org/team you end up on and your ability to steer away from drama/doomed projects and manage your time/expectations/career trajectory. It's very easy to go into big tech and end up working 80 hour weeks forever chasing the next promotion, but you can also get to a terminal level, establish healthy working hours/expectations, and have a very manageable work/life balance. Granted, I haven't really been in that world since the first big round of layoffs post-COVID, and I'm sure things are probably tougher across the board at this point

Well, not directly, but now thanks to crypto they can do ransomware attacks on the institutions you depend on and extort them for huge sums of money

I've been mostly remote since about 2013. I would have to be in extremely dire straits to ever go back to an office, especially since I've now moved back closer to my family and there's not really a tech industry to speak of around here. Overall no regrets, but I will say that I could definitely have made/be making more money if I'd stuck around the SF bay area and its drastically larger employer base. For me that's been fine, this career still pays better than I had any right to imagine I'd ever make, but it's definitely something to keep in mind--if you're not already in the salary ranges you want to be in, it is probably going to be harder to get there if you limit yourself to remote opportunities

Your problem is you're using regular human intuition while your employers operate under the logic of capitalism. Labor power is a commodity like any other, and if its supply curve shifts to the right as AI purports to do then it's not only possible but expected for price to go down as quantity sold increases.

IME big tech companies, although nowadays it seems like things are getting tighter everywhere

I've done, rough guess, somewhere in the neighborhood of 200-250 technical interviews at big tech companies. Not once has anyone ever reached out to me about a candidate before I interviewed them to try to influence me, nor have I ever heard of it happening. Obviously I can't say it's never happened in the history of ever, but this would be an absolutely insane thing to do at any company I've interviewed at, big tech or startup

Did they do it at Facebook in, like, 2008 or something? By the time I got onboarded for interviews (probably late 2012, early 2013-ish) that would have been considered wildly inappropriate, and if someone came at me trying to influence an interview for a friend I'd have asked them wtf they were thinking

A lot of these comments have me wondering if this is "go pound the pavement with you resume" style outdated advice, or if startups and smaller companies are really just their own world or what. I spent most of my career in big tech companies (Facebook, Snap, Stripe) and my experience has been that when I started in 2012 having a referral was pretty much a guarantee you'd get a phone screen, and that was it. Maybe I just don't know people with enough pull, but no one who's ever referred me for a position has seemed to make much of an impact on the hiring process, and likewise anyone I've referred has never seemed to have any kind of meaningful inside track beyond just getting past the auto-filters. And the longer I've been in the game, the value of a referral just seems to go lower and lower and lower. Around like 2015/2016, it was no guarantee my referrals would even get a phone screen. By the time COVID hit, I was seeing them get rejected without even getting a call from a recruiter

It's wild to me that they haven't special cased identifying this lowest hanging of fruit that people always reach for as soon as a new model comes out

The fact that your stick shrug guy is missing an arm due to markdown escaping is really just the cherry on top

I'm about as far as you can get from a gambler when it comes to these things. I tell people they should sell their equity when liquid so much that I should just get it mapped to a macro on my keyboard. But as far as I'm concerned, holding is the obvious choice here. It's $3600, I would rather watch the shares go to zero than let someone buy early engineer equity from me for the price of a gaming PC

Fwiw, I've been in the industry since 2012 and I've gotten every job but one, including the one I just started a few months ago, from submitting my regular ol human written resume to open listings. It's definitely possible

Have I ever left a job because the vibes were bad? No. *Should* I have? Oh yes, a couple of times

> This is why pretty much nothing really works anymore. This is a wild thing to write about services that handle unfathomable levels of traffic on a daily basis. Mature big tech companies are very good at letting the average application level engineer mostly ignore systems design concerns because they have infra teams that do such a good job of it that they can just give the rest of the company access to services and libraries that will scale up from a request per day to billions without blinking

This acronym has irritated the shit out of me from the beginning, because I started hearing it initially as a collection of high-growth big tech stocks to include in a portfolio--which, okay, fair enough--and then it seems like somewhere along the way it just got co-opted into employment discussions to mean "big, prestigious, high paying tech firms you should want to work for." And that never made sense! From the very beginning, there were a ton of companies off the list that were just as prestigious/high paying, arguably significantly more so, than some of the companies on the list. Because that's not what the acronym was for in the first place!

lmao, "if we didn't lie to our employees they might react rationally in their own best interests, so of course we can't tell them the truth."

Marginal tax brackets are exactly the point here. If you're in the 0% bracket, each marginal dollar you make increases your net income by $1...until you hit the next bracket. If you're in the 40% bracket, each marginal dollar only increases your net income by 60 cents.

Not that git is impossible to learn or anything, but a factor that gets underlooked IMO is complexity of a task *relative to how much you do it*. When it comes to something like git, I do some basic stuff all the time. Adding and committing files, pushing and pulling branches, rebasing, cherry picking, reverting. When you start getting into the more esoteric features, on the one hand sure, they're a lot less complicated than, say, learning a programming language. But they're also things that I very rarely need to do, so if I end up spending a couple hours in the git manual to figure out how to handle this one weird usecase, I'm probably gonna have to repeat those couple hours with the manual next time I need to do it because I'll have completely forgotten how to do it by then

It's coordinated in the same way everything in tech is defacto coordinated, everyone has the same investors who read the same blogs, and when a trend hits they all start telling all their portfolio companies to get on it. It's the same reason everyone started doing layoffs almost simultaneously whether they actually needed to or not. Also pretty much the reason softbank collapsed