Wow... I really relate to this. I'm 50 as well, and I started coding in 1985 when I was 10... I remember literally every evolutionary leap forward and my experience with this change has been a bit different.
Steve Yegge recently did an interview on vibe coding (https://www.youtube.com/watch?v=zuJyJP517Uw) where he says, "arch mage engineers who fell out-of-love with the modern complexity of shipping meaningful code are rediscovering the magic that got them involved as engineers in the first place" <-- paraphrased for brevity.
I vividly remember, staying up all night to hand-code assembler primitive rendering libraries, the first time I built a voxel rendering engine and thinking it was like magic what you could do on a 486... I remember the early days at Relic, working on Homeworld and thinking we were casting spells, not writing software. Honestly, that magic faded and died for me. I don't personally think there is magic in building a Docker container. Call me old-fashioned.
These days, I've never been more excited about engineering. The tedium of the background wiring is gone. I'm back to creating new, magical things - I'm up at 2 AM again, sitting at my desk in the dark, surrounded by the soft glow of monitors and casting spells again.
I think this works unironically. My mother is an avid gardener and can spend 8 hours a day gardening. When her life circumstances allowed for it, she hired a once a week gardener to do the tasks she didn't like (or had difficulties doing as a small woman), and still gardens the same amount. I've teased her for hiring a gardener, but she swears it's a huge help and boost to her gardening quality of life.
this is a great analogy despite it possibly coming off as snark.
I think it's hard for some people to grasp that programmers are motivated by different things. Some are motivated by shipping products to users, others are motivated to make code that's a giant elegant cathedral, still others love glorious hacks to bend the machine into doing things it was never really intended to do. And I'm sure I'm missing a few other categories.
I think the "AI ain't so bad" crowd are the ones who get the most satisfaction out of shipping product to users as quickly as possible, and that's totally fine. But I really wish they'd allow those of us who don't fall into that category to grieve just a little bit. This future isn't what I signed up for.
It's one thing to design a garden and admire the results, but some people get into their "zen happy place" by pulling up weeds.
I agree with this, I put myself in the "glorious hacks to bend the machine into doing things it was never really intended to do" camp, so the end game is somthing cool, now I can do 3 cool things before lunch instead of 3 cool things a year
*I'm so excited about landscape design. Can't wait to do more. Employing a gardener to do the gardening for me is really making me enjoy landscape design again!
The proper analogy would be you can now remove all weeds with the swipe of your hand and cut all your hedges with another swipe, you still are gardening you can do it quicker and therefore explore different possibilities.
Imagine though instead of 1 garden you can make 10 or 30 gardens in the same time that are more extravagant than your 1 garden was. At any point in time you can dive back in 1 of them and start plucking away
Well it's more like employing a gardener makes me enjoy landscaping again. It's not like we ever found writing words on a keyboard all that great, it's fundamentally just about having an idea and turning it into something real.
As mosburger says, this is a great analogy. Do you think that the great artists paint, sculpt, and draw everything by hand, by themselves? Of course not... they never did, and they don't today. You're being offered the ability to join their ranks.
It's your studio now. You have a staff of apprentices standing by, eager for instructions and commands. And you act like it's the worst thing that ever happened to you.
[55yo] My sense is that those problems we worked on in the 80s and 90s were like the perfectly balanced MMORPG. The challenges were tough, but with grit, could be overcome and you felt like you could build something amazing and unique. My voxel moment was passing parameters in my compilers class in college. I sat down to do it and about 12 hours later I got it working, not knowing if I could even do it.
With AI, it is like coding is on GOD mode and sure I can bang out anything I want, but so can anyone else and it just doesn't feel like an accomplishment.
>With AI, it is like coding is on GOD mode and sure I can bang out anything I want, but so can anyone else and it just doesn't feel like an accomplishment.
I think it's possible that we'll get to the point where "so can anyone else" becomes true, but it isn't today for most software. There's significant understanding required to ask for the right things and understand whether you're actually getting them.
That said, I think the accomplishment comes more so from the shaping of the idea. Even without the typing of code, I think that's where most of the interesting work lies. It's possible that AI develops "taste" such that it can sufficiently do this work, but I'm skeptical it happens in the near term.
I think there's still quite a chasm out there. Domain knowledge, an informed and opinionated view on how something should function, and overall tech knowledge are still key. Having those three things continues to greatly differentiate people of equal coding skill, as they always have.
> I don't personally think there is magic in building a Docker container. Call me old-fashioned.
This seems like a false dichotomy. You don't have to do this. It is still possible to build magical things. But agents aren't it, I don't think.
It is honestly extremely depressing to read this coming from a founder of Relic. Relic built magic. Dawn of War and Company of Heroes formed an important part of my teenage years. I formed connections, spent thousands of hours enjoying them together with other people, and pushed myself hard to become one of the top 100 players on the CoH leaderboards. Those competitive multiplayer games taught me everything there was to know about self-improvement, and formed the basis of my growth as an individual - learning that if I put my mind to it, I could be among the best at something, informed my worldview and led me to a life of perpetually pushing myself to further self-improvement, and from there I learned to code, draw, and play music. All of that while being part of amazing communities where I formed friendships that lasted decades.
All of this to say, Relic was magic. The work Relic did profoundly impacted my life. I wonder if you really believe your current role, "building trust infrastructure for AI agents", is actually magic? That it's going to profoundly impact the lives of thousands or millions?
I'm sorry for the jumbled nature of this post. I am on my phone, so I can't organize my thoughts as well as I would like. I am grateful to you for founding Relic, and this post probably comes off stupidly combative and ungrateful. But I would simply like to pose to you, to have a long think if what you're doing now is really where the magic is.
Edit: On further consideration, it's not clear the newly-created account I'm responding to is actually Alex Garden. The idea of potentially relating this personal anecdote to an impersonator is rather embarrassing, but I will nonetheless leave this up in the hope that if there are people who built magical things reading this, regardless of whether they're Alex Garden or someone else, that it might just inspire them to introspection about what building magic means, about the impact software can have on people's lives even if you don't see it, and whether this "agent" stuff is really it.
>The idea of potentially relating this personal anecdote to an impersonator is rather embarrassing
Good news! You've also related it to the roughly ~3-10M monthly HN readers who are not (potentially) impersonating the founder of a beloved game studio.
Also: I think you're probably safe. I'm sure someone at some point has come to HN to LARP as some prominent person in tech that they don't happen, at that specific moment, to actually be... but I can't really think of it happening before, nor would I expect it to take the form of a particularly thoughtful comment if a troll did that. Though with AI these days, who knows? I might myself just be one of a swarm of clawd/molt/claw things. In which case I'd be the last to even know it.
Oh-- as for being depressed about their docker/wiring things up sentiment. Try not to be, and instead, consider: Is it a surprise that someone who founded such a place as relic was occasionally-- even often-- frustrated at the things they had to clear away to build the thing they actually wanted to build? People who want to build amazing experiences may not love having to clear clutter that gets in their way. Other people want to build the tools that clear clutter, or other things that keep the whole system going. Those are beautiful too.
I started a bit younger and am a bit older, and relate. But only so much. I started programming in 3rd grade (also BASIC) when I found a computer and learned how to play a game on it, then found the source code for the game and randomly started changing it. In 7th grade I was paid to port a BASIC program to C (super new at the time), which I did on paper because I didn't own a computer (I used the money to buy my first). To be clear, I was really bad a programming for a long time and simply kept at it until I wasn't.
I love messing about with computers still. I can work at the byte level on ESP-32s on tiny little devices, and build massive computation engines at the time time on the same laptop. It's amazing.
I feel for those who have lost their love of this space, but I have to be honest: it's not the space that's the problem. Try something new, try something different and difficult or ungainly. Do what you rail against. Explore.
> I don't personally think there is magic in building a Docker container. Call me old-fashioned.
I still vividly remember setting up gcc in a docker container to cross compile custom firmware for my cannon camera and thinking about the amount of pain my local system would have been in if I had to do all the toolchain work in my host OS. Don't know if it felt like magic, but it sure didn't hurt like the alternative!
In the second half of my 40s now and I'm in the same boat. I started slapping keys on a c64 when I was 2 years old. Really enjoyed software development until 10-15 years ago. With the current LLM tooling available the number of systems I've been able to build that are novel and tackle significant problems has been kind of mind blowing over the past 8 months or so.
Staying up late, hacking away at stuff like I used to, and it's been a blast.
Finally, Homeworld was awesome and it felt magical playing it.
Wow, Alex Garden on Hackernews. Hello fellow canuck. I'm now getting up there, still a few years shy of y'all but not much. I came up through the 90s and early 2000s, all web/linux stuff, irc servers, bash scripts, python, weird php hacks, whatever, I was a kid. I'd lose track of time, It was Monday night after high school then all of a sudden it was Sunday morning and I was talking on irc about the crazy LAMP stack I'd put together. 2am? pfft, what is sleep?! Sadly with very strong dyslexia and dyscalculia, being a real programmer was never in the cards for me, I understood how everything worked, I can explain the whole thing end to end in great depth, but ask me predictably how to do a table in html or some fairly simple CSS, and I'll be there for hours. I'm grateful the rest of my life allowed me to be programmer adjacent and spend so much time around developers, but always a little frustrated I couldn't pick up the hammer myself.
These days, I've never been more excited about building. The frustration of being slow with the code is gone. I'm back to creating new, magical things - I'm up at 2 AM again, sitting at my desk in the dark, surrounded by the soft glow of monitors and casting spells.
Why is your last paragraph nearly identical to the last paragraph you are replying to? It might have been a strange quirk, but there’s also been the suggestion that the post you’re replying to is an imposter, so it gets weirder that you also did that.
I'm 45 yo. And also started programming quite early around 1988. In my case it was GWBAsic games and then C ModeX and A
Later Allegro based games.
Things got so boring in the last 15 years, I got some joy in doing AI research (ML, agents, Genetic Algorithms, etc).
But now, it's so cool how I can again think about something and build it so easily. I'm really excited of what I can do now. And im ot talking about the next billion dollar startup and whatnot. But the small hacky projects that LLMs made capable.yo build in no time.
My advice to everyone feeling existential vertigo over these tools is to remain confident and trust in yourself. If you were a smart dev before AI, chances are you will remain a smart dev with AI.
My experience so far is that to a first approximation, the quality of the code/software generated with AI corresponds to the quality of the developer using the AI tool surprisingly well. An inexperienced, bad dev will still generate a sub-par result while a great dev can produce great results.
The choices involved in using these tools are also not as binary as they are often made out to be, especially since agents have taken off. You can very much still decide to dedicate part of your day to chiseling away at important code to make it just right and make sure your brain is engaged in the result and exploring and growing with the problem at hand, while feeding background queues of agents with other tasks.
I would in fact say the biggest challenge of the AI tool revolution in terms of what to adapt to is just good ol' personal time management.
> If you were a smart dev before AI, chances are you will remain a smart dev with AI.
I don't think that's what people are upset about, or at least it's not for me. For me it's that writing code is really enjoyable, and delegating it to AI is hell on earth.
This is a part of it, but I also feel like a Luddite (the historical meaning, not the derogatory slang).
I do use these tools, clearly see their potential, and know full well where this is going: capital is devaluing labor. My skills will become worthless. Maybe GP is right that at first only skilled developers can wield them to full effect, but it's obviously not going to stop there.
If I could destroy these things - as the Luddites tried - I would do so, but that's obviously impossible.
For now I'm forced to use them to stay relevant, and simply hope I can hold on to some kind of employment long enough to retire (or switch careers).
The historical luddites are literally the human death drive externalized. Reject them and all of their garbage ideas with extreme prejudice.
Related, the word “meritocracy” was coined in a book which was extremely critical of the whole concept. AI thankfully destroys it. Good riddance, don’t let the door hit your ass on the way out.
You can reject the ideas in the aggregate. Regardless, for the individual, your skills are being devalued, and what used to be a reliable livelihood tied to a real craft is going to disappear within a decade or so. Best of luck
I resonate with that. I also find writing code super pleasurable. It's immediate stress relief for me, I love the focus and the flow. I end long hands-on coding sessions with a giddy high.
What I'm finding is that it's possible to integrate AI tools into your workflow in a big way without giving up on doing that, and I think there's a lot to say for a hybrid approach. The result of a fully-engaged brain (which still requires being right in there with the problem) using AI tools is better than the fully-hands-off way touted by some. Stay confident in your abilities and find your mix/work loop.
It's also possible to get a certain version of the rewards of coding from instrumenting AI tools. E.g. slicing up and sizing tasks to give to background agents that you can intuit from experience they'll be able to actually hand in a decent result on is similar to structuring/modularization exercises (e.g. with the goal to be readable or maintainable) in writing code, feelings-wise.
I really feel this. Claude is going to forget whatever correction I give it, unless I take the time and effort to codify it in the prompt.
And LLMs are going to continue to get better (though the curve feels like it's flattening), regardless of whatever I do to "mentor" my own session. There's no feeling that I'm contributing to the growth of an individual, or the state-of-the-art of the industry.
LLMs are similar in a lot of ways to the labor outsourcing that happened a generation or two ago. Except that instead of this development lifting a billion people out of poverty in the third world a handful of rich people will get even more rich and everyone else will have higher energy bills.
i agree. it seem like an expectation these days to use AI sometimes... for me i am happy not using it at all, i like to be able to say "I made this" :)
Yes, absolutely. I think the companies that don't understand software, don't value software and that think that all tech is fundamentally equivalent, and who will therefore always choose the cheaper option, and fire all their good people, will eventually fail.
And I think AI is in fact a great opportunity for good devs to produce good software much faster.
I think the issue is that given the speed the bad dev can generate sub-par results that at face value look good enough overwhelm any procedures in place.
Pair that with management telling us to go with AI to go as fast as possible means that there is very little time to do course correction.
I agree with the quality comments. The problem with AI coding isn't so much the slop, it's the developers not realizing its slop and trying to pass it off as a working product in code reviews. Some of the stuff I've reviewed in the past 6 months has been a real eye opener.
For me the problem is simple: we are in an active prisoner's dilemma with AI adoption where the outcome is worse collectively by not asking the right questions for optimal human results, we are defecting and using ai selfishly because we are rewarded by it. There's lots of potential for our use to be turned against us as we train these models for companies that have no commitment to give to the common good or return money to us or to common welfare if our jobs are disrupted and an AI replaces us fully.
What the author describes is also the feeling when you shift from being a developer all day to being a team lead or manager. When you become a lead you have to let go and get comfortable with the idea that the code is not going to be how you would do it. You can look at code produced by your team and attempt to replace it all with your craftsmanship but you're just setting yourself up to fail. The right approach is use your wisdom to make the team better, not the code. I think a lot of that applies to using AI when coding.
I'm turning 50 in April and am pretty excited about AI coding assistants. They make a lot of personal projects I've wanted to do but never had the time feasible.
Most of my career has been as an individual engineer, but the past few years I have been a project manager. I find this to be very much like using AI for coding.
Which also makes me refute the idea that AI coding is just another rung up on the programming abstraction ladder. Depending on how much you delegate to AI, I don't think it's really programming at all. It's project management. That's not a bad thing! But it's not really still programming.
Even just in the context of my human team, I feel less mentally engaged with the code. I don't know what everything does. (In principle, I could know, but I don't.) I see some code written in a way that differs from how I would have done it. But I'm not the one working day-in, day-out with the code. I'll ask questions, make suggestions, but I'm not going to force something unless I think it's really super important.
That said, I don't 100% like this. I enjoy programming. I enjoy computer science. I especially enjoy things more down the paths of algorithm design, Lisp, and the intersection of programming with mathematics. On my team, I do still do some programming. I could delegate it entirely, but I indulge myself and do a little bit.
I personally think that's a good path with AI too. I think we're at the point where, for many software application tasks, the programming could be entirely hands-off. Let AI do it all. But if I wish to, why not indulge in doing some myself also? Yeah, I know, I know, I'll get "left behind in the dust" and all of that. I'm not sure that I'm in that much of a hurry to churn out 50,000 lines of code a day; I'm cool with 45,100.
A lot of us resist the pressure to move to management or technical leadership for just these reasons. Programming people isn't the same as programming computers.
But the LLMs outnumber us. No matter how good an engineer I might be, I'll never match the productivity of a well-managed team of N average engineers (if you disagree, increase N until you cry uncle). Sure, there will be mythical man-month problems. But the optimal N is surely greater than 1, and I'll never be more than 1.
Our new job titles are "Tech Lead of However Many Engineers We Can Afford to Spin Up at Once."
It's fun managing a bunch of inexperienced juniors when there are no consequences (aka the infamous personal projects). It's a lot more stressful when it matters.
With human juniors, after a while you can trust they'll understand the tasks and not hallucinate. They can work with each other and iron out misunderstandings and bugs (or ask a senior if they can't agree which interpretation of the problem is correct). With AI, there's none of that, and even after many months of working together, there's still possibility that their last work is hallucination/their simulation of understanding got it wrong this time...
But the release of new models are generic. They don’t represent understanding in your specific codebase. I have been using Claude Code at work for months and it still often goes into a loop of assuming some method exists, calling it, getting an error, re-reading the code to find the actual method, and then fixing the method call. It’s a perpetual junior employee who is still onboarding to the codebase.
Yeah, I've experienced similar stuff. Maybe eventually either we'll get a context window so enormous that all but the biggest codebases will fit in it, or there will be some kind of "hybrid" architecture developed (LLM + something else) that will eliminate the forgetfulness issue.
It's also that when you move to being a leader, you suddenly have to learn to quantify and measure your productivity in a different way, which for a while can really do a number on your self-image.
What does it mean to be a productive developer in an AI tooling age? We don't quite know yet and it's also shifting all the time, so it becomes difficult to sort yourself into the range stably. For a lot of accomplished folks this is the first time they've felt that level of insecurity in a while, and it takes some getting used to.
> What the author describes is also the feeling when you shift from being a developer all day to being a team lead or manager.
I think that's very true. But... there's a reason I'm not a team lead or manager. I've done it in the past and I hate it. I enjoy doing the work, not tasking others with doing work.
I'm 60, started with a Tandy Model I in junior high, learned 6809 assembly for my Color Computer, loved the fact we could put certain values in particular memory positions and change the video mode and put pixels to the screen. It's been decades of losing that level of control, but for me coding is the fun part. I've never lost that spark of enjoyment and really obsession I felt early on. I enjoy the supposedly boring job of writing SQL and C with embedded SQL and working with business concepts to produce solutions. Coding is the fun part for me, even now.
I got moved up the chain to management and later worked to get myself moved back down to a dev role because I missed it and because I was running into the Peter Principle. I use AI to learn new concepts, but mostly as a search engine. I love the tech behind it, but I don't want it coding for me any more than I want it playing my video games for me. I was hoping AI would show up as robots doing my laundry, not doing the thing I most enjoy.
I am much younger than the author, but I've been coding for most of my life and I find close to no joy in using AIs. For me coding has always been about the nitty-gritty quirkiness of computers, languages, solving issues and writing new cool things for the sake of it. It was always more about the journey than the end goal, and AI basically hollows out all of the interesting bits about coding. It feels like skipping straight to the end of a book, or somewhat like that.
I don't know if I am the only one, but developing with chatbots in my experience turns developing software into something that feels more akin to filling out forms or answering to emails. I grieve for the day we'll lose what was once a passion of mine, but unfortunately that's how the world has always worked. We can only accept that times change, and we should follow them instead of complaining about it.
> For me coding has always been about the nitty-gritty quirkiness of computers, languages, solving issues and writing new cool things for the sake of it.
Same. It scratches my riddle-solving itch in a way that the process of "prompt-honing" has yet to do.
I'm lucky because I work as an independent consultant. I get paid to deliver solutions, but I get to choose how to create those solutions. I write whatever code I want however I want. As long as it solves the problem, no one cares.
I started programming in 1980, and I having just as much fun now as I did then. I literally cannot wait to sit down at my IDE and start writing.
But that was not always true. When I worked for a larger company, even some startups, it was not always fun. There's something about having full control over my environment that makes the work feel like play.
If you feel like programming isn't fun anymore, maybe switching to a consulting gig will help. It will give you the independence and control that you might be craving.
I have a hard time telling whether agentic coding tools will take a big bite out of the demand for software consultants. If the market is worried about SaaS because people think companies will use AI to code tools internally vs buying them, I would think the same would apply to consultants.
I’ve seen the code current tools produce if you’re not careful, or if you’re in a domain where training data is scarce. I could see a world where a couple of years from now companies need to bring outside people to fix vibe coded software that managed to gain traction. Hard to tell.
Not going to pull age or title rank here -- but I suggest if your use of AI feels empty, take advantage of its speed and plasticity and iterate upon its output more, shape the code results. Use it as a sculptor might too -- begin with its output and make the code your own. I particularly like this latter approach when I am tasked with use of a language I view as inferior and/or awkward. While this might read as idealistic, and I agree that there are situations where this interaction is infeasible or inappropriate, you should also be encountering problems where AI decidedly falls on its face and you need to intervene.
It seems AI is putting senior developers into two camps. Both groups relate to the statement, "I started programming when I was seven because a machine did exactly what I told it to, felt like something I could explore and ultimately know, and that felt like magic. I’m fifty now, and the magic is different, and I’m learning to sit with that."
The difference is that the first camp is re-experiencing that feeling of wonder while the second camp is lamenting it. I thankfully fall in the first camp. AI is allowing me to build things I couldn't, not due to a lack of skills, but a lack of time. Do you want to spend all your time building the app user interface, or do you want to focus on that core ability that makes your program unique? Most of us want the latter, but the former takes up so much time.
It sounds like you don’t particularly care about the user interface, and that’s why you’re okay with delegating it. I think the developers who don’t like delegating to AI are the ones who care about and have strong opinions about all the parts. To them there are no unimportant parts where the details don’t matter.
Similarly, I'm using it to write apps in non-native languages, like rust. My first foray into it led to finding poor documentation examples. AI allows me to create without spending large swaths of time learning minutia.
I'm enjoying it to a point, but yes, it does eliminate that sense of accomplishment - when you've spent many late nights working on something complex, and finally finish it. That's pretty much gone.
I have given the topic some thoughts. I concluded that the ONLY way for ordinary people (non-genius, IQ <= 120) to be really good, be really close to the genius, is to sit down, condensate the past 40 or so year's tech history of three topics (Comp-Arch, OS and Compiler) into a 4-5 years of self-education.
Such education is COMPLETELY different from the one they offered in school, but closer to those offered in premium schools (MIT/Berkeley). Basically, I'd call it "Software engineering archaeology". Students are supposed to take on ancient software, compile them, and figure out how to add new features.
For example, for the OS kernel branch:
- Course 0: MIT xv6 lab, then figure out which subsystem you are interested in (fs? scheduler? drivers?)
- Course 0.5: System programming for modern Linux and NT, mostly to get familiar with user space development and syscalls
- Course 1: Build Linux 0.95, run all of your toolchains in a docker container. Move it to 64-bit. Say you are interested in fs -- figure out the VFS code and write a couple of fs for it. Linux 0.95 only has Minix fs so there are a lot of simpler options to choose from.
- Course 2: Maybe build a modern Linux, like 5.9, and then do the same thing. This time the student is supposed to implement a much more sophiscated fs, maybe something from the SunOS or WinNT that was not there.
- Course 3 & 4: Do the same thing with leaked NT 3.5 and NT 4.0 kernel. It's just for personal use so I wouldn't worry about the lawyers.
For reading, there are a lot of books about Linux kernels and NT kernels.
Thank you for writing this. My feelings are very similar to the ones described by the author and the timeline almost matches. The thrill of tecnology for me started to fast decay since the early 2010s and now I see it as a no-return stage. I still have fun with my retro hardware & software but I am no longer an active practitioner and I have pivoted my attention and my efforts somewhere else. Unfortunately, I no longer feel excited for the future decades of tech and I am distancing myself from it.
I think this is something else, though. Even before AI really hit sweng, there were early signs of a collective tech depression a la "The best idea we can come up with is strapping screens to people's heads?", the "Are we the bad guys?" convo around social media, the crypto brain drain, etc. The queue of Next Big Things has increasingly felt more forced and controversial to many, and being in tech last lost much of its lustre to them.
I think it's healthy for everyone to evaluate whether one's personal reaction to AI is colored by this trend, or whether it's really being evaluated independently. Because while I share many of the negative feelings listed earlier, to me AI does still feel different; it has a lot more real utility.
What else do you do to make rent ? I feel the same way as you and I have no idea what else pays well for quality craftsmanship. I am staring at the abyss of hyper intelligent people with posh resumes and now wondering what to do.
it isn't all funeral marches and group crying sessions.
And don't let the blog post fool you , it is a rant about AI -- otherwise we would have heard complaints about the last 200 paradigm shifts in the industry over the past thirty years.
Sure, we got our share of dilbert-style agile/waterfall/tdd jokes shoved in our face, but no one wrote a blog post about how their identity was usurped by the waterfall model .
>And different in a way that challenges the identity I built around it and doesn’t satisfy in the way it did.
Everyone should do their own thing, but might I suggest that it is dangerous for anyone in this world to use a single pillar as their foundation for all identity and plinth of their character.
but no one wrote a blog post about how their identity was usurped by the waterfall model
I don’t know about that.
Waterfall mostly died before the rise of blogs, of course, but around the dawn of Agile I remember lots of posts about how nothing was properly designed any more, nothing was ever finished, and you never knew what the specification was.
They used to be real engineers, but now it was just all chaos! They couldn’t design anything any more!
Thanks for reminding me of the word plinth. I agree with the author that the job is less fun now, less interesting. I'm doing and accomplishing more, and it matters less. And unfortunately, having other ways of defining your identity doesn't really help, for me. What it does is make those other aspects of myself relatively more attractive as careers, in comparison to this one. Although then again, I suppose it's helping in the way you intend: I could leave (and I might), I could adapt. So I'm feeling none of the fear or anxiety about AI. Just something that I think is roughly boredom.
Oh my god. This is me. If I were any better at writing, I could have written this, the author is even the same age as me (well, a year younger) and followed a similar trajectory. And a lot of what I've been feeling lately feels similar to burnout (in fact I've been calling it that), but it really isn't burnout. It's... this, whatever this is... a "fallow period" is a good term.
And I feel like an old man grumbling about things changing, but... it's not the same. I started programming in BASIC on my Tandy 1000 and went to college and learned how to build ISA cards with handwritten oscilloscope software in the Computer Engineering lab. My first job was writing firmware. I've climbed so far up the abstraction chain over a thirty year career and I guess I don't feel the same energy from writing software that first got me into this, and it's getting harder to force myself to press on.
At my first full time job in the early 2000s I was tasked with building a webscraper. We worked for law firms representing Fortune 500 companies and they wanted to know who was running "pump and dump" stock schemes on stocks using Yahoo Finance message boards.
At the time, I didn't know the LWP::Simple module existed in Perl so I ended up writing my own socket based HTTP library to pull down the posts, store them in a database etc. I loved that project as it taught me a lot about HTTP, networking, HTML, parsing and regexes.
Nowadays, I use playwright to scrape websites for thing I care about (e.g. rental prices at the Jersey Shore etc). I would never think to re-do my old HTTP library today while still loving the speed of modern automation tools.
Now, I too have felt the "but I loved coding!" sense of loss. I temper that with the above story that we will probably love what comes next too (eventually).
> I started programming when I was seven because a machine did exactly what I told it to, felt like something I could explore and ultimately know, and that felt like magic. I’m fifty now, and the magic is different, and I’m learning to sit with that.
Don't take this the wrong way but this is more of an age thing rather than a technology advancement thing.
Kids growing up nowadays that are interested in computers grow up feeling the same magic. That magic is partly derived from not truly understanding the thing you are doing and creating a mental "map" by yourself. There is nothing intrinsic to computing nowadays that makes it less magic than fiddling around with config.sys, in 50 years there will be old programmers reminiscing of "Remember when all new models were coming out every few months and we could fiddle around with the vector dimensionality and chunking length to get the best of gpt-6.2 RAG? Those were the times".
> There is nothing intrinsic to computing nowadays that makes it less magic than fiddling around with config.sys
There definitely is: the rent-seeking behavior is out of control. As a kid I could fiddle with config.sys (or rather autoexec.bat) while nowadays wrestling a file path out of my phone is a battle and the system files of my phone are kept from me.
config.sys was understandable. Now your computer has thousands (probably more) of config.sys-sized components and you are still only one person. The classic UI may improve your ability to find the components (sometimes) but can't reduce the complexity of either the components themselves or their quantity. AI makes it possible to deal with this complexity in a functional way.
Your last point is probably correct though, because AI will also allow systems to become orders of magnitude more complex still. So like the early days of the internet, these are still the fun days of AI, when the tool is overpowered compared to its uses.
LLM are not AI, but are a great context search tool when they work.
When people first contact ML, they fool themselves into believing it is intelligent... rather than a massive plagiarism and copyright IP theft machine.
Fun is important, but people thinking zero workmanship generated content is sustainable are still in the self-delusion stage marketers promote.
I am not going to cite how many fads I've seen cycle in popularity, but many have seen the current active cons before. A firm that takes a dollar to make a dime in revenue is by definition a con kids. =3
The contrast between this and https://news.ycombinator.com/item?id=46923543 (Software engineering is back) is kind of stark. I am using frontier models to get fun technical projects done that I simply didn't have time for since my late teens. It is still possible to understand an architecture down to the hardware if you want to, but it can happen a lot faster. The specifications are queryable now. Obscure bugs that at least one person has seen in the past are seconds away instead of minutes or hours of searching. Even new bugs have extra eyes on them. I haven't written a new operating system yet but it's now a tractable problem. So is using Lean or Julia or some similar system to formally specify it. So far I've been digging into modern multithreaded cache performance which is just as fascinating as directly programming VGA and sound was in the early PC days. Linux From Scratch is still up to date. You can get FPGAs that fit in your USB port [0]. Technical depth and low-level understanding is wherever you want to look for it.
I was 7 in 1987, learned LOGO and C64 BASIC that year, and I relate to this article as well.
It feels as though a window is closing upon the feeling that software can be a powerful voice for the true needs of humanity. Those of us who can sense the deepest problems and implications well in advance are already rare. We are no more immune to the atrophy of forgetting than anyone.
But there is a third option beyond embrace or self-extinguish. The author even uses the word, implying that consumers wanted computers to be nothing more than an appliance.
The third option is to follow in the steps of fiction, the Butlerians of Dune, to transform general computation into bounded execution. We can go back to the metal and create a new kind of computer; one that does have a kind of permanence.
From that foundation, we can build a new kind of software, one that forces users to treat the machine as appliance.
It has never been done. Maybe it won't even work. But, I need to know. It feels meaningful and it has me writing my first compiler after 39 years of software development. It feels like fighting back.
Was this text run through LLM before posting? I recognize that writing style honestly; or did we simply speak to machines enough to now speak like machines?
Yes. This is absolutely chatgpt-speak. I see it everywhere now. It's inescapable.
At least this appears to be largely human authored and have some substance, which is generally not the case when I see these LLM-isms.
I don't disagree that technology is less fun in an AI era. The question is, what other careers are out there for someone who wants to make things?
About a decade ago, I went through a career crisis where I couldn't decide what job to do - whether technology was really the best choice for my particular temperament and skills.
Law? Too cutthroat. Civil service? Very bureaucratic. Academia? Bad pay. Journalism? An industry in decline.
It is a shame, what is happening. But I still think, even with AI hollowing out the fun parts, tech remains the best job for a smart, motivated person who's willing to learn new things.
Fact is, the tech sector is filled with folks that find zero joy in what they do, chose a career for financial reasons, and end up being miserable to everyone including themselves.
The ex-service people would call these folks entitled Shitbirds, as no matter the situation some will complain about everything. Note, everyone still does well in most large corporate settings, but some are exhausting to be around on a project. =3
The reason we don’t have the right to be lazy is because of the people who find “meaning” in toil. I do not want to work and AI is the most anti work technology in human history.
Bertrand Russel literally wrote a book called “in defense of idleness” because he knew that heavy hitters like him had to defend work abolitionism. The “work is good” crowd is why we can’t have nice things. You guys are time thief’s and ontologically evil. May all work supporters reincarnate as either durian fruits or cockroaches.
You seem very passionate about your opinions, but are you happy?
The fact remains LLM can't reach comparable human error rates without consuming 75% of the energy output of our entire local galaxy.
While I find true Neuromorphic computing topics more interesting, the emergence of the LLM "AI" true believer is deeply concerning to those that understand how they are actually built. =3
I'm the exact age as the author and this post could have been written by me (if I could write). It echoes my story and sentiment exactly right down to cutting my literal baby teeth on a rubber key ZX Spectrum.
The anxiety I have that the author might not be explicitly stating is that as we look for places we add genuine value in the crevices of frontier models' shortcomings those crevices are getting more narrow by the day and a bit harder to find.
Just last night I worked with Claude and at the end of the evening I had it explain to me what we actually did. It was a "Her" (as in the movie) moment for me where the AI was now handholding me and not the other way around.
> The anxiety I have that the author might not be explicitly stating is that as we look for places we add genuine value in the crevices of frontier models' shortcomings those crevices are getting more narrow by the day and a bit harder to find.
That's exactly it. And then people say "pivot to planning / overall logic / high-level design," but how long do we have before upper management decides that AI is good enough at that stuff, too, and shows us all the door?
If they believe they can get a product that's 95% of what an experienced engineer would give them for 5% of the cost, why bother keeping the engineer around?
A blacksmith was a person that picked up chunks of carbon and heated them to they were glowing red and beat the iron to submission with a hammer in their hands.
Today iron is produced by machines in factories by the mega-tonne.
We just happened to live in the age where code when from being beaten by hand to a mass produced product.
Especially anyone in their 40s or 50s who is close enough to retirement that a career shift is unappealing but far enough from retirement that a layoff now would meaningfully change that timeline or QOL. I don't blame people for feeling uneasy.
I'm probably 7 or 8 years from an easy retirement myself, so I can appreciate how that feels. Nobody really wants to feel disruption at this age, especially when they're the breadwinner for a family.
No. By this logic, if they wanted to stay with the times they should have sought capital investment for their own industrial forges, joined their local lodges, climbed the ranks, lobbied their governments for loose safety regulations, and plied their workers with propaganda about how "we're in a recession and have to tighten our belts".
Think of the wonderful world we could have if everyone just got their shit together and became paper trillionaire technocrats.
Some of them feel bad about it and some of them refined metallurgy to build Saturn V rockets and go to space. We are very much living in the new space race. The discussion here is split 50/50 between the “Thank you! I feel the same way” folks and the “I am having the time of my life!” folks.
I'm the exact same demographic as the author, just turned 50, writing code since childhood in BASIC. I'm dealing with the AI in programming issue by ignoring it.
I still enjoy the physical act of programming so I'm unsure why I should do anything that changes that. To me it's akin to asking a painter to become a photographer. Both are artists but the craft is different.
Even if the AI thing is here to stay, I think there will be room for people who program by hand for the same reason there's still room for people who paint, despite the invention of the camera.
But then, I'm somebody who doesn't even use an IDE. If I find an IDE obtrusive then I'm certain I'll find an AI agent even more so.
50 myself, and started coding with a Commodore 64, but only really picked it up seriously with the advent of open source software, and that feeling of being able to dig around any component of the system I wanted to was exhilarating.
I think that's one of the biggest things that gives me pause about AI: the fact that, if they prove to be a big productivity boost, you're beholden to huge corporations, and not just for a one-time purchase, but on an ongoing basis.
Maybe the open source models will improve, but if keeps being driven by raw compute power and big numbers, it seems to tilt things very much in favor of those with lots and lots of capital to deploy.
There's nothing "hollowed out" about directing an AI effectively, the feedback is as quick and tight as it always was. The trick is that you don't just "vibe code" and let the AI one-shot the whole thing: you should propose the change first and ask the AI about a good, detailed plan for implementing it. Then you review what the robot has proposed (which is trivial compared to revising code!) make sensible changes, ask for feedback again, and repeat. By the time the AI bot has to write actual code, it's not running on vibes anymore: it's been told exactly what to do and how to assess the result. You spend more time upfront, but a lot less on fixing the AI's mistakes.
> you should propose the change first and ask the AI about a good, detailed plan for implementing
Why ask though?
If I’m familiar with a project, more often than not, I usually have a very good idea of the code I have to write within minutes of reading the ticket. Most of the time taken is finding the impact of the change, especially with dependencies that are present in the business domain, but are not reflected in the code.
I don’t need to ask what to code. I can deduce it as easily as doing 2+2. What I’m seeking is a reason not to write it the way I envisioned it. And if those reasons are technical, it’s not often a matter of code.
Because that's how you ensure that the AI has the right idea about what to do. If the proposed plan has problems, you work with the AI to fix them before setting it to work. AI is not as smart as you, so it needs to be told how to go about doing things.
Any change that I’ve done which resulted in more than a a 10 lines diff are done with tools (copy-paste, vim-fu, refactor tools or script, snippets, code generators,…) Why would I spend time babysitting an LLM when I could have just done it myself? The purpose of automation is to lighten my workload, not to add to it.
An LLM is also a code generator. There is a scale of changes where using one is just not worthwhile (quite possibly around the 10 lines mark, as you said) but other than that, why would you want to write code yourself line-by-line that you could just generate?
If vendors can't be bothered to use a C compiler from the last decade, I don't think they'll be adopting AI anytime soon.
At my work, as of 2026, we only now have a faction riled up about evangelizing clean code, OOP, and C++ design patterns. I hope the same delay keeps for all the rest of the "abstraction tower".
The issue is that AI will be creating software at whatever abstraction layer it is asked to produce. Right down to ASM maybe even machine code if someone actually wanted or needed that. Perhaps not the AI of today but given a few years I'll be quite surprised if it still can't.
If we can take a computer as powerful as today’s laptops and make it crawl because of the amount of inefficiencies in software like Teams, I’m not holding breath for embedded. If you apply the same kind of engineering principle as Anthropic, you’ll be laughed out of the room.
I too have felt these feelings (though I'm much younger than the author). I think as I've grown older I have to remind myself
1. I shouldn't be so tied to what other people think of me (craftsman, programmer, low level developer)
2. I shouldn't measure my satisfaction by comparing my work to others'. Quality still matters especially in shared systems, but my responsibility is to the standards I choose to hold, not to whether others meet them. Plus there are still community of people that still care about this (handmade network, openbsd devs, languages like Odin) that I can be part of it I want to
3. If my values are not being met either in my work or personal life I need to take ownership of that myself. The magic is still there, I just have to go looking for it
I found that feeling again while building a game on the EVM. All of the constraints were new and different. Solidity feels somewhere between and high and low level language, not as abstracted as most popular languages today but a solid step above writing assembly.
A lot of people started building projects like mine when the EVM was newer. Some managed to get a little bit of popularity, like Dark Forest. But most were never noticed. The crypto scene has distracted everyone from the work of tinkerers and artists who just wanted to play with a new paradigm. The whole thing became increasingly toxic.
It was like one last breath of fresh cool air before the pollution of AI tools arrived on the scene. It's a bitter sweet feeling.
I humbly submit this interview with Grady Booch (if you know, you know) talking about the "3rd golden age of software engineering - thanks to AI": https://youtu.be/OfMAtaocvJw
I feel like the conversation does a good job of couching the situation we find ourselves in.
I think one of the big distinctions between people who like building with AI and those who don't, is that the people who are pro-AI are building their own ideas, of which they have many.
The people who are anti-AI are largely building other people's ideas, for work. And they have no desire to ramp up velocity, and it's not helpful to them anyway because of bureaucratic processes that are the real bottleneck to what they're building.
I am a little older than OP. I don't think I've ever had that feeling about a programming project for work that came from someone else.
Generally, I get that feeling from work projects that I've self-initiated to solve a problem. Fortunately, I get the chance to do this a lot. With the advent of agentic coding, I am able to solve problems at a much higher rate.
Quite often, I'll still "raw dog" a solution without AI (except for doc lookups) for fun, kind of as a way to prove to myself I can still do it when the power's out.
I'm ~40ish but middle career and not in management. I envy this author, whatever joy he found in solving little puzzles and systems was extinguished in me very early in my career in an intense corporate environment. I was never one to love fussing much with code, but I do love solving system scale problems, which also involve code. I don't feel I am losing anything, the most annoying parts of code I deal with are now abstracted into human language and specs, and I can now architect/build more creatively than before. So I am happy. But, I was one of those types that never had a true passion for "code" and have meant plenty of people that do have that, and I feel for them. I worry for people that carved out being really good at programming as a niche, but you enter a point in your career where that becomes much less important than being able to execute and define requirements and understand business logic. And yea, that isn't very romantic or magical, but I find passion outside of what pays my bills, so I lost that ennui feeling a while ago.
"Over four decades I’ve been through more technology transitions than I can count. New languages, new platforms, new paradigms. CLI to GUI. Desktop to web. Web to mobile. Monoliths to microservices. Tapes, floppy discs, hard drives, SSDs. JavaScript frameworks arriving and dying like mayflies."... made me think of
I've seen things you people wouldn't believe. Attack ships on fire off the shoulder of Orion. I watched C-beams glitter in the dark near the Tannhäuser Gate. All those moments will be lost in time, like tears in rain. Time to die.
where we came from and where we're going this whole time in my career those things are kind of hard to pinpoint. Abstraction is killing us for sure. Time to market above all else. It's no wonder why software in cars, appliances and medical equipment is a factor that is killing people.
maybe we just change, honestly. i think when i were younger there was nothing to lose, time felt unlimited, no "career" to gamble with, no billion dollar idea, just learning and tinkering and playing with whatever was out there because it was cool and interesting to me. in some respects i miss that.
not sure how that relates to llms but it does become an unblocker to regain some of that "magic", but also i know to deep dive requires an investment i cannot shortcut.
the new generation of devs are already playing with things few dinosaurs will get to experience fully, having sunk decades into the systems built and afraid to let it go. some of that is good (to lean on experience) and some of it holding us back.
I prefer to see it as the automtion of the IT age.
All other professions had their time when technology came and automated things.
For example wood carvers, blacksmiths, butchers, bakers, candlestickmakers etc etc. All of those professions have been mostly taken over by machines in factories.
I view 'ai' as new machines in factories for producing code. We have reached the point where we have code factories which can produce things much more efficiently and quicker than any human can alone.
Where the professions still thrive is in the artisan market. There is always demand for hand crafted things which have been created with love and care.
I am hoping this stays true for my coding analogy. Then people who really care about making a good product will still have a market from customers who want something different from the mass produced norm.
I think the issue at the core of the analogy is that factories, traditional factories, excel at making a ton of one thing (or small variations thereof). The big productivity gains came from highly reliable, repeatable processes that do not accommodate substantial variation. This rigidity of factory production is what drives the existence of artisan work: it can always easily distinguish itself from the mass product.
This does not seem true for AI writing software. It's neither reliable nor rigid.
What assembly lines and factories did for other manufacturing processes is to make it feasable for any person to be able to make those things. In the past only very skilled professionals were able to create such things, but mechanisation and breaking down manufacturing processes into small chunks made the same things be able to be achieved by low skilled workers.
IMO that is exactly what is happening here. Ai is making coding apps possible for the normal person. Yes they will need to be supervised and monitored, just like workers in a factory. But groups of normal low skilled workers will be able to create large pieces of software via ai, whic has only ever been possible by skilled teams of professinoals before.
Yes, I think that's how it will go, like all those other industries. There will be an artisanal market, that's much smaller, where the (fewer) participants charge higher prices. So it'll (ironically?) end up being just another wealth concentrator. A few get richer doing artisanal work while most have their wage depressed and/or leave the market.
> For example wood carvers, blacksmiths, butchers, bakers, candlestickmakers etc etc.
Very, very few of those professions are thriving. Especially if we are talking true craftsmanship and not stuffing the oven with frozen pastries to create the smell and the corresponding illusion of artisinal work.
They are thriving where I live. There is a huge artisinal market for hand crafted things. There are many markets, craft centers, art fairs, regular classes from professionals teaching amateurs etc. In most rural communities I have visited it is similar.
> The feedback loop has changed. The intimacy has gone. The thing that kept me up at night for decades — the puzzle, the chase, the moment where you finally understand why something isn’t working — that’s been compressed into a prompt and a response
It's so strange to read because to me its never been more fun to make software, its especially never been easier for an individual. The boring parts are being automated so I can work on the bespoke and artistic parts. The feedback loop is getting shorter to making something nice and workable. The investigation tools for profiling and pinpointing performance bottlenecks are better than ever, where Claude is just one new part of it.
I turn 52 this year. I also started at 10 years old programming in a combination of AppleSoft BASIC and assembly language and typing machine code out of books so I could use Double Hires graphics since it wasn’t supported by BASIc and doing my own assembly language programming.
I stuck with C and C++ as my bread and butter from 1996-2011 with other languages in between.
I don’t miss “coding” because of AI. My vision has been larger than what I could do myself without delegating for over a decade - before LLMs.
“coding” and/or later coordinating with people (dotted line) reporting to me has been a necessary evil until a year or two ago to see my vision go to implementation.
I absolutely love this new world. For loops and while loops and if statements don’t excite me in my 50s. Seeing my vision come to life faster than I ever could before and having it well archited does.
I love talking to “the business” and solving XYProblems and getting to a solution 3x faster
I am in a very similar boat, age and experience-wise. I would like to work backward from the observation that there is no resource constraints and we're collectively hopelessly lost up the abstraction Jenga tower.
I observe that the way we taught math was not oriented on the idea that everyone would need to know trigonometric functions or how to do derivatives. I like to believe math curricula was centered around standardizing a system of thinking about maths and those of us who were serious about our educational development would all speak the same language. It was about learning a language and laying down processes that everyone else could understand. And that shaped us, and it's foolish to challenge or complain about that or, God forbid, radically change the way we teach math subjects because it damages our ability to think alike. (I know the above is probably completely idealistic verging on personal myth, but that's how I choose to look at it.)
In my opinion, we never approached software engineering the same way. We were so focused on the compiler and the type calculus, and we never taught people about what makes code valuable and robust. If I had FU money to burn today, I'd start a Mathnasium company focused around making kids into systems integrators with great soft skills and the ability to produce high quality software. I would pitch this business under the assumption that the jenga tower is going to be collapsing pretty much continuously for the next 25-50 years and civilization needs absolute unit super developers coming out of nowhere who will be able to make a small fortune helping companies dig their way out of 75 years of tech debt.
You can still have fun programming. Just sit down and write some code. Ain't nobody holding a gun to your head forcing you to use AI in your projects.
And the part of programming that wasn't your projects, whether back in the days of TPS reports and test coverage meetings, or in the age of generative AI, that bit was always kinda soul draining.
Starting code when I was 14, sold my first bit of code at 17, which was written in 6502 assembler.
40+ years later, been through many BASICs, C, C++ (CFront on onwards) and now NodeJS, and I still love writing code.
Tinkering with RPi, getting used to having a coding assistant, looking forward to having some time to work on other fun projects and getting back into C++ sooooon.
Well-written and it expresses a mood, a feeling, a sense of both loss and awe. I was there too in the 8-bit era, fully understanding every byte of RAM and ROM.
The sense of nostalgia that can turn too easily into a lament is powerful and real. But for me this all came well before AI had become all consuming... It's the just the latest manifestation of the process. I knew I didn't really understand computers anymore, not in the way I used to. I still love coding and building but it's no longer central to my job or lif3. It's useful, I enjoy it but at the same time I also marvel at the future that I find myself living in. I've done things with AI that I wouldn't have dared to start for lack of time. It's amazing and transformative and I love that too.
But I will always miss the Olden Days. I think more than anything it's the nostalgia for the 8-bit era that made me enjoy Stranger Things so much. :)
Fantastic Article, well written, thoughtful. Here are a couple of my favorite quotes:
* "Then it professionalised. Plug and Play arrived. Windows abstracted everything. The Wild West closed. Computers stopped being fascinating, cantankerous machines that demanded respect and understanding, and became appliances. The craft became invisible."
* "The machines I fell in love with became instruments of surveillance and extraction. The platforms that promised to connect us were really built to monetise us. The tinkerer spirit didn’t die of natural causes — it was bought out and put to work optimising ad clicks."
* "Previous technology shifts were “learn the new thing, apply existing skills.” AI isn’t that. It’s not a new platform or a new language or a new paradigm. It’s a shift in what it means to be good at this."
* "They’re writing TypeScript that compiles to JavaScript that runs in a V8 engine written in C++ that’s making system calls to an OS kernel that’s scheduling threads across cores they’ve never thought about, hitting RAM through a memory controller with caching layers they couldn’t diagram, all while npm pulls in 400 packages they’ve never read a line of... But sure. AI is the moment they lost track of what’s happening."
* "Typing was never the hard part."
* "I don’t have a neat conclusion. I’m not going to tell you that experienced developers just need to “push themselves up the stack” or “embrace the tools” or “focus on what AI can’t do.” All of that is probably right, and none of it addresses the feeling."
To relate to the author, I think with a lot of whats going on I feel the same about, but other parts I feel differently than they do. There appears to be a shallowness with this... yes we can build faster than ever, but so much of what we are building we should really be asking ourselves why do we have to build this at all? Its like sitting through the meeting that could have been an email, or using hand tools for 3 hours because the power tool purchase/rental is just obscenely expensive for the ~20min you need it.
I think the true genuinely-love-programming type of people will increasingly have to do what so many other people do, and that's separation of work and personal enjoyment. You might have to AI-architect your code at work, and hand code your toy projects on the weekend.
I'm roughly the same (started at 9, currently 48), but programming hasn't really changed for me. What's changed is me having to have pointless arguments with people who obviously have no clue what they're talking about but feel qualified either because:
a) They asked an LLM
b) "This is what all our competitors are doing"
c) They saw a video on Youtube by some big influencer
d) [...insert any other absurd reason...]
True story:
In one of our recent Enterprise Architecture meetings, I was lamenting the lack of a plan to deal with our massive tech debt, and used an example of a 5000 line regulatory reporting stored procedure written 10 years ago that noone understood. I was told my complaint was irrelevant because I could just dump it into ChatGPT and it would explain it to me. These are words uttered by a so-called Senior Developer, in an Enterprise Architecture meeting.
Was he entirely wrong? Have you tried to dump the stored proc into a frontier model and ask it to refactor? You'd probably have neat 20 stored procs with well laid out logic in minutes.
I wouldn't keep a ball of mud just because LLMs can usually make sense of them but to refactor such code debt is becoming increasingly trivial.
Yes. I mean... of course he was?. Firstly, I had already gone through this process with multiple LLMs, from various perspectives, including using Deep Research models to find out if any other businesses faced similar issues, and/or if products existed that could help with this. That lead me down a rabbit hole of data science products related to regulatory reporting of a completely different nature which was effectively useless. tl;dr: Virtually all LLMs - after understanding the context - recommended us doing thing we had already been urging the business to do - hire a Technical BA with experience in this field. And yes, that's what we ended up doing.
Now, give you some ideas about why his idea was obviously absurd:
- He had never seen the SP
- He didn't understand anything about regulatory reporting
- He didn't understand anything about financial derivatives
- He didn't understand the difference between Transact SQL and ANSI SQL
- No consideration given to IP
- etc etc
Those are the basics. Let's jump a little bit into the detail. Here's a rough snippet of what the SP looks like:
SELECT
CASE
WHEN t.FLD4_TXT IN ('CCS', 'CAC', 'DEBT', ..... 'ZBBR') THEN '37772BCA2221'
WHEN t.FLD4_TXT IN ('STCB') AND ISNULL(s.FLD5_TXT, s.FLD1_TXT) = 'X' THEN 'EUMKRT090011'
END as [Id When CounterParty Has No Valid LEI in Region]
-- remember, this is around 5000 lines long ....
Yes, that's a typical column name that has rotted over time, so noone even knows if it's still correct. Yes, those are typical CASE statements (170+ of them at last count, and no, they are not all equal or symmetric).
So... you're not just dealing with incredibly unwieldy and non-standard SQL (omitted), noone really understands the business rules either.
So again... yes he was entirely wrong. There is nothing "trivial" about refactoring things that noone understands.
I'm a few years behind you. I got started on my uncle's handed down vic 20 in the late 80s.
The culture change in tech has been the toughest part for me. I miss the combination of curiosity, optimism, creativity, and even the chaos that came with it. Nowadays it's much harder to find organizations like that.
> I wrote my first line of code in 1983. I was seven years old, typing BASIC into a machine that had less processing power than the chip in your washing machine
I think there may be a counterpoint hiding in plain sight here: back in 1983 the washing machine didn't have a chip in it. Now there are more low-level embedded CPUs and microcontrollers to develop for than before, but maybe it's all the same now. Unfathomable levels of abstraction, uniformly applied by language models?
Yeah I could use Cursor or whatever but I don't, I like writing code. I guess that makes me a luddite or something, although I still develop agents. I enjoy architecting things (I don't consider myself an architect) I'm talking about my hobby hardware projects.
I'm 55 and I started at age 13 on a TI-99/4A, then progressed through Commodore 64, Amiga 2000, an Amiga XT Sidecar, then a real XT, and on and on. DOS, Windows, Unix, the first Linux. I ran a tiny BBS and felt so excited when I heard the modem singing from someone dialing in. The first time I "logged into the Internet" was to a Linux prompt. Gopher was still a bigger thing than the nascent World-Wide Web.
The author is right. The magic has faded. It's sad. I'm still excited about what's possible, but it'll never create that same sense of awe, that knowledge that you can own the entire system from the power coming from the wall to the pixels on your screen.
Similar story for myself. It was long and tedious for my mental model to go from Basic, to Pascal, to C, and finally to ASM as a teen.
My recent experience is the opposite. With LLMs, I'm able to delve into the deepest parts of code and systems I never had time to learn. LLMs will get you to the 80% pretty quick - compiles and sometimes even runs.
I know exactly how you feel. I don't know how many hours I sat in front of this debugger (https://www.jasik.com) poking around and trying to learn everything at a lower level. Now its so different.
Yeah. Different is the word. In many ways it’s just another abstraction but we’re not machines and this, to me at least, just gives a very different feel.
Same, but it changed when I was 17 and again when I was 27 and then 37 and so on. It has always been changing dramatically, but this latest leap is just so incredibly different that it seems unique.
Abstractions can take away but many add tremendous value.
For example, the author has coded for their entire career on silicon-based CPUs but never had to deal with the shittiness of wire-wrapped memory, where a bit-flip might happen in one place because of a manufacturing defect and good luck tracking that down. Ever since lithography and CPU packaging, the CPU is protected from the elements and its thermal limits are well known and computed ahead of time and those limits baked into thermal management so it doesn’t melt but still goes as fast as we understand to be possible for its size, and we make billions of these every day and have done for over 50 years.
Moving up the stack you can move your mouse “just so” and click, no need to bit-twiddle the USB port (and we can talk about USB negotiation or many other things that happen on the way) and your click gets translated into an action and you can do this hundreds of times a day without disturbing your flow.
Or javascript jit compilation, where the js engine watches code run and emits faster versions of it that make assumptions about types of variables - with escape hatches if the code stops behaving predictably so you don’t get confusing bugs that only happen if the browser jitted some code. Python has something similar. Thanks to these jit engines you can write ergonomic code that in the typical scenario is fast enough for your users and gets faster with each new language release, with no code changes.
Lets talk about the decades of research that went into autoregressive transformer models, instruction tuning, and RLHF, and then chat harnesses. Type to a model and get a response back, because behind the scenes your message is prefixed with “User: “, triggering latent capabilities in the model to hold its end of a conversation. Scale that up and call it a “low key research preview” and you have ChatGPT. Wildly simple idea, massive implications.
These abstractions take you further from the machine and yet despite that they were adopted en masse. You have to account for the ruthless competition out there - each one would’ve been eliminated if they hadn’t proven to be worth something.
You’ll never understand the whole machine so just work at the level you’re comfortable with and peer behind the curtain if and when you need (eg. when optimizing or debugging).
I'm 43. Took a year or so off from contracting after being flat out for years without taking any breaks, just poked around with some personal projects, did some stuff for my wife's company, petitioned the NHS to fix some stuff. Used Claude Code for much of it. Travelled a bit too.
I feel like I turned around and there seem to be no jobs now (500+ applications deep is a lot when you've always been given the first role you'd applied to) unless you have 2+ years commercial AI experience, which I don't, or perhaps want to sit in a SOC, which I don't. It's like a whole industry just disappeared while I had my back turned.
I looked at Java in Google Trends the other day, it doesn't feel like it was that long ago that people were bemoaning how abstracted that was, but it was everywhere. It doesn't seem to be anymore. I've tried telling myself that maybe it's because people are using LLMs to code, so it's not being searched for, but I think the game's probably up, we're in a different era now.
Not sure what I'm going to do for the next 20 years. I'm looking at getting a motorbike licence just to keep busy, but that won't pay the bills.
I’m 45 and contracted for over a decade before switching to product development. I used to still get inquiries from former customers, mainly for Java and Android work. But since about two years, it’s completely dried up. Anecdotally I’ve been hearing from friends who are still in the contracting/freelancing business that things are very tough right now. It makes sense to me, contractors are usually the first thing businesses cut when they’re either lowering their spending or becoming more efficient themselves.
I don't know what these people from our now traditional daily lamentation session are coding where Claude can do all the work for them just with a few prompts and minimal reviews.
Claude is a godsend to me, but fuck, it is sometimes dumb as door, loves to create regressions, is a fucking terrible designer. Small, tiny changes? Those are actually the worse, it is easy for claude, on the first setback, decides to burn the whole world and start from zero again. Not to mention when it gets stuck in an eternal loop where it increasingly degenerates the code.
If I care about what I deliver, I have to actively participate in coding.
I'm 57 and wrote my first line of BASIC in 1980, so while I can still chime in on this specific demographic I feel that I ought to. So im like this guy, but like a lot of other people in my specific demographic we aren't writing these long melancholy blog posts about AI because it's not that big of a deal. As an OSS maintainer most of my work is a lot of boring slog adding features to libraries to suit new features in upstream dependencies, nitpicky things people point out, new docs, tons of tedium. Claude helps a ton with all of that. no way is Claude doing the real architectural puzzle stuff, that's still fully on me! I can just use Claude to help implement it. It's like the ultimate junior programmer assistant. It's certainly a new, different and unique experience in one's programming career but it really feels like another tool, like an autocomplete or code refactoring tool that is just a lot better, with similar caveats. I mean in my career, I've had to battle the whole time people who don't "get" source code control (starting with me), who don't "get" IDEs (starting with me), people who dont "get" distributed version control (same), people who don't "get" ORMs (oh yes, same for me though this one I took much more dramatic steps to appreciate them), people who don't "get" code formatters, now we're battling people who don't "get" LLMs used for coding, in that sense the whole thing doesnt feel like that novel of a situation.
it's the LLMs that are spitting out fake photos and videos and generating lots of shitty graphics for local businesses, that's where I'm still wielding a pitchfork...
It'd be more strange if the thing you learned 43 years ago was exactly the same today. We should expect change. When that change is positive we call it progress.
In the grand scheme of things it wouldn’t actually be that strange: generations and generations of humans were mostly farmers and mostly did the same thing as their parents. Of course technology developed but lots of people did the same job with the same methods their whole lives.
But everybody on this site lived through the first half of a logistic curve so that perspective seems strange to us.
I think more than ever programmers need jobs where performance matters and the naive way the AI does things doesn't cut it. When no one cares about things other than correctness your job turns into AI Slop. The good news right now is that AI tends to produce things that AI struggles to do well with so large scale projects often descend into crap. You can write a C-compiler for $20,000 with an explosive stack of agents, but that C-compiler isn't anywhere close to efficient or performant.
As model costs come down that $20,000 will become a viable number for doing entirely AI-generate coding. So more than ever you don't want to be doing work that the AI is good enough at. Either jobs where performance matters or being able to code the stack of agents needed to produce high quality code in an application context.
I wonder what other “crevices” (as the author put it) exist.
Another commentor mentioned embedded, and after a brief phase of dabbling in that, mainly with nRF5x micros, I tend to agree. Less training data and obtuse tooling.
"They’re writing TypeScript that compiles to JavaScript that runs in a V8 engine written in C++ that’s making system calls to an OS kernel that’s scheduling threads across cores they’ve never thought about, hitting RAM through a memory controller with caching layers they couldn’t diagram, all while npm pulls in 400 packages they’ve never read a line of."
and they still call themselves 'full stack developers' :eyeroll:
>But sure. AI is the moment they lost track of what’s happening. The abstraction ship sailed decades ago.
Bullshit. While abstraction has increased over time, AI is no mere incremental change. And the almost natural language interaction with an agent is not the same as Typescript over assembly (not to mention you could very well right C or Rust and the like, and know most of the details of the machine by heart, and no, microcode and low level abstractions are not a real counter-argument to that). Even less so if agents turn autonomous and you just herd them onto completion.
This LLM stuff is a little weird. Previously we had Python which was pretty close to pseudocode but you could run it directly. Now, these LLMs are one step more abstract, but their outputs aren’t runnable directly, they produce possibly incorrect code-like-text. Actually this seems like good news for programmers since you have to read the code in the lower-level language that gets produced.
The irony of these "My craft is dead" posts is that they consistently, heavily leverage AI for their writing. So you're crying about losing one craft to AI while using AI to kill another. It's disingenuous. And yes it is so damn obvious.
I think the Oxide computer LLM guidelines are wise on this front:
> Finally, LLM-generated prose undermines a social contract of sorts: absent LLMs, it is presumed that of the reader and the writer, it is the writer that has undertaken the greater intellectual exertion. (That is, it is more work to write than to read!) For the reader, this is important: should they struggle with an idea, they can reasonably assume that the writer themselves understands it — and it is the least a reader can do to labor to make sense of it.
The heavy use of LLMs in writing makes people rightfully distrustful that they should put the time in to try to read what's written there.
Using LLMs for coding is different in many ways from writing, because the proof is more there in the pudding - you can run it, you can test it, etc. But the writing _is_ the writing, and the only way to know it's correct is to put in the work.
That doesn't mean you didn't put in the work! But I think it's why people are distrustful and have a bit of an allergic reaction to LLM-generated writing.
I mean, obviously you can't know your actual error rates, but it seems useful to estimate a number for this and to have a rough intuition for what your target rate is.
I agree with that for programming, but not for writing. The stylistic tics are obtrusive and annoying, and make for bad writing. I think I'm sympathetic to the argument this piece is making, but I couldn't make myself slog through the LinkedIn-bot prose.
I felt the same. I resonate with the message, but it really rings hollow with so much AI directing.
I'd wish people would stop doing that. AI writing isn't even particularly good. Its not like it makes you into Dostoevsky, it just sloppifies your writing with the same lame mannerisms ("wasn't just X — it was Y"), the same short paragraphs, the same ems.
The author admits that they used AI but I found it not that obvious. What are telltale signs in this case? While the writing style is a little bit over-stylized (exactly three examples in a sentence, Blade Runner reference), I might write in a similar style about a topic that im very emotional about. The actual content feels authentic to me.
(1) The pattern "It's not just a X---It's a Y" is super common in LLM-generated text for some reason. Complete with em dash. (I like em dashes and I wish LLMs weren't ruining them for the rest of us)
"Upgrading your CPU wasn’t a spec sheet exercise — it was transformative."
"You weren’t just a user. You were a systems engineer by necessity."
"The tinkerer spirit didn’t die of natural causes — it was bought out and put to work optimising ad clicks."
And in general a lot of "It's not <alternative>, it's <something else>", with or without an em dash:
"But it wasn’t just the craft that changed. The promise changed."
it's really verbose. One of those in a piece might be eye-catching and make someone think, but an entire blog post made up of them is _tiresome_.
(2) Phrasing like this seems to come out of LLMs a lot, particularly ChatGPT:
"I don’t want to be dishonest about this. "
(3) Lots of use of very short catch sentences / almost sentence fragments to try to "punch up" the writing. Look at all of the paragraphs after the first in the section "The era that made me":
"These weren’t just products. " (start of a paragraph)
"And the software side matched." (next P)
"Then it professionalised."
"But it wasn’t just the craft that changed."
"But I adapted." (a few paragraphs after the previous one)
And .. more. It's like the LLM latched on to things that were locally "interesting" writing, but applies them globally, turning the entire thing into a soup of "ah-ha! hey! here!" completely ignorant of the terrible harm it does to the narrative structure and global readability of the piece.
I'm weird about this, I choose to use AI to get feedback on my writing, but refuse to just copy and paste the AIs words. I only do it if its a short work email and I really dont care about its short lived lifespan, if its supposed to be an email where the discussion continues, then I refine it. I can write a LOT. If HN has edit count logs, I've probably got the high score.
This is not either of those. This is the equivalent of a eulogy to a passion and a craft. Using an LLM to write it: entire sections, headers, sentences - is an insult to the craft.
The post in the same vain, "We mourn our craft", did a much better job at this communicating the point without the AI influence.
At least then you’re being honest about you hating your intended audience, and not proudly posting the slop vomited forth from your algorithmic garbage machine as if it were something that deserved the time, thought and consideration of your equals.
“... when I was 7. I'm 50 now and the thing I loved has changed”
Welcome to the human condition, my friend. The good news is that a plurality of novels, TV shows, country songs, etc. can provide empathy for and insight into your experience.
There's 3-4 of these posts a day - why don't people spend more time hand-building things for fun in their free time? That's what led a lot of us to this career path to start with. I have a solid mix of hand-code and AI-assisted projects in my free time.
>>The machines I fell in love with became instruments of surveillance and extraction.
Surveillance and Extraction
"We were promised flying cars", and what we got was "investors" running the industry off the cliff into cheap ways to extract money from people instead of real innovation.
This isn't new. It's the same feeling the first commercial programmers had working in assembly, or machine code, once compilers became available. Ultimately I think even Mel Kaye forsook being able to handpick memory locations for optimum drum access before his retirement, in favor of being able to build vastly more complex software than before.
AI has just vastly extended your reach. No sense crying about it. It is literally foolish to lament the evolution of our field into something more.
As someone who has always enjoyed designing things, but was never really into PUZZLES, I always felt like an outsider in the programming domain. People around me really enjoyed the "fun" of programming, whereas I was more interested in the Engineering of the thing - balancing tradeoffs until within acceptable margins and then actually calling it "DONE". People around me rarely called things "done", they rewrote it and rewrote it so that it kept satisfying their need for puzzle-solving (today, it's Ruby, tomorrow, it's rewritten in Scala, and the day after that, it's Golang or Zig!)
I feel that LLMs have finally put the ball in MY court. I feel sorry for the others, but you can always find puzzles in the toy section of the bookstore.
Don't program as a career, but am also 50 and programming since TRS-80. AI has transformed this era, and I LOVE IT! I can focus on making and not APIs or syntax or all of the bootstrapping.
Professional development is changing dramatically. Nothing stops anyone from coding "the old way," though. Your hobby project remains yours, exactly the way you want it. Your professional project, on the other hand, was never about you in the first place. It's always about the customer/audience/user, period full stop.
I have the opposite take. There’s nothing stopping you from jumping into any component to polish things up. You can code whatever you wish. And AI takes away nearly all of the drudgery : boilerplate, test cases, inspecting poor documentation, absurd tooling.
It also lets me focus more on improving things since I feel more liberated to scrap low quality components. I’m much braver to take on large refactors now – things that would have taken days now take minutes.
In many ways AI has made up for my growing lack of patience and inability to stay on task until 3am.
11 and now 45. I am still interested in it, but I feel like in my 20s I would get a dopamine rush when a row showed up in a database. In my 30s I would get that only if a message passed through a system and updated on-screen analytics within 10 seconds. Thank god for LLMs because all of it became extremely boring, I can't stand having to get these little milestones each new company or each new product I'm working on. At least with LLMs the dopamine hit comes from being in awe of the code that gets generated and realizing it found every model, every messaging system interface, every API, and figuring out how to make it backwards compatible, updating the UI - something that would take half a day, now in 5 minutes or less.
I’m 50 too and I’ve complained and yearned about the “old” days too, a lot of this is nostalgia as we reminisce about periods of time in our youth when we had the exuberance and time to play and build with technology of our own time
Working in AI startups strangely enough I see a lot of the same spirit of play and creativity applied to LLM based tools - I mean what is OpenClaw but a fun experiment
Those kids these days are going to reminisce about the early days of AI when prompts would be handwritten and LLMs would hallucinate
I’m not really sure 1983, 1993 or 2003 really was that gold of age but we look at it with rose colored glasses
I'd feel the same when I was younger. Over time I've realized that they are the lucky ones. You too, if you're lucky, will one day be an old man doing old man things.
Wow... I really relate to this. I'm 50 as well, and I started coding in 1985 when I was 10... I remember literally every evolutionary leap forward and my experience with this change has been a bit different.
Steve Yegge recently did an interview on vibe coding (https://www.youtube.com/watch?v=zuJyJP517Uw) where he says, "arch mage engineers who fell out-of-love with the modern complexity of shipping meaningful code are rediscovering the magic that got them involved as engineers in the first place" <-- paraphrased for brevity.
I vividly remember, staying up all night to hand-code assembler primitive rendering libraries, the first time I built a voxel rendering engine and thinking it was like magic what you could do on a 486... I remember the early days at Relic, working on Homeworld and thinking we were casting spells, not writing software. Honestly, that magic faded and died for me. I don't personally think there is magic in building a Docker container. Call me old-fashioned.
These days, I've never been more excited about engineering. The tedium of the background wiring is gone. I'm back to creating new, magical things - I'm up at 2 AM again, sitting at my desk in the dark, surrounded by the soft glow of monitors and casting spells again.
I just told my gardener to cut the grass and work on some flower installations.
I'm so excited about gardening again. Can't wait to do some. Employing a gardener to do my gardening for me is really making me enjoy gardening again!
I think this works unironically. My mother is an avid gardener and can spend 8 hours a day gardening. When her life circumstances allowed for it, she hired a once a week gardener to do the tasks she didn't like (or had difficulties doing as a small woman), and still gardens the same amount. I've teased her for hiring a gardener, but she swears it's a huge help and boost to her gardening quality of life.
this is a great analogy despite it possibly coming off as snark.
I think it's hard for some people to grasp that programmers are motivated by different things. Some are motivated by shipping products to users, others are motivated to make code that's a giant elegant cathedral, still others love glorious hacks to bend the machine into doing things it was never really intended to do. And I'm sure I'm missing a few other categories.
I think the "AI ain't so bad" crowd are the ones who get the most satisfaction out of shipping product to users as quickly as possible, and that's totally fine. But I really wish they'd allow those of us who don't fall into that category to grieve just a little bit. This future isn't what I signed up for.
It's one thing to design a garden and admire the results, but some people get into their "zen happy place" by pulling up weeds.
I agree with this, I put myself in the "glorious hacks to bend the machine into doing things it was never really intended to do" camp, so the end game is somthing cool, now I can do 3 cool things before lunch instead of 3 cool things a year
Having opencode doesn't preclude me from making elegant code. It just takes away the carpel tunnel.
*I'm so excited about landscape design. Can't wait to do more. Employing a gardener to do the gardening for me is really making me enjoy landscape design again!
The proper analogy would be you can now remove all weeds with the swipe of your hand and cut all your hedges with another swipe, you still are gardening you can do it quicker and therefore explore different possibilities.
Imagine though instead of 1 garden you can make 10 or 30 gardens in the same time that are more extravagant than your 1 garden was. At any point in time you can dive back in 1 of them and start plucking away
Oh, the joy that awaits you when you come back home to discover how the gardener interpreted "please trim the hedge by the gate a little".
Well it's more like employing a gardener makes me enjoy landscaping again. It's not like we ever found writing words on a keyboard all that great, it's fundamentally just about having an idea and turning it into something real.
As mosburger says, this is a great analogy. Do you think that the great artists paint, sculpt, and draw everything by hand, by themselves? Of course not... they never did, and they don't today. You're being offered the ability to join their ranks.
It's your studio now. You have a staff of apprentices standing by, eager for instructions and commands. And you act like it's the worst thing that ever happened to you.
Is this sarcasm? I can't tell.
No, it's not.
If you want things to stay the same forever, you shouldn't go into technology, art, or gardening. Try plumbing, masonry, or religion.
[55yo] My sense is that those problems we worked on in the 80s and 90s were like the perfectly balanced MMORPG. The challenges were tough, but with grit, could be overcome and you felt like you could build something amazing and unique. My voxel moment was passing parameters in my compilers class in college. I sat down to do it and about 12 hours later I got it working, not knowing if I could even do it.
With AI, it is like coding is on GOD mode and sure I can bang out anything I want, but so can anyone else and it just doesn't feel like an accomplishment.
>With AI, it is like coding is on GOD mode and sure I can bang out anything I want, but so can anyone else and it just doesn't feel like an accomplishment.
I think it's possible that we'll get to the point where "so can anyone else" becomes true, but it isn't today for most software. There's significant understanding required to ask for the right things and understand whether you're actually getting them.
That said, I think the accomplishment comes more so from the shaping of the idea. Even without the typing of code, I think that's where most of the interesting work lies. It's possible that AI develops "taste" such that it can sufficiently do this work, but I'm skeptical it happens in the near term.
I think there's still quite a chasm out there. Domain knowledge, an informed and opinionated view on how something should function, and overall tech knowledge are still key. Having those three things continues to greatly differentiate people of equal coding skill, as they always have.
but so can anyone else and it just doesn't feel like an accomplishment.
So it's not enough that you get to do cool stuff, the important part is that nobody else gets to. Is that it?
If so, other sites beckon.
> I don't personally think there is magic in building a Docker container. Call me old-fashioned.
This seems like a false dichotomy. You don't have to do this. It is still possible to build magical things. But agents aren't it, I don't think.
It is honestly extremely depressing to read this coming from a founder of Relic. Relic built magic. Dawn of War and Company of Heroes formed an important part of my teenage years. I formed connections, spent thousands of hours enjoying them together with other people, and pushed myself hard to become one of the top 100 players on the CoH leaderboards. Those competitive multiplayer games taught me everything there was to know about self-improvement, and formed the basis of my growth as an individual - learning that if I put my mind to it, I could be among the best at something, informed my worldview and led me to a life of perpetually pushing myself to further self-improvement, and from there I learned to code, draw, and play music. All of that while being part of amazing communities where I formed friendships that lasted decades.
All of this to say, Relic was magic. The work Relic did profoundly impacted my life. I wonder if you really believe your current role, "building trust infrastructure for AI agents", is actually magic? That it's going to profoundly impact the lives of thousands or millions?
I'm sorry for the jumbled nature of this post. I am on my phone, so I can't organize my thoughts as well as I would like. I am grateful to you for founding Relic, and this post probably comes off stupidly combative and ungrateful. But I would simply like to pose to you, to have a long think if what you're doing now is really where the magic is.
Edit: On further consideration, it's not clear the newly-created account I'm responding to is actually Alex Garden. The idea of potentially relating this personal anecdote to an impersonator is rather embarrassing, but I will nonetheless leave this up in the hope that if there are people who built magical things reading this, regardless of whether they're Alex Garden or someone else, that it might just inspire them to introspection about what building magic means, about the impact software can have on people's lives even if you don't see it, and whether this "agent" stuff is really it.
Oh man, modding Company of Heroes was one of the things that got me into programming. I look back fondly on those memories.
>The idea of potentially relating this personal anecdote to an impersonator is rather embarrassing
Good news! You've also related it to the roughly ~3-10M monthly HN readers who are not (potentially) impersonating the founder of a beloved game studio.
Also: I think you're probably safe. I'm sure someone at some point has come to HN to LARP as some prominent person in tech that they don't happen, at that specific moment, to actually be... but I can't really think of it happening before, nor would I expect it to take the form of a particularly thoughtful comment if a troll did that. Though with AI these days, who knows? I might myself just be one of a swarm of clawd/molt/claw things. In which case I'd be the last to even know it.
Oh-- as for being depressed about their docker/wiring things up sentiment. Try not to be, and instead, consider: Is it a surprise that someone who founded such a place as relic was occasionally-- even often-- frustrated at the things they had to clear away to build the thing they actually wanted to build? People who want to build amazing experiences may not love having to clear clutter that gets in their way. Other people want to build the tools that clear clutter, or other things that keep the whole system going. Those are beautiful too.
I started a bit younger and am a bit older, and relate. But only so much. I started programming in 3rd grade (also BASIC) when I found a computer and learned how to play a game on it, then found the source code for the game and randomly started changing it. In 7th grade I was paid to port a BASIC program to C (super new at the time), which I did on paper because I didn't own a computer (I used the money to buy my first). To be clear, I was really bad a programming for a long time and simply kept at it until I wasn't.
I love messing about with computers still. I can work at the byte level on ESP-32s on tiny little devices, and build massive computation engines at the time time on the same laptop. It's amazing.
I feel for those who have lost their love of this space, but I have to be honest: it's not the space that's the problem. Try something new, try something different and difficult or ungainly. Do what you rail against. Explore.
That's what it's always been about.
> I don't personally think there is magic in building a Docker container. Call me old-fashioned.
I still vividly remember setting up gcc in a docker container to cross compile custom firmware for my cannon camera and thinking about the amount of pain my local system would have been in if I had to do all the toolchain work in my host OS. Don't know if it felt like magic, but it sure didn't hurt like the alternative!
For me it's both - I mourn the loss of my craft ( and my identity ) but I'm also enjoying the "magic".
Last night I was thinking about this "xswarm" screen saver I had in 1992 on my DEC Ultrix workstation. I googled for the C source code and found it.
I asked Claude to convert it to Java, which it did in a few seconds. I compiled and ran it, and there it was again, like magic
In the second half of my 40s now and I'm in the same boat. I started slapping keys on a c64 when I was 2 years old. Really enjoyed software development until 10-15 years ago. With the current LLM tooling available the number of systems I've been able to build that are novel and tackle significant problems has been kind of mind blowing over the past 8 months or so.
Staying up late, hacking away at stuff like I used to, and it's been a blast.
Finally, Homeworld was awesome and it felt magical playing it.
I'm feeling the same.
AI development actually feels like a similar rate of change. It took 8 years to go from the Atari 2600 to the Amiga.
An 8 year old computer doesn't quite capture the difference today.
Wow, Alex Garden on Hackernews. Hello fellow canuck. I'm now getting up there, still a few years shy of y'all but not much. I came up through the 90s and early 2000s, all web/linux stuff, irc servers, bash scripts, python, weird php hacks, whatever, I was a kid. I'd lose track of time, It was Monday night after high school then all of a sudden it was Sunday morning and I was talking on irc about the crazy LAMP stack I'd put together. 2am? pfft, what is sleep?! Sadly with very strong dyslexia and dyscalculia, being a real programmer was never in the cards for me, I understood how everything worked, I can explain the whole thing end to end in great depth, but ask me predictably how to do a table in html or some fairly simple CSS, and I'll be there for hours. I'm grateful the rest of my life allowed me to be programmer adjacent and spend so much time around developers, but always a little frustrated I couldn't pick up the hammer myself.
These days, I've never been more excited about building. The frustration of being slow with the code is gone. I'm back to creating new, magical things - I'm up at 2 AM again, sitting at my desk in the dark, surrounded by the soft glow of monitors and casting spells.
Why is your last paragraph nearly identical to the last paragraph you are replying to? It might have been a strange quirk, but there’s also been the suggestion that the post you’re replying to is an imposter, so it gets weirder that you also did that.
I thought I was being cute. :) I'm not a bot. Although I did think the original msg read a bit LLM, I try not to judge these days.
Go Canada! I personally can't wait to see what happens to the world when all of us find the passion to create again.
Yes yes yes!!!
I'm 45 yo. And also started programming quite early around 1988. In my case it was GWBAsic games and then C ModeX and A Later Allegro based games.
Things got so boring in the last 15 years, I got some joy in doing AI research (ML, agents, Genetic Algorithms, etc).
But now, it's so cool how I can again think about something and build it so easily. I'm really excited of what I can do now. And im ot talking about the next billion dollar startup and whatnot. But the small hacky projects that LLMs made capable.yo build in no time.
Yeah it’s drugs and or religion. Feels pretty good.
My advice to everyone feeling existential vertigo over these tools is to remain confident and trust in yourself. If you were a smart dev before AI, chances are you will remain a smart dev with AI.
My experience so far is that to a first approximation, the quality of the code/software generated with AI corresponds to the quality of the developer using the AI tool surprisingly well. An inexperienced, bad dev will still generate a sub-par result while a great dev can produce great results.
The choices involved in using these tools are also not as binary as they are often made out to be, especially since agents have taken off. You can very much still decide to dedicate part of your day to chiseling away at important code to make it just right and make sure your brain is engaged in the result and exploring and growing with the problem at hand, while feeding background queues of agents with other tasks.
I would in fact say the biggest challenge of the AI tool revolution in terms of what to adapt to is just good ol' personal time management.
> If you were a smart dev before AI, chances are you will remain a smart dev with AI.
I don't think that's what people are upset about, or at least it's not for me. For me it's that writing code is really enjoyable, and delegating it to AI is hell on earth.
This is a part of it, but I also feel like a Luddite (the historical meaning, not the derogatory slang).
I do use these tools, clearly see their potential, and know full well where this is going: capital is devaluing labor. My skills will become worthless. Maybe GP is right that at first only skilled developers can wield them to full effect, but it's obviously not going to stop there.
If I could destroy these things - as the Luddites tried - I would do so, but that's obviously impossible.
For now I'm forced to use them to stay relevant, and simply hope I can hold on to some kind of employment long enough to retire (or switch careers).
> know full well where this is going: capital is devaluing labor
But now you too can access AI labor. You can use it for yourself directly.
> If I could destroy these things - as the Luddites tried - I would do so, but that's obviously impossible.
Certainly, you must realize how much worse life would be for all of us had the Luddites succeeded.
The historical luddites are literally the human death drive externalized. Reject them and all of their garbage ideas with extreme prejudice.
Related, the word “meritocracy” was coined in a book which was extremely critical of the whole concept. AI thankfully destroys it. Good riddance, don’t let the door hit your ass on the way out.
https://en.wikipedia.org/wiki/The_Rise_of_the_Meritocracy
You can reject the ideas in the aggregate. Regardless, for the individual, your skills are being devalued, and what used to be a reliable livelihood tied to a real craft is going to disappear within a decade or so. Best of luck
I resonate with that. I also find writing code super pleasurable. It's immediate stress relief for me, I love the focus and the flow. I end long hands-on coding sessions with a giddy high.
What I'm finding is that it's possible to integrate AI tools into your workflow in a big way without giving up on doing that, and I think there's a lot to say for a hybrid approach. The result of a fully-engaged brain (which still requires being right in there with the problem) using AI tools is better than the fully-hands-off way touted by some. Stay confident in your abilities and find your mix/work loop.
It's also possible to get a certain version of the rewards of coding from instrumenting AI tools. E.g. slicing up and sizing tasks to give to background agents that you can intuit from experience they'll be able to actually hand in a decent result on is similar to structuring/modularization exercises (e.g. with the goal to be readable or maintainable) in writing code, feelings-wise.
Hope: I want to become a stronger dev.
Reality: Promoted to management (of AI) without the raise or clout or the reward of mentoring.
> ...the reward of mentoring.
I really feel this. Claude is going to forget whatever correction I give it, unless I take the time and effort to codify it in the prompt.
And LLMs are going to continue to get better (though the curve feels like it's flattening), regardless of whatever I do to "mentor" my own session. There's no feeling that I'm contributing to the growth of an individual, or the state-of-the-art of the industry.
LLMs are similar in a lot of ways to the labor outsourcing that happened a generation or two ago. Except that instead of this development lifting a billion people out of poverty in the third world a handful of rich people will get even more rich and everyone else will have higher energy bills.
There's room for both. Give AI the boilerplate, save the exciting stuff for you.
exactly
thankfully I started down the FIRE route 20 years ago and now am more or less continuing to work because I want to
which will end for my employer if they insist on making me output generative excrement
i agree. it seem like an expectation these days to use AI sometimes... for me i am happy not using it at all, i like to be able to say "I made this" :)
Yes, absolutely. I think the companies that don't understand software, don't value software and that think that all tech is fundamentally equivalent, and who will therefore always choose the cheaper option, and fire all their good people, will eventually fail.
And I think AI is in fact a great opportunity for good devs to produce good software much faster.
I think the issue is that given the speed the bad dev can generate sub-par results that at face value look good enough overwhelm any procedures in place.
Pair that with management telling us to go with AI to go as fast as possible means that there is very little time to do course correction.
I think no one is better positioned to use these tools than experienced developers.
I agree with the quality comments. The problem with AI coding isn't so much the slop, it's the developers not realizing its slop and trying to pass it off as a working product in code reviews. Some of the stuff I've reviewed in the past 6 months has been a real eye opener.
For me the problem is simple: we are in an active prisoner's dilemma with AI adoption where the outcome is worse collectively by not asking the right questions for optimal human results, we are defecting and using ai selfishly because we are rewarded by it. There's lots of potential for our use to be turned against us as we train these models for companies that have no commitment to give to the common good or return money to us or to common welfare if our jobs are disrupted and an AI replaces us fully.
What the author describes is also the feeling when you shift from being a developer all day to being a team lead or manager. When you become a lead you have to let go and get comfortable with the idea that the code is not going to be how you would do it. You can look at code produced by your team and attempt to replace it all with your craftsmanship but you're just setting yourself up to fail. The right approach is use your wisdom to make the team better, not the code. I think a lot of that applies to using AI when coding.
I'm turning 50 in April and am pretty excited about AI coding assistants. They make a lot of personal projects I've wanted to do but never had the time feasible.
Most of my career has been as an individual engineer, but the past few years I have been a project manager. I find this to be very much like using AI for coding.
Which also makes me refute the idea that AI coding is just another rung up on the programming abstraction ladder. Depending on how much you delegate to AI, I don't think it's really programming at all. It's project management. That's not a bad thing! But it's not really still programming.
Even just in the context of my human team, I feel less mentally engaged with the code. I don't know what everything does. (In principle, I could know, but I don't.) I see some code written in a way that differs from how I would have done it. But I'm not the one working day-in, day-out with the code. I'll ask questions, make suggestions, but I'm not going to force something unless I think it's really super important.
That said, I don't 100% like this. I enjoy programming. I enjoy computer science. I especially enjoy things more down the paths of algorithm design, Lisp, and the intersection of programming with mathematics. On my team, I do still do some programming. I could delegate it entirely, but I indulge myself and do a little bit.
I personally think that's a good path with AI too. I think we're at the point where, for many software application tasks, the programming could be entirely hands-off. Let AI do it all. But if I wish to, why not indulge in doing some myself also? Yeah, I know, I know, I'll get "left behind in the dust" and all of that. I'm not sure that I'm in that much of a hurry to churn out 50,000 lines of code a day; I'm cool with 45,100.
I find that AI allows me to get into algorithm design more, and the intersection of math and programming more, by avoiding boilerplate.
You can indulge even more by letting AI take care of the easy stuff so you can focus on the hard stuff.
A lot of us resist the pressure to move to management or technical leadership for just these reasons. Programming people isn't the same as programming computers.
But the LLMs outnumber us. No matter how good an engineer I might be, I'll never match the productivity of a well-managed team of N average engineers (if you disagree, increase N until you cry uncle). Sure, there will be mythical man-month problems. But the optimal N is surely greater than 1, and I'll never be more than 1.
Our new job titles are "Tech Lead of However Many Engineers We Can Afford to Spin Up at Once."
It's fun managing a bunch of inexperienced juniors when there are no consequences (aka the infamous personal projects). It's a lot more stressful when it matters.
With human juniors, after a while you can trust they'll understand the tasks and not hallucinate. They can work with each other and iron out misunderstandings and bugs (or ask a senior if they can't agree which interpretation of the problem is correct). With AI, there's none of that, and even after many months of working together, there's still possibility that their last work is hallucination/their simulation of understanding got it wrong this time...
The equivalent of "employee development" with AI is just the release schedule of new models, I guess.
But the release of new models are generic. They don’t represent understanding in your specific codebase. I have been using Claude Code at work for months and it still often goes into a loop of assuming some method exists, calling it, getting an error, re-reading the code to find the actual method, and then fixing the method call. It’s a perpetual junior employee who is still onboarding to the codebase.
Yeah, I've experienced similar stuff. Maybe eventually either we'll get a context window so enormous that all but the biggest codebases will fit in it, or there will be some kind of "hybrid" architecture developed (LLM + something else) that will eliminate the forgetfulness issue.
It's also that when you move to being a leader, you suddenly have to learn to quantify and measure your productivity in a different way, which for a while can really do a number on your self-image.
What does it mean to be a productive developer in an AI tooling age? We don't quite know yet and it's also shifting all the time, so it becomes difficult to sort yourself into the range stably. For a lot of accomplished folks this is the first time they've felt that level of insecurity in a while, and it takes some getting used to.
> What the author describes is also the feeling when you shift from being a developer all day to being a team lead or manager.
I think that's very true. But... there's a reason I'm not a team lead or manager. I've done it in the past and I hate it. I enjoy doing the work, not tasking others with doing work.
I'm 60, started with a Tandy Model I in junior high, learned 6809 assembly for my Color Computer, loved the fact we could put certain values in particular memory positions and change the video mode and put pixels to the screen. It's been decades of losing that level of control, but for me coding is the fun part. I've never lost that spark of enjoyment and really obsession I felt early on. I enjoy the supposedly boring job of writing SQL and C with embedded SQL and working with business concepts to produce solutions. Coding is the fun part for me, even now.
I got moved up the chain to management and later worked to get myself moved back down to a dev role because I missed it and because I was running into the Peter Principle. I use AI to learn new concepts, but mostly as a search engine. I love the tech behind it, but I don't want it coding for me any more than I want it playing my video games for me. I was hoping AI would show up as robots doing my laundry, not doing the thing I most enjoy.
I am much younger than the author, but I've been coding for most of my life and I find close to no joy in using AIs. For me coding has always been about the nitty-gritty quirkiness of computers, languages, solving issues and writing new cool things for the sake of it. It was always more about the journey than the end goal, and AI basically hollows out all of the interesting bits about coding. It feels like skipping straight to the end of a book, or somewhat like that.
I don't know if I am the only one, but developing with chatbots in my experience turns developing software into something that feels more akin to filling out forms or answering to emails. I grieve for the day we'll lose what was once a passion of mine, but unfortunately that's how the world has always worked. We can only accept that times change, and we should follow them instead of complaining about it.
> For me coding has always been about the nitty-gritty quirkiness of computers, languages, solving issues and writing new cool things for the sake of it.
Same. It scratches my riddle-solving itch in a way that the process of "prompt-honing" has yet to do.
I'm lucky because I work as an independent consultant. I get paid to deliver solutions, but I get to choose how to create those solutions. I write whatever code I want however I want. As long as it solves the problem, no one cares.
I started programming in 1980, and I having just as much fun now as I did then. I literally cannot wait to sit down at my IDE and start writing.
But that was not always true. When I worked for a larger company, even some startups, it was not always fun. There's something about having full control over my environment that makes the work feel like play.
If you feel like programming isn't fun anymore, maybe switching to a consulting gig will help. It will give you the independence and control that you might be craving.
I have a hard time telling whether agentic coding tools will take a big bite out of the demand for software consultants. If the market is worried about SaaS because people think companies will use AI to code tools internally vs buying them, I would think the same would apply to consultants.
I’ve seen the code current tools produce if you’re not careful, or if you’re in a domain where training data is scarce. I could see a world where a couple of years from now companies need to bring outside people to fix vibe coded software that managed to gain traction. Hard to tell.
Not going to pull age or title rank here -- but I suggest if your use of AI feels empty, take advantage of its speed and plasticity and iterate upon its output more, shape the code results. Use it as a sculptor might too -- begin with its output and make the code your own. I particularly like this latter approach when I am tasked with use of a language I view as inferior and/or awkward. While this might read as idealistic, and I agree that there are situations where this interaction is infeasible or inappropriate, you should also be encountering problems where AI decidedly falls on its face and you need to intervene.
It seems AI is putting senior developers into two camps. Both groups relate to the statement, "I started programming when I was seven because a machine did exactly what I told it to, felt like something I could explore and ultimately know, and that felt like magic. I’m fifty now, and the magic is different, and I’m learning to sit with that."
The difference is that the first camp is re-experiencing that feeling of wonder while the second camp is lamenting it. I thankfully fall in the first camp. AI is allowing me to build things I couldn't, not due to a lack of skills, but a lack of time. Do you want to spend all your time building the app user interface, or do you want to focus on that core ability that makes your program unique? Most of us want the latter, but the former takes up so much time.
It sounds like you don’t particularly care about the user interface, and that’s why you’re okay with delegating it. I think the developers who don’t like delegating to AI are the ones who care about and have strong opinions about all the parts. To them there are no unimportant parts where the details don’t matter.
Similarly, I'm using it to write apps in non-native languages, like rust. My first foray into it led to finding poor documentation examples. AI allows me to create without spending large swaths of time learning minutia.
I'm enjoying it to a point, but yes, it does eliminate that sense of accomplishment - when you've spent many late nights working on something complex, and finally finish it. That's pretty much gone.
delegating UI to the 'not worth my time' pile is how you end up with a poor UI
I have given the topic some thoughts. I concluded that the ONLY way for ordinary people (non-genius, IQ <= 120) to be really good, be really close to the genius, is to sit down, condensate the past 40 or so year's tech history of three topics (Comp-Arch, OS and Compiler) into a 4-5 years of self-education.
Such education is COMPLETELY different from the one they offered in school, but closer to those offered in premium schools (MIT/Berkeley). Basically, I'd call it "Software engineering archaeology". Students are supposed to take on ancient software, compile them, and figure out how to add new features.
For example, for the OS kernel branch:
- Course 0: MIT xv6 lab, then figure out which subsystem you are interested in (fs? scheduler? drivers?)
- Course 0.5: System programming for modern Linux and NT, mostly to get familiar with user space development and syscalls
- Course 1: Build Linux 0.95, run all of your toolchains in a docker container. Move it to 64-bit. Say you are interested in fs -- figure out the VFS code and write a couple of fs for it. Linux 0.95 only has Minix fs so there are a lot of simpler options to choose from.
- Course 2: Maybe build a modern Linux, like 5.9, and then do the same thing. This time the student is supposed to implement a much more sophiscated fs, maybe something from the SunOS or WinNT that was not there.
- Course 3 & 4: Do the same thing with leaked NT 3.5 and NT 4.0 kernel. It's just for personal use so I wouldn't worry about the lawyers.
For reading, there are a lot of books about Linux kernels and NT kernels.
Thank you for writing this. My feelings are very similar to the ones described by the author and the timeline almost matches. The thrill of tecnology for me started to fast decay since the early 2010s and now I see it as a no-return stage. I still have fun with my retro hardware & software but I am no longer an active practitioner and I have pivoted my attention and my efforts somewhere else. Unfortunately, I no longer feel excited for the future decades of tech and I am distancing myself from it.
I think this is something else, though. Even before AI really hit sweng, there were early signs of a collective tech depression a la "The best idea we can come up with is strapping screens to people's heads?", the "Are we the bad guys?" convo around social media, the crypto brain drain, etc. The queue of Next Big Things has increasingly felt more forced and controversial to many, and being in tech last lost much of its lustre to them.
I think it's healthy for everyone to evaluate whether one's personal reaction to AI is colored by this trend, or whether it's really being evaluated independently. Because while I share many of the negative feelings listed earlier, to me AI does still feel different; it has a lot more real utility.
What else do you do to make rent ? I feel the same way as you and I have no idea what else pays well for quality craftsmanship. I am staring at the abyss of hyper intelligent people with posh resumes and now wondering what to do.
What do you do for living now (if anything)?
6 or 7 , 38 now -- and having a blast.
it isn't all funeral marches and group crying sessions.
And don't let the blog post fool you , it is a rant about AI -- otherwise we would have heard complaints about the last 200 paradigm shifts in the industry over the past thirty years.
Sure, we got our share of dilbert-style agile/waterfall/tdd jokes shoved in our face, but no one wrote a blog post about how their identity was usurped by the waterfall model .
>And different in a way that challenges the identity I built around it and doesn’t satisfy in the way it did.
Everyone should do their own thing, but might I suggest that it is dangerous for anyone in this world to use a single pillar as their foundation for all identity and plinth of their character.
but no one wrote a blog post about how their identity was usurped by the waterfall model
I don’t know about that.
Waterfall mostly died before the rise of blogs, of course, but around the dawn of Agile I remember lots of posts about how nothing was properly designed any more, nothing was ever finished, and you never knew what the specification was.
They used to be real engineers, but now it was just all chaos! They couldn’t design anything any more!
Thanks for reminding me of the word plinth. I agree with the author that the job is less fun now, less interesting. I'm doing and accomplishing more, and it matters less. And unfortunately, having other ways of defining your identity doesn't really help, for me. What it does is make those other aspects of myself relatively more attractive as careers, in comparison to this one. Although then again, I suppose it's helping in the way you intend: I could leave (and I might), I could adapt. So I'm feeling none of the fear or anxiety about AI. Just something that I think is roughly boredom.
Oh my god. This is me. If I were any better at writing, I could have written this, the author is even the same age as me (well, a year younger) and followed a similar trajectory. And a lot of what I've been feeling lately feels similar to burnout (in fact I've been calling it that), but it really isn't burnout. It's... this, whatever this is... a "fallow period" is a good term.
And I feel like an old man grumbling about things changing, but... it's not the same. I started programming in BASIC on my Tandy 1000 and went to college and learned how to build ISA cards with handwritten oscilloscope software in the Computer Engineering lab. My first job was writing firmware. I've climbed so far up the abstraction chain over a thirty year career and I guess I don't feel the same energy from writing software that first got me into this, and it's getting harder to force myself to press on.
At my first full time job in the early 2000s I was tasked with building a webscraper. We worked for law firms representing Fortune 500 companies and they wanted to know who was running "pump and dump" stock schemes on stocks using Yahoo Finance message boards.
At the time, I didn't know the LWP::Simple module existed in Perl so I ended up writing my own socket based HTTP library to pull down the posts, store them in a database etc. I loved that project as it taught me a lot about HTTP, networking, HTML, parsing and regexes.
Nowadays, I use playwright to scrape websites for thing I care about (e.g. rental prices at the Jersey Shore etc). I would never think to re-do my old HTTP library today while still loving the speed of modern automation tools.
Now, I too have felt the "but I loved coding!" sense of loss. I temper that with the above story that we will probably love what comes next too (eventually).
> I started programming when I was seven because a machine did exactly what I told it to, felt like something I could explore and ultimately know, and that felt like magic. I’m fifty now, and the magic is different, and I’m learning to sit with that.
Don't take this the wrong way but this is more of an age thing rather than a technology advancement thing.
Kids growing up nowadays that are interested in computers grow up feeling the same magic. That magic is partly derived from not truly understanding the thing you are doing and creating a mental "map" by yourself. There is nothing intrinsic to computing nowadays that makes it less magic than fiddling around with config.sys, in 50 years there will be old programmers reminiscing of "Remember when all new models were coming out every few months and we could fiddle around with the vector dimensionality and chunking length to get the best of gpt-6.2 RAG? Those were the times".
> There is nothing intrinsic to computing nowadays that makes it less magic than fiddling around with config.sys
There definitely is: the rent-seeking behavior is out of control. As a kid I could fiddle with config.sys (or rather autoexec.bat) while nowadays wrestling a file path out of my phone is a battle and the system files of my phone are kept from me.
config.sys was understandable. Now your computer has thousands (probably more) of config.sys-sized components and you are still only one person. The classic UI may improve your ability to find the components (sometimes) but can't reduce the complexity of either the components themselves or their quantity. AI makes it possible to deal with this complexity in a functional way.
Your last point is probably correct though, because AI will also allow systems to become orders of magnitude more complex still. So like the early days of the internet, these are still the fun days of AI, when the tool is overpowered compared to its uses.
> Don't take this the wrong way but this is more of an age thing rather than a technology advancement thing.
I am much younger than the poster you are replying to, but I feel much the same.
LLM are not AI, but are a great context search tool when they work.
When people first contact ML, they fool themselves into believing it is intelligent... rather than a massive plagiarism and copyright IP theft machine.
Fun is important, but people thinking zero workmanship generated content is sustainable are still in the self-delusion stage marketers promote.
https://medium.com/ideas-into-action/ikigai-the-perfect-care...
I am not going to cite how many fads I've seen cycle in popularity, but many have seen the current active cons before. A firm that takes a dollar to make a dime in revenue is by definition a con kids. =3
"The Ice King"
https://www.youtube.com/watch?v=6HVYHNTDOFs
The contrast between this and https://news.ycombinator.com/item?id=46923543 (Software engineering is back) is kind of stark. I am using frontier models to get fun technical projects done that I simply didn't have time for since my late teens. It is still possible to understand an architecture down to the hardware if you want to, but it can happen a lot faster. The specifications are queryable now. Obscure bugs that at least one person has seen in the past are seconds away instead of minutes or hours of searching. Even new bugs have extra eyes on them. I haven't written a new operating system yet but it's now a tractable problem. So is using Lean or Julia or some similar system to formally specify it. So far I've been digging into modern multithreaded cache performance which is just as fascinating as directly programming VGA and sound was in the early PC days. Linux From Scratch is still up to date. You can get FPGAs that fit in your USB port [0]. Technical depth and low-level understanding is wherever you want to look for it.
[0] https://www.crowdsupply.com/sutajio-kosagi/fomu
> Obscure bugs that at least one person has seen in the past are seconds away instead of minutes or hours of searching.
This is a huge one for me. Claude is significantly better at Googling than I am.
I was 7 in 1987, learned LOGO and C64 BASIC that year, and I relate to this article as well.
It feels as though a window is closing upon the feeling that software can be a powerful voice for the true needs of humanity. Those of us who can sense the deepest problems and implications well in advance are already rare. We are no more immune to the atrophy of forgetting than anyone.
But there is a third option beyond embrace or self-extinguish. The author even uses the word, implying that consumers wanted computers to be nothing more than an appliance.
The third option is to follow in the steps of fiction, the Butlerians of Dune, to transform general computation into bounded execution. We can go back to the metal and create a new kind of computer; one that does have a kind of permanence.
From that foundation, we can build a new kind of software, one that forces users to treat the machine as appliance.
It has never been done. Maybe it won't even work. But, I need to know. It feels meaningful and it has me writing my first compiler after 39 years of software development. It feels like fighting back.
Was this text run through LLM before posting? I recognize that writing style honestly; or did we simply speak to machines enough to now speak like machines?
Yes. This is absolutely chatgpt-speak. I see it everywhere now. It's inescapable. At least this appears to be largely human authored and have some substance, which is generally not the case when I see these LLM-isms.
I don't disagree that technology is less fun in an AI era. The question is, what other careers are out there for someone who wants to make things?
About a decade ago, I went through a career crisis where I couldn't decide what job to do - whether technology was really the best choice for my particular temperament and skills.
Law? Too cutthroat. Civil service? Very bureaucratic. Academia? Bad pay. Journalism? An industry in decline.
It is a shame, what is happening. But I still think, even with AI hollowing out the fun parts, tech remains the best job for a smart, motivated person who's willing to learn new things.
> Academia? Bad pay.
I think that bad pay is preferable to no fun. Of course, academia isn’t exactly a bed of roses either.
People get bored if they don't find real meaning in their work.
https://medium.com/ideas-into-action/ikigai-the-perfect-care...
Fact is, the tech sector is filled with folks that find zero joy in what they do, chose a career for financial reasons, and end up being miserable to everyone including themselves.
The ex-service people would call these folks entitled Shitbirds, as no matter the situation some will complain about everything. Note, everyone still does well in most large corporate settings, but some are exhausting to be around on a project. =3
The reason we don’t have the right to be lazy is because of the people who find “meaning” in toil. I do not want to work and AI is the most anti work technology in human history.
Bertrand Russel literally wrote a book called “in defense of idleness” because he knew that heavy hitters like him had to defend work abolitionism. The “work is good” crowd is why we can’t have nice things. You guys are time thief’s and ontologically evil. May all work supporters reincarnate as either durian fruits or cockroaches.
You seem very passionate about your opinions, but are you happy?
The fact remains LLM can't reach comparable human error rates without consuming 75% of the energy output of our entire local galaxy.
While I find true Neuromorphic computing topics more interesting, the emergence of the LLM "AI" true believer is deeply concerning to those that understand how they are actually built. =3
https://www.youtube.com/watch?v=ERiXDhLHxmo
https://www.youtube.com/watch?v=_zfN9wnPvU0
https://www.youtube.com/watch?v=Xx4Tpsk_fnM
I'm the exact age as the author and this post could have been written by me (if I could write). It echoes my story and sentiment exactly right down to cutting my literal baby teeth on a rubber key ZX Spectrum.
The anxiety I have that the author might not be explicitly stating is that as we look for places we add genuine value in the crevices of frontier models' shortcomings those crevices are getting more narrow by the day and a bit harder to find.
Just last night I worked with Claude and at the end of the evening I had it explain to me what we actually did. It was a "Her" (as in the movie) moment for me where the AI was now handholding me and not the other way around.
> The anxiety I have that the author might not be explicitly stating is that as we look for places we add genuine value in the crevices of frontier models' shortcomings those crevices are getting more narrow by the day and a bit harder to find.
That's exactly it. And then people say "pivot to planning / overall logic / high-level design," but how long do we have before upper management decides that AI is good enough at that stuff, too, and shows us all the door?
If they believe they can get a product that's 95% of what an experienced engineer would give them for 5% of the cost, why bother keeping the engineer around?
A blacksmith was a person that picked up chunks of carbon and heated them to they were glowing red and beat the iron to submission with a hammer in their hands.
Today iron is produced by machines in factories by the mega-tonne.
We just happened to live in the age where code when from being beaten by hand to a mass produced product.
And so the change of technology goes.
And the blacksmiths losing their jobs are not allowed to feel bad about it?
Especially anyone in their 40s or 50s who is close enough to retirement that a career shift is unappealing but far enough from retirement that a layoff now would meaningfully change that timeline or QOL. I don't blame people for feeling uneasy.
I'm probably 7 or 8 years from an easy retirement myself, so I can appreciate how that feels. Nobody really wants to feel disruption at this age, especially when they're the breadwinner for a family.
You either become a foreman operating the machines or a Luddite burning them.
No. By this logic, if they wanted to stay with the times they should have sought capital investment for their own industrial forges, joined their local lodges, climbed the ranks, lobbied their governments for loose safety regulations, and plied their workers with propaganda about how "we're in a recession and have to tighten our belts".
Think of the wonderful world we could have if everyone just got their shit together and became paper trillionaire technocrats.
The software world pretty much demanded this outcome.
Go back 10 years and post "SWE's should form labor unions"
Then watch as your post drops to [dead] and people scream "How dare you rob me of theoretical millions of dollars I'll be making".
I wonder how many of these same downvoters are now worried about getting replaced with AI.
Some of them feel bad about it and some of them refined metallurgy to build Saturn V rockets and go to space. We are very much living in the new space race. The discussion here is split 50/50 between the “Thank you! I feel the same way” folks and the “I am having the time of my life!” folks.
Blacksmiths were replaced by factories which produced deterministic products with 100% predictability.
AI can't produce code yet with 100% predictability. If that day ever arrives, the blacksmith analogy will be apt.
>with 100% predictability.
Not sure what world you're from, but lots of products get sent back to the manufacture because they break.
I'm 46 but same. I'm not quite as melancholy about it, but I do feel a lot of this.
I'm the exact same demographic as the author, just turned 50, writing code since childhood in BASIC. I'm dealing with the AI in programming issue by ignoring it.
I still enjoy the physical act of programming so I'm unsure why I should do anything that changes that. To me it's akin to asking a painter to become a photographer. Both are artists but the craft is different.
Even if the AI thing is here to stay, I think there will be room for people who program by hand for the same reason there's still room for people who paint, despite the invention of the camera.
But then, I'm somebody who doesn't even use an IDE. If I find an IDE obtrusive then I'm certain I'll find an AI agent even more so.
50 myself, and started coding with a Commodore 64, but only really picked it up seriously with the advent of open source software, and that feeling of being able to dig around any component of the system I wanted to was exhilarating.
I think that's one of the biggest things that gives me pause about AI: the fact that, if they prove to be a big productivity boost, you're beholden to huge corporations, and not just for a one-time purchase, but on an ongoing basis.
Maybe the open source models will improve, but if keeps being driven by raw compute power and big numbers, it seems to tilt things very much in favor of those with lots and lots of capital to deploy.
There's nothing "hollowed out" about directing an AI effectively, the feedback is as quick and tight as it always was. The trick is that you don't just "vibe code" and let the AI one-shot the whole thing: you should propose the change first and ask the AI about a good, detailed plan for implementing it. Then you review what the robot has proposed (which is trivial compared to revising code!) make sensible changes, ask for feedback again, and repeat. By the time the AI bot has to write actual code, it's not running on vibes anymore: it's been told exactly what to do and how to assess the result. You spend more time upfront, but a lot less on fixing the AI's mistakes.
> you should propose the change first and ask the AI about a good, detailed plan for implementing
Why ask though?
If I’m familiar with a project, more often than not, I usually have a very good idea of the code I have to write within minutes of reading the ticket. Most of the time taken is finding the impact of the change, especially with dependencies that are present in the business domain, but are not reflected in the code.
I don’t need to ask what to code. I can deduce it as easily as doing 2+2. What I’m seeking is a reason not to write it the way I envisioned it. And if those reasons are technical, it’s not often a matter of code.
Because that's how you ensure that the AI has the right idea about what to do. If the proposed plan has problems, you work with the AI to fix them before setting it to work. AI is not as smart as you, so it needs to be told how to go about doing things.
Any change that I’ve done which resulted in more than a a 10 lines diff are done with tools (copy-paste, vim-fu, refactor tools or script, snippets, code generators,…) Why would I spend time babysitting an LLM when I could have just done it myself? The purpose of automation is to lighten my workload, not to add to it.
An LLM is also a code generator. There is a scale of changes where using one is just not worthwhile (quite possibly around the 10 lines mark, as you said) but other than that, why would you want to write code yourself line-by-line that you could just generate?
Same as assembly programmers felt when C came along I guess
A lot of that magic still remains in embedded.
If vendors can't be bothered to use a C compiler from the last decade, I don't think they'll be adopting AI anytime soon.
At my work, as of 2026, we only now have a faction riled up about evangelizing clean code, OOP, and C++ design patterns. I hope the same delay keeps for all the rest of the "abstraction tower".
The issue is that AI will be creating software at whatever abstraction layer it is asked to produce. Right down to ASM maybe even machine code if someone actually wanted or needed that. Perhaps not the AI of today but given a few years I'll be quite surprised if it still can't.
If we can take a computer as powerful as today’s laptops and make it crawl because of the amount of inefficiencies in software like Teams, I’m not holding breath for embedded. If you apply the same kind of engineering principle as Anthropic, you’ll be laughed out of the room.
I too have felt these feelings (though I'm much younger than the author). I think as I've grown older I have to remind myself
1. I shouldn't be so tied to what other people think of me (craftsman, programmer, low level developer)
2. I shouldn't measure my satisfaction by comparing my work to others'. Quality still matters especially in shared systems, but my responsibility is to the standards I choose to hold, not to whether others meet them. Plus there are still community of people that still care about this (handmade network, openbsd devs, languages like Odin) that I can be part of it I want to
3. If my values are not being met either in my work or personal life I need to take ownership of that myself. The magic is still there, I just have to go looking for it
Did hardware engineers back in the 1970s-80s* think that software took the joy out of their craft? What do those engineers now think in retrospect?
*I'm picking that era because it seems to be when most electronic machines' business logic moved from hardware to software.
I found that feeling again while building a game on the EVM. All of the constraints were new and different. Solidity feels somewhere between and high and low level language, not as abstracted as most popular languages today but a solid step above writing assembly.
A lot of people started building projects like mine when the EVM was newer. Some managed to get a little bit of popularity, like Dark Forest. But most were never noticed. The crypto scene has distracted everyone from the work of tinkerers and artists who just wanted to play with a new paradigm. The whole thing became increasingly toxic.
It was like one last breath of fresh cool air before the pollution of AI tools arrived on the scene. It's a bitter sweet feeling.
I humbly submit this interview with Grady Booch (if you know, you know) talking about the "3rd golden age of software engineering - thanks to AI": https://youtu.be/OfMAtaocvJw
I feel like the conversation does a good job of couching the situation we find ourselves in.
I think one of the big distinctions between people who like building with AI and those who don't, is that the people who are pro-AI are building their own ideas, of which they have many.
The people who are anti-AI are largely building other people's ideas, for work. And they have no desire to ramp up velocity, and it's not helpful to them anyway because of bureaucratic processes that are the real bottleneck to what they're building.
Not everyone falls into these silos, of course.
I am a little older than OP. I don't think I've ever had that feeling about a programming project for work that came from someone else.
Generally, I get that feeling from work projects that I've self-initiated to solve a problem. Fortunately, I get the chance to do this a lot. With the advent of agentic coding, I am able to solve problems at a much higher rate.
Quite often, I'll still "raw dog" a solution without AI (except for doc lookups) for fun, kind of as a way to prove to myself I can still do it when the power's out.
I'm ~40ish but middle career and not in management. I envy this author, whatever joy he found in solving little puzzles and systems was extinguished in me very early in my career in an intense corporate environment. I was never one to love fussing much with code, but I do love solving system scale problems, which also involve code. I don't feel I am losing anything, the most annoying parts of code I deal with are now abstracted into human language and specs, and I can now architect/build more creatively than before. So I am happy. But, I was one of those types that never had a true passion for "code" and have meant plenty of people that do have that, and I feel for them. I worry for people that carved out being really good at programming as a niche, but you enter a point in your career where that becomes much less important than being able to execute and define requirements and understand business logic. And yea, that isn't very romantic or magical, but I find passion outside of what pays my bills, so I lost that ennui feeling a while ago.
"Over four decades I’ve been through more technology transitions than I can count. New languages, new platforms, new paradigms. CLI to GUI. Desktop to web. Web to mobile. Monoliths to microservices. Tapes, floppy discs, hard drives, SSDs. JavaScript frameworks arriving and dying like mayflies."... made me think of
I've seen things you people wouldn't believe. Attack ships on fire off the shoulder of Orion. I watched C-beams glitter in the dark near the Tannhäuser Gate. All those moments will be lost in time, like tears in rain. Time to die.
where we came from and where we're going this whole time in my career those things are kind of hard to pinpoint. Abstraction is killing us for sure. Time to market above all else. It's no wonder why software in cars, appliances and medical equipment is a factor that is killing people.
maybe we just change, honestly. i think when i were younger there was nothing to lose, time felt unlimited, no "career" to gamble with, no billion dollar idea, just learning and tinkering and playing with whatever was out there because it was cool and interesting to me. in some respects i miss that.
not sure how that relates to llms but it does become an unblocker to regain some of that "magic", but also i know to deep dive requires an investment i cannot shortcut.
the new generation of devs are already playing with things few dinosaurs will get to experience fully, having sunk decades into the systems built and afraid to let it go. some of that is good (to lean on experience) and some of it holding us back.
I prefer to see it as the automtion of the IT age.
All other professions had their time when technology came and automated things.
For example wood carvers, blacksmiths, butchers, bakers, candlestickmakers etc etc. All of those professions have been mostly taken over by machines in factories.
I view 'ai' as new machines in factories for producing code. We have reached the point where we have code factories which can produce things much more efficiently and quicker than any human can alone.
Where the professions still thrive is in the artisan market. There is always demand for hand crafted things which have been created with love and care.
I am hoping this stays true for my coding analogy. Then people who really care about making a good product will still have a market from customers who want something different from the mass produced norm.
I think the issue at the core of the analogy is that factories, traditional factories, excel at making a ton of one thing (or small variations thereof). The big productivity gains came from highly reliable, repeatable processes that do not accommodate substantial variation. This rigidity of factory production is what drives the existence of artisan work: it can always easily distinguish itself from the mass product.
This does not seem true for AI writing software. It's neither reliable nor rigid.
What assembly lines and factories did for other manufacturing processes is to make it feasable for any person to be able to make those things. In the past only very skilled professionals were able to create such things, but mechanisation and breaking down manufacturing processes into small chunks made the same things be able to be achieved by low skilled workers.
IMO that is exactly what is happening here. Ai is making coding apps possible for the normal person. Yes they will need to be supervised and monitored, just like workers in a factory. But groups of normal low skilled workers will be able to create large pieces of software via ai, whic has only ever been possible by skilled teams of professinoals before.
Yes, I think that's how it will go, like all those other industries. There will be an artisanal market, that's much smaller, where the (fewer) participants charge higher prices. So it'll (ironically?) end up being just another wealth concentrator. A few get richer doing artisanal work while most have their wage depressed and/or leave the market.
> For example wood carvers, blacksmiths, butchers, bakers, candlestickmakers etc etc.
Very, very few of those professions are thriving. Especially if we are talking true craftsmanship and not stuffing the oven with frozen pastries to create the smell and the corresponding illusion of artisinal work.
They are thriving where I live. There is a huge artisinal market for hand crafted things. There are many markets, craft centers, art fairs, regular classes from professionals teaching amateurs etc. In most rural communities I have visited it is similar.
> The feedback loop has changed. The intimacy has gone. The thing that kept me up at night for decades — the puzzle, the chase, the moment where you finally understand why something isn’t working — that’s been compressed into a prompt and a response
It's so strange to read because to me its never been more fun to make software, its especially never been easier for an individual. The boring parts are being automated so I can work on the bespoke and artistic parts. The feedback loop is getting shorter to making something nice and workable. The investigation tools for profiling and pinpointing performance bottlenecks are better than ever, where Claude is just one new part of it.
The irony is that you could still code the way you always did, where you control every pixel. Nothing is stopping you.
But you would not be able to make anything anywhere near as complex as you can with modern tools.
I turn 52 this year. I also started at 10 years old programming in a combination of AppleSoft BASIC and assembly language and typing machine code out of books so I could use Double Hires graphics since it wasn’t supported by BASIc and doing my own assembly language programming.
I stuck with C and C++ as my bread and butter from 1996-2011 with other languages in between.
I don’t miss “coding” because of AI. My vision has been larger than what I could do myself without delegating for over a decade - before LLMs.
“coding” and/or later coordinating with people (dotted line) reporting to me has been a necessary evil until a year or two ago to see my vision go to implementation.
I absolutely love this new world. For loops and while loops and if statements don’t excite me in my 50s. Seeing my vision come to life faster than I ever could before and having it well archited does.
I love talking to “the business” and solving XYProblems and getting to a solution 3x faster
I am in a very similar boat, age and experience-wise. I would like to work backward from the observation that there is no resource constraints and we're collectively hopelessly lost up the abstraction Jenga tower.
I observe that the way we taught math was not oriented on the idea that everyone would need to know trigonometric functions or how to do derivatives. I like to believe math curricula was centered around standardizing a system of thinking about maths and those of us who were serious about our educational development would all speak the same language. It was about learning a language and laying down processes that everyone else could understand. And that shaped us, and it's foolish to challenge or complain about that or, God forbid, radically change the way we teach math subjects because it damages our ability to think alike. (I know the above is probably completely idealistic verging on personal myth, but that's how I choose to look at it.)
In my opinion, we never approached software engineering the same way. We were so focused on the compiler and the type calculus, and we never taught people about what makes code valuable and robust. If I had FU money to burn today, I'd start a Mathnasium company focused around making kids into systems integrators with great soft skills and the ability to produce high quality software. I would pitch this business under the assumption that the jenga tower is going to be collapsing pretty much continuously for the next 25-50 years and civilization needs absolute unit super developers coming out of nowhere who will be able to make a small fortune helping companies dig their way out of 75 years of tech debt.
You can still have fun programming. Just sit down and write some code. Ain't nobody holding a gun to your head forcing you to use AI in your projects.
And the part of programming that wasn't your projects, whether back in the days of TPS reports and test coverage meetings, or in the age of generative AI, that bit was always kinda soul draining.
Starting code when I was 14, sold my first bit of code at 17, which was written in 6502 assembler.
40+ years later, been through many BASICs, C, C++ (CFront on onwards) and now NodeJS, and I still love writing code.
Tinkering with RPi, getting used to having a coding assistant, looking forward to having some time to work on other fun projects and getting back into C++ sooooon.
What's not to love?
Well-written and it expresses a mood, a feeling, a sense of both loss and awe. I was there too in the 8-bit era, fully understanding every byte of RAM and ROM.
The sense of nostalgia that can turn too easily into a lament is powerful and real. But for me this all came well before AI had become all consuming... It's the just the latest manifestation of the process. I knew I didn't really understand computers anymore, not in the way I used to. I still love coding and building but it's no longer central to my job or lif3. It's useful, I enjoy it but at the same time I also marvel at the future that I find myself living in. I've done things with AI that I wouldn't have dared to start for lack of time. It's amazing and transformative and I love that too.
But I will always miss the Olden Days. I think more than anything it's the nostalgia for the 8-bit era that made me enjoy Stranger Things so much. :)
Fantastic Article, well written, thoughtful. Here are a couple of my favorite quotes:
To relate to the author, I think with a lot of whats going on I feel the same about, but other parts I feel differently than they do. There appears to be a shallowness with this... yes we can build faster than ever, but so much of what we are building we should really be asking ourselves why do we have to build this at all? Its like sitting through the meeting that could have been an email, or using hand tools for 3 hours because the power tool purchase/rental is just obscenely expensive for the ~20min you need it.I think the true genuinely-love-programming type of people will increasingly have to do what so many other people do, and that's separation of work and personal enjoyment. You might have to AI-architect your code at work, and hand code your toy projects on the weekend.
I'm roughly the same (started at 9, currently 48), but programming hasn't really changed for me. What's changed is me having to have pointless arguments with people who obviously have no clue what they're talking about but feel qualified either because:
a) They asked an LLM
b) "This is what all our competitors are doing"
c) They saw a video on Youtube by some big influencer
d) [...insert any other absurd reason...]
True story:
In one of our recent Enterprise Architecture meetings, I was lamenting the lack of a plan to deal with our massive tech debt, and used an example of a 5000 line regulatory reporting stored procedure written 10 years ago that noone understood. I was told my complaint was irrelevant because I could just dump it into ChatGPT and it would explain it to me. These are words uttered by a so-called Senior Developer, in an Enterprise Architecture meeting.
Was he entirely wrong? Have you tried to dump the stored proc into a frontier model and ask it to refactor? You'd probably have neat 20 stored procs with well laid out logic in minutes.
I wouldn't keep a ball of mud just because LLMs can usually make sense of them but to refactor such code debt is becoming increasingly trivial.
> Was he entirely wrong?
Yes. I mean... of course he was?. Firstly, I had already gone through this process with multiple LLMs, from various perspectives, including using Deep Research models to find out if any other businesses faced similar issues, and/or if products existed that could help with this. That lead me down a rabbit hole of data science products related to regulatory reporting of a completely different nature which was effectively useless. tl;dr: Virtually all LLMs - after understanding the context - recommended us doing thing we had already been urging the business to do - hire a Technical BA with experience in this field. And yes, that's what we ended up doing.
Now, give you some ideas about why his idea was obviously absurd:
- He had never seen the SP
- He didn't understand anything about regulatory reporting
- He didn't understand anything about financial derivatives
- He didn't understand the difference between Transact SQL and ANSI SQL
- No consideration given to IP
- etc etc
Those are the basics. Let's jump a little bit into the detail. Here's a rough snippet of what the SP looks like:
Yes, that's a typical column name that has rotted over time, so noone even knows if it's still correct. Yes, those are typical CASE statements (170+ of them at last count, and no, they are not all equal or symmetric).So... you're not just dealing with incredibly unwieldy and non-standard SQL (omitted), noone really understands the business rules either.
So again... yes he was entirely wrong. There is nothing "trivial" about refactoring things that noone understands.
I'm a few years behind you. I got started on my uncle's handed down vic 20 in the late 80s.
The culture change in tech has been the toughest part for me. I miss the combination of curiosity, optimism, creativity, and even the chaos that came with it. Nowadays it's much harder to find organizations like that.
> I wrote my first line of code in 1983. I was seven years old, typing BASIC into a machine that had less processing power than the chip in your washing machine
I think there may be a counterpoint hiding in plain sight here: back in 1983 the washing machine didn't have a chip in it. Now there are more low-level embedded CPUs and microcontrollers to develop for than before, but maybe it's all the same now. Unfathomable levels of abstraction, uniformly applied by language models?
I am younger than the author but damn this somehow hit me hard. I do remember growing up as a kid with a 486...
Yeah I could use Cursor or whatever but I don't, I like writing code. I guess that makes me a luddite or something, although I still develop agents. I enjoy architecting things (I don't consider myself an architect) I'm talking about my hobby hardware projects.
I'm 55 and I started at age 13 on a TI-99/4A, then progressed through Commodore 64, Amiga 2000, an Amiga XT Sidecar, then a real XT, and on and on. DOS, Windows, Unix, the first Linux. I ran a tiny BBS and felt so excited when I heard the modem singing from someone dialing in. The first time I "logged into the Internet" was to a Linux prompt. Gopher was still a bigger thing than the nascent World-Wide Web.
The author is right. The magic has faded. It's sad. I'm still excited about what's possible, but it'll never create that same sense of awe, that knowledge that you can own the entire system from the power coming from the wall to the pixels on your screen.
Similar story for myself. It was long and tedious for my mental model to go from Basic, to Pascal, to C, and finally to ASM as a teen.
My recent experience is the opposite. With LLMs, I'm able to delve into the deepest parts of code and systems I never had time to learn. LLMs will get you to the 80% pretty quick - compiles and sometimes even runs.
I know exactly how you feel. I don't know how many hours I sat in front of this debugger (https://www.jasik.com) poking around and trying to learn everything at a lower level. Now its so different.
Yeah. Different is the word. In many ways it’s just another abstraction but we’re not machines and this, to me at least, just gives a very different feel.
Cool, at 7? I started at 9 and I'm 53 now. And Claude does all the things. Need to get adjusted to that though. Still not there.
Last year I found out that I always was a creator, not a coder.
Same, but it changed when I was 17 and again when I was 27 and then 37 and so on. It has always been changing dramatically, but this latest leap is just so incredibly different that it seems unique.
Abstractions can take away but many add tremendous value.
For example, the author has coded for their entire career on silicon-based CPUs but never had to deal with the shittiness of wire-wrapped memory, where a bit-flip might happen in one place because of a manufacturing defect and good luck tracking that down. Ever since lithography and CPU packaging, the CPU is protected from the elements and its thermal limits are well known and computed ahead of time and those limits baked into thermal management so it doesn’t melt but still goes as fast as we understand to be possible for its size, and we make billions of these every day and have done for over 50 years.
Moving up the stack you can move your mouse “just so” and click, no need to bit-twiddle the USB port (and we can talk about USB negotiation or many other things that happen on the way) and your click gets translated into an action and you can do this hundreds of times a day without disturbing your flow.
Or javascript jit compilation, where the js engine watches code run and emits faster versions of it that make assumptions about types of variables - with escape hatches if the code stops behaving predictably so you don’t get confusing bugs that only happen if the browser jitted some code. Python has something similar. Thanks to these jit engines you can write ergonomic code that in the typical scenario is fast enough for your users and gets faster with each new language release, with no code changes.
Lets talk about the decades of research that went into autoregressive transformer models, instruction tuning, and RLHF, and then chat harnesses. Type to a model and get a response back, because behind the scenes your message is prefixed with “User: “, triggering latent capabilities in the model to hold its end of a conversation. Scale that up and call it a “low key research preview” and you have ChatGPT. Wildly simple idea, massive implications.
These abstractions take you further from the machine and yet despite that they were adopted en masse. You have to account for the ruthless competition out there - each one would’ve been eliminated if they hadn’t proven to be worth something.
You’ll never understand the whole machine so just work at the level you’re comfortable with and peer behind the curtain if and when you need (eg. when optimizing or debugging).
Or to take a moment to marvel.
Are you me?
I'm 49.... Started at 12... In the same boat
First 286 machine had a CMOS battery that was loose so I had to figure that out to make it boot into ms-dos
This time it does feel different and while I'm using them ai more than ever, it feels soulless and empty even when I 'ship' something
> the VGA Mode X tricks in Doom
Doom does not use mode-X :P ! It uses mode-Y.
That being said as a 47 years old having given 40 years to this thing as well, I can relate to the feeling.
I'm 47 and excited to live in a time of the moat important innovation since the printing press.
Is there some magic lost also when using AI to write your blog post?
I miss human writing. I miss the different voices.
Seriously I thought I was going crazy with this. So many "it's not just x it's y". Short punchy sentences. Emdashes galore.
I'm 43. Took a year or so off from contracting after being flat out for years without taking any breaks, just poked around with some personal projects, did some stuff for my wife's company, petitioned the NHS to fix some stuff. Used Claude Code for much of it. Travelled a bit too.
I feel like I turned around and there seem to be no jobs now (500+ applications deep is a lot when you've always been given the first role you'd applied to) unless you have 2+ years commercial AI experience, which I don't, or perhaps want to sit in a SOC, which I don't. It's like a whole industry just disappeared while I had my back turned.
I looked at Java in Google Trends the other day, it doesn't feel like it was that long ago that people were bemoaning how abstracted that was, but it was everywhere. It doesn't seem to be anymore. I've tried telling myself that maybe it's because people are using LLMs to code, so it's not being searched for, but I think the game's probably up, we're in a different era now.
Not sure what I'm going to do for the next 20 years. I'm looking at getting a motorbike licence just to keep busy, but that won't pay the bills.
I’m 45 and contracted for over a decade before switching to product development. I used to still get inquiries from former customers, mainly for Java and Android work. But since about two years, it’s completely dried up. Anecdotally I’ve been hearing from friends who are still in the contracting/freelancing business that things are very tough right now. It makes sense to me, contractors are usually the first thing businesses cut when they’re either lowering their spending or becoming more efficient themselves.
I don't know what these people from our now traditional daily lamentation session are coding where Claude can do all the work for them just with a few prompts and minimal reviews.
Claude is a godsend to me, but fuck, it is sometimes dumb as door, loves to create regressions, is a fucking terrible designer. Small, tiny changes? Those are actually the worse, it is easy for claude, on the first setback, decides to burn the whole world and start from zero again. Not to mention when it gets stuck in an eternal loop where it increasingly degenerates the code.
If I care about what I deliver, I have to actively participate in coding.
I'm 57 and wrote my first line of BASIC in 1980, so while I can still chime in on this specific demographic I feel that I ought to. So im like this guy, but like a lot of other people in my specific demographic we aren't writing these long melancholy blog posts about AI because it's not that big of a deal. As an OSS maintainer most of my work is a lot of boring slog adding features to libraries to suit new features in upstream dependencies, nitpicky things people point out, new docs, tons of tedium. Claude helps a ton with all of that. no way is Claude doing the real architectural puzzle stuff, that's still fully on me! I can just use Claude to help implement it. It's like the ultimate junior programmer assistant. It's certainly a new, different and unique experience in one's programming career but it really feels like another tool, like an autocomplete or code refactoring tool that is just a lot better, with similar caveats. I mean in my career, I've had to battle the whole time people who don't "get" source code control (starting with me), who don't "get" IDEs (starting with me), people who dont "get" distributed version control (same), people who don't "get" ORMs (oh yes, same for me though this one I took much more dramatic steps to appreciate them), people who don't "get" code formatters, now we're battling people who don't "get" LLMs used for coding, in that sense the whole thing doesnt feel like that novel of a situation.
it's the LLMs that are spitting out fake photos and videos and generating lots of shitty graphics for local businesses, that's where I'm still wielding a pitchfork...
It'd be more strange if the thing you learned 43 years ago was exactly the same today. We should expect change. When that change is positive we call it progress.
In the grand scheme of things it wouldn’t actually be that strange: generations and generations of humans were mostly farmers and mostly did the same thing as their parents. Of course technology developed but lots of people did the same job with the same methods their whole lives.
But everybody on this site lived through the first half of a logistic curve so that perspective seems strange to us.
I think more than ever programmers need jobs where performance matters and the naive way the AI does things doesn't cut it. When no one cares about things other than correctness your job turns into AI Slop. The good news right now is that AI tends to produce things that AI struggles to do well with so large scale projects often descend into crap. You can write a C-compiler for $20,000 with an explosive stack of agents, but that C-compiler isn't anywhere close to efficient or performant.
As model costs come down that $20,000 will become a viable number for doing entirely AI-generate coding. So more than ever you don't want to be doing work that the AI is good enough at. Either jobs where performance matters or being able to code the stack of agents needed to produce high quality code in an application context.
I wonder what other “crevices” (as the author put it) exist.
Another commentor mentioned embedded, and after a brief phase of dabbling in that, mainly with nRF5x micros, I tend to agree. Less training data and obtuse tooling.
A bit younger, and exact opposite. Probably the most excited I've ever been about the state of development!
'It’s not a “back in my day” piece.'
That's exactly what it is.
"They’re writing TypeScript that compiles to JavaScript that runs in a V8 engine written in C++ that’s making system calls to an OS kernel that’s scheduling threads across cores they’ve never thought about, hitting RAM through a memory controller with caching layers they couldn’t diagram, all while npm pulls in 400 packages they’ve never read a line of."
and they still call themselves 'full stack developers' :eyeroll:
>But sure. AI is the moment they lost track of what’s happening. The abstraction ship sailed decades ago.
Bullshit. While abstraction has increased over time, AI is no mere incremental change. And the almost natural language interaction with an agent is not the same as Typescript over assembly (not to mention you could very well right C or Rust and the like, and know most of the details of the machine by heart, and no, microcode and low level abstractions are not a real counter-argument to that). Even less so if agents turn autonomous and you just herd them onto completion.
This LLM stuff is a little weird. Previously we had Python which was pretty close to pseudocode but you could run it directly. Now, these LLMs are one step more abstract, but their outputs aren’t runnable directly, they produce possibly incorrect code-like-text. Actually this seems like good news for programmers since you have to read the code in the lower-level language that gets produced.
It's not like it's changing by itself, you can always opt out of the slop race and scratch your itches instead.
https://gitlab.com/codr7/rem
The irony of these "My craft is dead" posts is that they consistently, heavily leverage AI for their writing. So you're crying about losing one craft to AI while using AI to kill another. It's disingenuous. And yes it is so damn obvious.
If you bothered to read it you’d find that I am embracing the tools and I still feel there is craft. It’s just different.
But snark away. It’s lazy. And yes it is so damn tedious.
I think the Oxide computer LLM guidelines are wise on this front:
> Finally, LLM-generated prose undermines a social contract of sorts: absent LLMs, it is presumed that of the reader and the writer, it is the writer that has undertaken the greater intellectual exertion. (That is, it is more work to write than to read!) For the reader, this is important: should they struggle with an idea, they can reasonably assume that the writer themselves understands it — and it is the least a reader can do to labor to make sense of it.
https://rfd.shared.oxide.computer/rfd/0576#_llms_as_writers
The heavy use of LLMs in writing makes people rightfully distrustful that they should put the time in to try to read what's written there.
Using LLMs for coding is different in many ways from writing, because the proof is more there in the pudding - you can run it, you can test it, etc. But the writing _is_ the writing, and the only way to know it's correct is to put in the work.
That doesn't mean you didn't put in the work! But I think it's why people are distrustful and have a bit of an allergic reaction to LLM-generated writing.
Speaking directly, if I catch the scent of ChatGPT, it's over.
People put out AI text, primarily, to run hustles.
So its writing style is a kind of internet version of "talking like a used car salesman".
With some people that's fine, but anyone with a healthy epistemic immune system is not going to listen to you.
If you want to save a few minutes, you'll just have to accept that.
What's your target false positive rate?
I mean, obviously you can't know your actual error rates, but it seems useful to estimate a number for this and to have a rough intuition for what your target rate is.
Did chatGPT write this response?
This is how LLMs poison the discourse.
I agree with that for programming, but not for writing. The stylistic tics are obtrusive and annoying, and make for bad writing. I think I'm sympathetic to the argument this piece is making, but I couldn't make myself slog through the LinkedIn-bot prose.
"But snark away. It’s lazy. And yes it is so damn tedious."
Looks like this comment is embracing the tools too?
I'd take cheap snark over something somebody didn't bother to write, but expect us to read.
Why should anyone bother to read what nobody wrote?
AI?
Having an LLM write your blog posts is also lazy, and it's damn tedious to read.
I felt the same. I resonate with the message, but it really rings hollow with so much AI directing.
I'd wish people would stop doing that. AI writing isn't even particularly good. Its not like it makes you into Dostoevsky, it just sloppifies your writing with the same lame mannerisms ("wasn't just X — it was Y"), the same short paragraphs, the same ems.
The author admits that they used AI but I found it not that obvious. What are telltale signs in this case? While the writing style is a little bit over-stylized (exactly three examples in a sentence, Blade Runner reference), I might write in a similar style about a topic that im very emotional about. The actual content feels authentic to me.
(1) The pattern "It's not just a X---It's a Y" is super common in LLM-generated text for some reason. Complete with em dash. (I like em dashes and I wish LLMs weren't ruining them for the rest of us)
"Upgrading your CPU wasn’t a spec sheet exercise — it was transformative."
"You weren’t just a user. You were a systems engineer by necessity."
"The tinkerer spirit didn’t die of natural causes — it was bought out and put to work optimising ad clicks."
And in general a lot of "It's not <alternative>, it's <something else>", with or without an em dash:
"But it wasn’t just the craft that changed. The promise changed."
it's really verbose. One of those in a piece might be eye-catching and make someone think, but an entire blog post made up of them is _tiresome_.
(2) Phrasing like this seems to come out of LLMs a lot, particularly ChatGPT:
"I don’t want to be dishonest about this. "
(3) Lots of use of very short catch sentences / almost sentence fragments to try to "punch up" the writing. Look at all of the paragraphs after the first in the section "The era that made me":
"These weren’t just products. " (start of a paragraph)
"And the software side matched." (next P)
"Then it professionalised."
"But it wasn’t just the craft that changed."
"But I adapted." (a few paragraphs after the previous one)
And .. more. It's like the LLM latched on to things that were locally "interesting" writing, but applies them globally, turning the entire thing into a soup of "ah-ha! hey! here!" completely ignorant of the terrible harm it does to the narrative structure and global readability of the piece.
I'm weird about this, I choose to use AI to get feedback on my writing, but refuse to just copy and paste the AIs words. I only do it if its a short work email and I really dont care about its short lived lifespan, if its supposed to be an email where the discussion continues, then I refine it. I can write a LOT. If HN has edit count logs, I've probably got the high score.
Imagine if people were complex creatures, feeling different emotions for different things, shocking right?
I can hate LLMs for killing my craft while simultaneously using it to write a "happy birthday" message for a relative I hate or some corpo speak.
This is not either of those. This is the equivalent of a eulogy to a passion and a craft. Using an LLM to write it: entire sections, headers, sentences - is an insult to the craft.
The post in the same vain, "We mourn our craft", did a much better job at this communicating the point without the AI influence.
Fair enough, agree on your second paragraph.
At least then you’re being honest about you hating your intended audience, and not proudly posting the slop vomited forth from your algorithmic garbage machine as if it were something that deserved the time, thought and consideration of your equals.
“... when I was 7. I'm 50 now and the thing I loved has changed”
Welcome to the human condition, my friend. The good news is that a plurality of novels, TV shows, country songs, etc. can provide empathy for and insight into your experience.
There's 3-4 of these posts a day - why don't people spend more time hand-building things for fun in their free time? That's what led a lot of us to this career path to start with. I have a solid mix of hand-code and AI-assisted projects in my free time.
>>The machines I fell in love with became instruments of surveillance and extraction.
Surveillance and Extraction
"We were promised flying cars", and what we got was "investors" running the industry off the cliff into cheap ways to extract money from people instead of real innovation.
> I started programming when I was seven because a machine did exactly what I told it to
What a poetic ending. So beautiful! And true, in my experience.
yeah coding is a lot more fun and useful now
It seems fun must be subjective. It seems less fun than ever to me.
At least parts of this were written with AI
I'm 57. I was there when the ZX81 came out.
I had my first paid programming job when I was 11, writing a database for the guy that we rented our pirate VHS tapes from.
AI is great.
This isn't new. It's the same feeling the first commercial programmers had working in assembly, or machine code, once compilers became available. Ultimately I think even Mel Kaye forsook being able to handpick memory locations for optimum drum access before his retirement, in favor of being able to build vastly more complex software than before.
AI has just vastly extended your reach. No sense crying about it. It is literally foolish to lament the evolution of our field into something more.
As someone who has always enjoyed designing things, but was never really into PUZZLES, I always felt like an outsider in the programming domain. People around me really enjoyed the "fun" of programming, whereas I was more interested in the Engineering of the thing - balancing tradeoffs until within acceptable margins and then actually calling it "DONE". People around me rarely called things "done", they rewrote it and rewrote it so that it kept satisfying their need for puzzle-solving (today, it's Ruby, tomorrow, it's rewritten in Scala, and the day after that, it's Golang or Zig!)
I feel that LLMs have finally put the ball in MY court. I feel sorry for the others, but you can always find puzzles in the toy section of the bookstore.
Programming is dead. In the last 4 days I've done 2 months of work. The future is finally here.
Bad times to be a programmer. Start learning business.
Don't program as a career, but am also 50 and programming since TRS-80. AI has transformed this era, and I LOVE IT! I can focus on making and not APIs or syntax or all of the bootstrapping.
I wondered when I would see someone else mention TRS-80. Recall those days like yesterday "Hey Mah! It works!"
Professional development is changing dramatically. Nothing stops anyone from coding "the old way," though. Your hobby project remains yours, exactly the way you want it. Your professional project, on the other hand, was never about you in the first place. It's always about the customer/audience/user, period full stop.
I have the opposite take. There’s nothing stopping you from jumping into any component to polish things up. You can code whatever you wish. And AI takes away nearly all of the drudgery : boilerplate, test cases, inspecting poor documentation, absurd tooling.
It also lets me focus more on improving things since I feel more liberated to scrap low quality components. I’m much braver to take on large refactors now – things that would have taken days now take minutes.
In many ways AI has made up for my growing lack of patience and inability to stay on task until 3am.
> In many ways AI has made up for my growing lack of patience and inability to stay on task until 3am.
That is called...programming.
Please stop upvoting these posts. We have gotten to the point where both the front page and new page is polluted with these laments
It’s literally the same argument over and over and it’s the same comments over and over and over
HN will either get back to interesting stuff or simply turn into a support group for aging “coders” that refuse to adapt
I’m going to start flagging these as spam
The article talks about human vs technology and the loss of connection between creation, intent, ownership, and control. Don't be condescending.
…like every other post of this kind
same bud.
maybe that just means it's a maturing field and we gotta adapt?
yes, the promise has changed, but you still gotta do it for the love of the game. anything else doesnt work.
11 and now 45. I am still interested in it, but I feel like in my 20s I would get a dopamine rush when a row showed up in a database. In my 30s I would get that only if a message passed through a system and updated on-screen analytics within 10 seconds. Thank god for LLMs because all of it became extremely boring, I can't stand having to get these little milestones each new company or each new product I'm working on. At least with LLMs the dopamine hit comes from being in awe of the code that gets generated and realizing it found every model, every messaging system interface, every API, and figuring out how to make it backwards compatible, updating the UI - something that would take half a day, now in 5 minutes or less.
I’m 50 too and I’ve complained and yearned about the “old” days too, a lot of this is nostalgia as we reminisce about periods of time in our youth when we had the exuberance and time to play and build with technology of our own time
Working in AI startups strangely enough I see a lot of the same spirit of play and creativity applied to LLM based tools - I mean what is OpenClaw but a fun experiment
Those kids these days are going to reminisce about the early days of AI when prompts would be handwritten and LLMs would hallucinate
I’m not really sure 1983, 1993 or 2003 really was that gold of age but we look at it with rose colored glasses
Old Man Yells at Clouds
I'd feel the same when I was younger. Over time I've realized that they are the lucky ones. You too, if you're lucky, will one day be an old man doing old man things.
Old Man Yells at Claude