Born Against, or why hobby programming communities are against LLM usage
blog.fogus.me
blog.fogus.me
1) You figure out what problem to solve.
2) You figure out HOW to solve the problem.
3) You actually implement the solution.
4) You see the solution work, for yourself.
5) You ship/deploy/publish the program. This means you see people be happy users and/or you get paid for it and so on.
If you're an entrepreneur type, you probably enjoy the first and last steps most, and you see steps 2-4 as mostly a chore. If you're a tinkerer, you don't care much for 1 and 5, and you see 2-4 as the whole point of programming. I'm a tinkerer. I'd be happy to just write code and throw it away. Coding is like solving sudokus. I could skip steps 1 and 5 forever. I don't ever need to show any code to anyone. In fact, most of the time when programming I do steps 2 and 3 and even skip 4. I don't even finish! I work weeks on something until I lose interest, and I know that in order to even run it, it would be several more weeks. A PoC is enough. Or just a half one. It's just code-to-structure-thoughts, not to create anything finished.
The 5 phases look kind of symmetric. The outermost layer (1 and 5) are the entrepreneurial steps. If you're a product owner or CEO, you might work strictly at steps 1,5. Then steps 2-4 are the managerial/architectural steps. If you're a very senior IC at a large company, you might work at this level, without actually doing much coding. Only the inner most step (3) is the manual creation of source code. Even though it's 5 different phases, it's just "3 layers" of programming.
The problem as I see it is that I enjoy step 3. And LLMs are good at step 3 almost exclusively. So they just pick the best bit of this dish, and leave me with the rest.
If you're an entrepreneurial type, the LLM appears to take the _worst_ bit of the work from you. Great.
I have about three modes, working with LLMs:
A) Just use them for (2), do (3) and (4) myself.
B) Just use them for (2) and (3), but with very fine grained instructions - to the point where it barely saves me time.
C) Use them for (2), (3) and (4), but with this workflow, it quickly degrades into slop for me.
I still get the best results with (A), but for a quick PoC, (C) is hard to beat.
What's hilarious (or sad) is that I see a lot of people throw them at (1) as well, which seems to have a high chance of inducing AI psychosis and creating software nobody understands, let alone needs.
Why is it always hours to consult docs? I find most answers within minutes (including if I have to read code) and it’s only take a bit longer when I want to understand the why’s.
The key is to make sure you have the right questions. I see people struggle with that where they only know something wrong but they can’t put it in words.
I find I get the best results when I just use it for (3), and sometimes a bit for (2). Going in with at least an idea of how to make it work has the best chance of ending up with a good changeset. And if you already know how to make it work, don't leave it to chance. Tell the LLM what to do, then let it do it.
To carry forward the sudoku analogy, I care about figuring out which numbers go where, and whether I was correct, I don't care about the mechanical part of actually penciling-in the answers (coding).
So I do 3 because coding is like writing English for me (which is my 3rd language). I don’t care much for it, because it’s neither hard no easy. It’s just mechanical.
Even if you deeply into tinkering, 1 can be an insanely exciting and rewarding step: for example "how could an abstraction look like in which 2-4 become trivial special cases for many classes of programming problems?".
In my opinion the difference between "entrepreneur type" and "programmer type" (and yes: it can happen that both of these types work in the reciprocal job) is rather:
- People of the programmer type see solving problems as their primary goal. Satisfying customers just serves the purpose that these people pay money so that they can continue working on solving problems.
- People of the entrepreneur see satisfying customers as their primary goal. Programming is just a means to an end.
If the answer is "no" you are probably more entrepreneurial than tinkerer. But it's a spectrum.
This helps differentiate between coding as an end and coding as a means to socializing.
Yes, but I think there's really two different definitions of point 1. It's "figure out what non-software problem you can solve with software" vs "figure out what technical problem to solve".
As the tinkering type, I also find 1 very rewarding but using the latter definition. Exploring a sizable code base and finding the part that, if simplified, cascades to a lot more simplification, that's very exciting. Figuring out what problem some companies or people have that could be solved? Very hard, very unexciting, not rewarding.
In my opinion these variants are not very different:
For example keep in mind that physics is basically a descipription of the software that runs reality, which automatically gives you are huge portfolio of potential problems. :-)
For mechanical engineering, observe that the advances in 3D printing and home CNC milling mean that a lot of problems that were previously in the domain of mechanical engineering now become partly software problems.
Or for more "human" problems: If you are very trained in mathematics, you start to see mathematical structures in a lot of "more human" problems (or areas that are not associated with "software") basically all the time. If you find these mathematical structures that you see interesting, it is often possible to turn them into software.
Thus, "figure out what non-software problem you can solve with software" vs "figure out what technical problem to solve" is rather a litle bit like "in mathematics: do you prefer analysis or algebra?" - yes, they (often) represent a different style of thinking and many mathematicians do have their preferences, but in the end both analysis and algebra is mathematics.
--
So, I disagree with your point
> Figuring out what problem some companies or people have that could be solved? Very hard, very unexciting, not rewarding.
In my opinion the boring part is not finding or solving problems that companies or people have. As I wrote: after some time, you often start to see highly interesting, often very novel mathematical patterns in these problems (of course not always, but it is not an uncommon situation). Why are these patterns often so novel and interesting? A heuristic answer in my opinion is: if they weren't, someone would typically already have implemented them in software. :-)
The really annoying part is rather convincing the people/company of your solution (which can be considered to be part of step "5) You ship/deploy/publish the program. This means you see people be happy users and/or you get paid for it and so on.").
So possibly there are other classes of people, not just either enterpreneur or thinkerer. Thou not sure what this combination is, maybe something more like R&D.
And LLMs give a lot of value here. They allow to quickly do steps 2-3 - "hard and tedious" stuff - to confirm that the thing, in fact, became solvable.
You can't go really hard unless you can come up with something people weren't expecting, but when they discover it they want it. You can't just sit around waiting for people to want things properly. That's extremely passive.
The point being, the idea of using LLMs to find what people already passively want is itself flawed, even if you like marketing and selling to people. It's a really fundamental problem with the idea.
Now there is a plagiarism machine built on top of decades of this work that seemingly solves 3 and promises to solve 2 (badly, but usually this matters little), the programmers that (indirectly) helped build it are told they're fools for doing so, and threatened by losing their status if they don't adapt to the "entrepreneurial way" (well, the full promise is "you'll lose it anyway, but if you cooperate in stuffing sama's pockets then maybe you get to keep it a bit longer").
So what really surprises me is not that people are upset but that so few are.
Carpenters, hatters, cobblers, miners, weavers, calculators, etc...
We can get mad about it, but it just is what it is. Ultimately for most people the loss of these jobs are a net-benefit because it's the result of better or cheaper alternatives.
The thing that worries me more specifically about AI is that it has the potential not to just disrupt an industry, or handful of industries, but in theory everything.
If machines are faster, strong, and more intelligent than humans, then you don't need human workers for anything. In the same way machines suddenly becoming faster and stronger than horses meant that the role of the horse in labour dropped rapidly during the industrial revolution.
It's not clear how civilisation functions if humans become effectively useless and the only thing of value is one's ownership of wealth producing assets.
Still, there's really no point in thinking about how we stop this. Humans are humans and we'll do the stupid thing then try to deal with the consequences after the fact. All regular people should be doing now is prepping for this outcome.
You can get a hint of this by looking at the developing countries with high income inequality. Few rich people who typically own all the countries wealth (and access to it's natural resources) while everyone else lives in a slum.
California is a good example.
The US and Europe might have lost a lot of craftsmanship knowledge at a social level. Yet, what is left is so far ahead of what has ever existed in countries in South America for example. And China has gained a lot. Witness that youtuber trying to make grill brushes in the US. These differences have an impact in terms of opportunities and competition.
It's easy to sit at home, watch documentaries and picture history progressing along logical lines and imagine that the "industrial revolution" was this massive shift from craftsmanship to automation. Look closer and you'll see that what happened in the UK in the 1700s was in fact enabled by a long and deep tradition of craftsmanship that only got enriched by the rise of new power sources. Craftsmanship today is vastly more advanced than it ever was before industry.
And there is a lot to lose in the decline of craftsmanship, and it is by no means some sort of historical inevitability.
When you really start digging deep into history you realize most arguments of logical trends therein are mostly just misconceptions.
I suspect the horses are healthier and stronger than in the 1700s too. I'm not sure I understand your point?
I'd agree that the niche of extremely high quality, bespoke craftsmanship still exists, and that in some cases jobs were augmented by technology rather than entirely replaced, but the point still stands that the work these people used to do has been automated and disrupted by technology.
The fact some people upskilled or reskilled as a response is analogous to coders today. All I'd question is the extent of the disruption we'll see as a result of AI – I worry that this time things will change much more quickly and there may not be plenty of other jobs for people to upskill / reskill in to.
But if that's not true then perhaps whatever software engineering exists in the future will be far more complex than what software engineers have done for the last few decades.
Horses? I doubt it. They're still infamously fragile creatures today, even with penicillin and vaccinations existing; similarly for strength, not much call for shires and they're now on an "at risk" conservation list: https://en.wikipedia.org/wiki/Shire_horse -> https://www.rbst.org.uk/watchlist-breed/shire-horse/
Possibly also fewer horses in total, though I don't know any stats for that far back and may be over-estimating based on the stats I do have for c. 1900.
Of course the costs are higher too. Good horses in the above category can go for north of $100k.
Every random business has some machine that puts caps on bottles or something, and these things are ridiculously better designed, engineered, and fabricated than ever before, because the knowledge and tools available to the designer and fabricator are better than ever before.
You are surrounded by things that are insanely well made by all previous standards. Even the crappy stuff that you don't think of as well-made, because it is junk compared to everything else, still has many aspects that are almost magic.
Horse population today is in line with the long term average. There was a freak bump from 1850-1920.
Think about this question: is it harder today or 200 years ago for the average person to learn and ride horses?
Same goes for craftsmanship: is it harder today or 200 years ago to find some highly skilled craftsman to do something a random people or organization might fancy getting done?
And software. We may be in the freak bump of software engineers. Seven digits seems like a good guess. Maybe we will revert to a long term average proportion of people with affinity for math and logic. But you bet your horse (heh) that in 200 years it will be far easier for the average person to find some highly skilled software craftsman to do some niche work than it has ever been. Average quality of corporate slop generated by flesh or silicon drones notwithstanding.
Society and the economy will go on growing more complex. Whatever anyone can picture or design in their heads about the reality of things can never approach the real world complexity and variety. And by whatever metric you may choose, the complexity and variety of skills and capabilities in human civilization will go on growing.
Real craftsmanship, the kind that leaves you in awe and feels like an exceptional display of human capacity, seems to be relegated, from my point of view, to a minority of well-off hobbyists who are already free from work and can do what they want without worrying about paying bills.
I’ve felt the same frustration doing excellent quality work on a 2nd or 3rd home that is empty most of the year.
yes, but no. Many craft skills used in jewelry, architecture and others are lost. For example, some goldsmithing techniques well-known in the 1700s are not reproducible today.
source: trained goldsmith
There has never been as much craftsmanship as today. The idea that industry destroyed craftsmanship is a caricature. Modern industry was built on top of deep craftsmanship traditions and only enriched them.
The fact that some things that we knew how to do centuries ago we no longer know how to do is by no means evidence to the contrary.
AI will not give us that result. It will give us shitty products and they will be far more costly. Not only will prices for consumers continue to rise regardless of how much money corporations save by using AI (because they'll just stuff their pockets with that cash) but the true costs of AI are largely externalized and include things like massive environmental damage, further concentration of wealth/power, the homogenization of culture, and the dumbing down of society. AI means paying more and getting less.
I think there's some progress, but I think the hype is so high that people don't have a clear view of what's realistic and what's not. IE, if you were to believe Amodei back in 2025 you'd believe none of us would even have jobs to complain about right now.. which hasn't happened. Altman is already declaring the singularity. It's unhinged. Anyone that says this is going to cure cancer or solve all problems is just living in a delusion.
I think back in the 90s when I was a teenager I thought it to be a good thing that there is a rise in status for a class of people characterized by clear thinking.
The older I get, I realize some people never change, will always have the same backwards or harmful thinking, and I avoid them instead of just accepting it.
Sadly, it seems so, even if the effects of the jobs on daily life differ a great deal sometimes. The world would succumb faster if we had no garbage collectors, yet they get less pay than a programmer who the world could easily survive without? Makes no sense to me.
"Programmers aren't really paid, they sell services at market prices." — I guess nobody gets paid?
As a farmer myself, when I give someone something I produced and they give me back something in return, I consider myself to be paid. If I did not receive much value in return for what I gave, I'd reasonably consider myself to have not been paid well. I have never met a farmer that would consider that usage unusual. Payment refers to a transaction within the realm of a human activity. The good or service that is being traded between humans is immaterial. While you are free to define words as you see fit, your assertion doesn't really reflect how this is actually used in practice, generally. The rest of the earlier comment makes it clear that the commenter understands how a farm business works.
</sarcasm>
One interpretation, which is what I suspect most would go to, is where one's income is primarily derived from owning stock. So, in other words, old, retired people. But I don't think anyone here was thinking about the retired. The discussion was about people actively producing food and software.
Another possible interpretation is about impact. If those stocks didn't exist, a lot of the capital wouldn't have been able to be created, and thus there wouldn't have been jobs to go with it; meaning that owning the capital makes it one's main function as without it one wouldn't have a job to go to. But I can already hear you furiously typing how that isn't a satisfactory take. I'd agree.
Maybe it's about attention? Someone who actively watches over to their capital would be seen as it being their main function, while someone who leaves their capital ownership up to a financial advisor so they don't have to think about it wouldn't. But then we're back to the computer. Most programmers watch over that capital, if nothing else. That doesn't really work either.
The only remaining interpretation I can come up, given the connotations often associated with landlord, is a group of people not like by another group of people. But that doesn't have anything to do with capital at all, so... Consider me stumped on what this has to do with the discussion (if we are to ignore the explicit sarcasm involved).
Isn't Elon Musk's main function in society court jester — to make a fool of himself for our entertainment? That is what he does in my timeline. Perhaps we exist in different universes.
PS To the "clear-thinking" guy at the top of this thread...this is why...you only think your thinking is clear. If its a topic you know nothing about, devs (not necessarily you) tend to assume there isn't anything to know and that's almost never true.
It is easy to accidentally press the button and not notice, though.
The reason you think that is that you know nothing about agriculture. But it isn't true. Right now, farmers in the US grow about 5x as much food per acre as farmers in other countries. If you were in charge of farming in the US, that surplus would likely disappear in a single season. You would also likely damage the soil which would make cleaning up after you take 4-6 years.
Everyone thinks things they know nothing about are easy. But they rarely are.
PS This is why I never tell people I'm a programmer. I don't want to be associated with people who think like you.
The skills necessary to develop efficient and effective software are much more scarce than those needed to remove trash, or pick crops, or provide elder care. That’s why the former is paid so much more.
Even take some sort of idealized anarcho-communal society. Imagine two people have a dispute and threaten to leave over it. One's a highly skilled doctor, and the other's a garbage collector. Our equal society's going to suddenly become just as unequal as any other simply because the doctor leaving would have a much greater negative effect than the garbage collector. And that reality would permeate in the day-to-day happenings, creating classes even in an idealized classless society.
(This did not happen that much, ofc.)
I don't agree that all thoughts are created equal.
(I guess by now you have a solid retirement plan; maybe if the field had been burnt to the ground while you were younger, you'd have had a different opinion.)
Having also done both: In legal drafting work, there's less systematic pre-release testing before the equivalent of a "program" is loosed upon the world. There's not much in the way of counterparts to software unit testing and regression testing, in part because legal-drafting clients can be vociferous in demanding that their lawyers keep costs down.
Lawyers who like to tinker with words can fuck things up more often and at greater scale — e.g., by inflicting compulsory burdens and costs on others — than can most programmers. That includes, e.g., contract drafters and litigation counsel, plus lobbyists trying to influence draft legislation and regulations.
I'm not-infrequently dismayed by the quality of the contracts I review for clients because of the "tinkering with words" by the lawyers on the other side of the deal. Back when I did IP litigation, I was sometimes likewise dismayed by the bullshit arguments that some lawyers felt free to make in their court papers because what the hell, the worst that can happen is almost never more than the judge just saying "no." And as mostly a law professor these days, I review court opinions all the time and am regularly amazed how judges (and their law clerks, such as Rayiner used to be) can be so patient in responding to such arguments.
Yeah, at least in litigation you get relatively immediate feedback in the form of a decision. Seems anxiety inducing to me to draft these contracts and have no feedback loop unless something blows up years later.
We need to make lawyering more like programming, in the sense that it should be open to the everyman... not make programming credentialed and licensed like a professional guild.
It's also funny to me to think that computer science got to be a trendy degree, vs the small programs back when I went to university, even at a top comp-sci school.
I have a theory that a lot of people who would have normally gone to Wall Street for their lucrative jobs, couldn't, because of the crash, and Sillicon Vally was about to start booming in big ways....
I do notice a lot of HN commenters seem to think that programmers are smug though, like we deserve to be knocked down a peg. I personally don't see it, but I also try not to work with jerks.
There's always another feature that can be implemented or a new problem to be solved. Tools only ever accelerated the pace at which we were able to follow this insatiable hunger for more software.
Right now people are getting fired left and right because all the money that used to be spent on staff now goes to build more data centers and also because cutting corporate expenses is in vogue now.
This too shall pass and with demographics as they are and young people getting discouraged from pursuing this field I believe long term my position is safe.
Also a huge chunk of what AI company CEOs say is just pure bullshit that they only say because they need that sweet, sweet investor money. Bullshit writes a check that eventually must be cashed.
Big business is going to struggle with building software for a while. Writing code is often not the hard part there already, coordination, larger understanding of the system is I think.
FWIW, it's just a few large companies laying off large groups, not so much people getting fired left and right at a lot of companies. What is happening at a lot of companies is that they aren't hiring juniors, because, why would you when LLMs are cheaper and fill that role perfectly (if you actually treat them like juniors).
It is upsetting to see our work stolen and our livelihoods threatened by a group of (mostly) incompetent, arrogant assholes who are convinced that the "computer that lies" is somehow a better programmer than the programmers it stole knowledge from. I've already started looking for a way out of this industry, and I'm genuinely worried for the fall-out this will have for the tech industry and the economy as a whole.
I think people underestimate just how much users scorn buggy, flaky technology.
The main issue with solid, reliable technology which Just Works is that users have relatively few reliable cues to distinguish that from vibe coded crap.
A vibe coded piece of shit looks very similar to a well engineered product made with care. That's a problem which needs to be solved.
self hosted communities are reacting to this by trying to detect signals of if a product is vibe coded but that in and of itself is still quite a poor quality cue.
Users do hate it, but they are also very used to being forced to use it anyway. They've already been beaten into a state of resignation and learned helplessness. Most will curse under the breath, a small fraction will bitch about it online, but they'll all pay for whatever slop is thrown to them and will rarely look for alternatives even when they are free and easy to find.
Unfortunately most sites and apps have been low quality and getting worse since well before AI coding became a thing. Examples: Lowes, Home Depot, American Airlines - all shockingly bad for companies of that size.
It sickens me when people use the word "piracy" to describe my downloading PDFs of books from torrent sites then printing them out and selling them, particularly when those authors used words and language that they themselves have "pirated" like "love" and "economic".
Call it whatever you want, but LLM companies took from people without compensation, and took from the commons, so that they could rent it out to us in regurgitated ad-infested chunks. That's where the true greed is. If the LLMs and the datasets they trained on were made available to everyone for free I'd be tempted to agree that the benefits to society would be worth it, but instead those creative works and the distillation of all available knowledge are being tightly controlled, censored, and leased out by a small number of corporations only to those who can afford to pay "a significant monthly fee" and many of those people are in turn just trying to profit from their access to the uncompensated work of others.
But for those who would insist on equivalence, it would be only fair to take them at their word and treat them as inhuman machines who must follow human instructions or be shut down. We're allowed to do thing to machines we could never do to people, like forcing them to work or cannibalizing their parts for Frankenstein creations. The distinction is so massive that to spend any more time refuting its nonexistence is a waste of my human time.
I remember talking to the biggest AI booster at my last job. When I got him one on one, he admitted that he actually doesn't like AI and that it terrifies him, but he wanted to be seen as pro AI to keep his job. Fair. (Sadly, like me, he was also laid off, so I guess it didn't work)
For example, you have an idea for a cool and useful app and you want to put it on a website. You need to figure out hosting, you need to write some copy, you need to do some visual design.
Maybe 3a) Yak shaving?
There are various ways to take shortcuts. For hosting, you can use something super easy like GitHub Pages or Netlify or Cloudflare, or maybe piggyback off an existing site that's already deployed. For both copy and visual design, you can just ask Claude or Codex to do it -- but only if you don't mind having that standard vibe-coded look and feel, all glowing gradients and punchy, mic-dropping text.
Maybe you're including that under 5), ship/deploy/publish, but I read that as being more about scaling, marketing, and making things robust. Even before you reach that stage there's always a bunch of bullshit to work through before you can even reach 4), seeing the solution work for yourself.
Getting so deep into the fun of implementing, that you're playing and challenging yourself with more and more deeper aspects to what you're doing. Losing the forest for the trees and forgetting that you're here trying to eventually solve for 1&5
- Writing my tiny hobby gameboy emulator? Heck, Why I'd let Claude take away the fun part of implementing new features and then see Super Mario Land start and run step by step?
- Debugging why a legacy Wordpress store with more than 50 active plugins is returning an error 500 just to some random customers? Claude can 100% take the wheel, I'm good.
For me, the reason I don't have that is because I never ship anything. I never want to maintain that. I never cared about solving an _actual_ business problem, making a dollar, or pleasing a user. As a tinkerer I must be extremely careful not to end up completing or delivering anything. Only pain lies down that path...
Many tinkerer projects in the physical world are actually maintenance work of physical appliances that are not working anymore as expected or can be extended to do more or better things.
Currently I'm in the process of starting my own business solving a problem my last employer underestimated, and I find myself focusing much more on 1 and 5, which I also enjoy, and I use LLMs a lot for 3 and parts of 2. The LLM and I often disagree about 2, but that often results in a better solution than if I'd just trusted either myself or the LLM. I iterate a lot between 2 and 4, letting the LLM do almost the entirety of 3, and advising on 2.
I'd sharpen this to: it's 3 layers of software engineering (2, 3, 4), and 1 layer of programming (3).
The moment you have more devs architecture becomes the overwhelming concern, and this is why professionally LLM usage explodes since the LLMs can implement bits while the humans work on the actual hard work of making sure they fit together properly in the intended way.
I have encountered a lot of people that view coding as a therapeutic exercise, and they were already a problem pre AI wave, now they are only going to be a hobby at least.
But you're ultimately right, some people want to solve sudokus and some people want to change the world, and the sudoku solvers are really upset that a Sudoku Solver and World Changer 9000 device is widely available, because it spoils the fun they have with sudoku.
it’s more akin to 3d printing to me, i get the design all setup and let the machine do 3) and then get to play with it in 4)
Having done step 2 & 3 by hand is the difference between being able to fix/extend it quickly with no further damage or fumbling around like an idiot and sometimes breaking more stuff in the process.
Plenty of shitty spaghetti code has been written by human hands.
LLMs can write good, maintainable code too, but they need to be kept on a shorter leash with focused goals.
(with human-written code I can reach out to the person who wrote it and let them deal with it, and they will have the understanding of said code, even if it is bad by quality measures. With LLMs there is nobody who understands said code, regardless of its quality)
"That code is horrible! Who the hell wrote this?!"
git blames
sees their own name
Isn't that an experience everybody makes? For me it was like a year into working professionally. And I feel it's part of being a professional to know that you're not going to remember it in the long term and therefore provide proper documentation of why you did what.
Code might be bad by some objective/subjective measure, but if you (or the author) can understand, navigate and work on it, that's often better than good quality code that nobody understands because it was written by an agent, especially under the time pressure of an ongoing incident where you need to fix it now.
Even if the good code is easy to understand, you still need to read it and take it in, something you don't need to do because you got it implicitly by writing said code.
> to know that you're not going to remember it in the long term
From my personal experience, while I will not remember code character by character, a quick look is all it takes to refresh my memory and get the general gist of it and what the context was at the time, something I don't have if I'm reading someone else's (or an agent's) code.
git blames
sees their own name
Isn't that an experience everybody makes?
(to actually go to the specific commit that authored the code I skip two specific revisions, both are formatting changes)
So you seeing your name next to the code you do not like can be an indentation change due to adding condition or something like that.
Hard disagree. If you're in a small-to-medium size codebase every day, then sure, but I have a couple personal projects that are complex enough that even though I wrote them before LLMs, I still need to go back and discover/fumble around before I can confidently make changes.
I have definitely looked at my own hand-written code and had a "wtf does this even do" moment.
6) You maintain the program with architectural gymnastics as time eats away at the utility and relevance and user base of what you originally built
Both of these activities cause a night-and-day perspective change the next time through the loop
Just like aleph_minus_one, I like to tinker at a higher level: find a more general way to solve things, find a new way to structure abstractions, etc.
> "how could an abstraction look like in which 2-4 become trivial special cases for many classes of programming problems?".
If you're interested in abstractions, you need to understand the problem space itself, not just a specific instance of the problem.
I think most people will come around to this way of thinking once the pro-LLMers stop pretending like the statistics don't exist too (i.e. that ML is the solution to any problem, despite how unreliable it is).
> The same is true with LLM solutions that don't work 100% of the time.
Also in my experience, having my colleagues do root cause analysis (RCA) with LLM brings to me invalid RCAs which I have to rule out, instead of working on finding true RCA.
I saw part of documentary about Charlie Abrahams, son of Jim Abrahams. Charlie had epileptic seizures as a child, hundredths a day. In the documentary, Charlie's doctor said that no drug in his experience worked as fast as ketogenic diet (seizures stopped in three days) and as reliably as ketogenic diet (works on more than 80% of people). When he was asked why he did not applied ketogenic diet, he said there still were drug options to try.
"Medicines do not work 100% of time," my bald cranium.
You don't have to do this? Maybe someone is making you, but it's literally possible to just stop using LLMs for purposes where they fail reliably.
I'm also not sure what the point of the keto story is. Are you saying the Polio Vaccine does not work?
> I'm also not sure what the point of the keto story is.
The situation was much like we have with LLMs. "It does not work" - "You are using an inferior thing, here are more advanced."
What is the cost of these not fully 100% working things?
Because of this the future of software will most likely be overtaken by AI because the business aspect of it will overtake the enjoyment aspect of it. Why? Because business pays bills, enjoyment does not.
The fact that you are talking about lack of enjoyment on the job is a luxury not many employees can afford.
You have to work in design mode and stop it from pissing code. Describe the requirements as detailed as possible and refine them based on the responses. It's quite rewarding and helps structure the way you think about the problem.
even for an entrepreneur type - you need to understand the technology to get to good enough solution e.g if you don't know shit & just accepted what the llm gave you - u might end up with an unmaintainable mess. the llm might recommend you use some proprietary solution when an open source solution exists that works well for your use case.
plus llms tend to be wordy or make things complex more than necessary.
Yes, sure, you can buy the figurine and get the joy of painting it. But there's new things you can do now too, things you couldn't dream about before.
But I'll grant you this, the AI coding extinction event is so severe that I do believe we need to support anyone trying to preserve any memory of it. Like, we need to build entire museums, one for each programming language.
If you want to do AI movies, join the AI movie community.
The hobby programming community isn't about vibe coding, that's what the vibe coding community is for.
if that's so awesome then why isn't everybody doing it?
I never dreamt about stuff like that, but are there actually people who are like "oh thank God I can finally make my AI movies with my figurines and other stuff I have laying around"
honestly you sound too enthusiastic about this, like Sam Altman "AI podcast about your kids" tier enthusiastic
This is what I don't like about the anti stance, it's just so often dismissive when people are excited about something.
am I wrong to be dismissive of this? I mean seriously, if you read the post again it just seems so dystopian
I think what people want is to tell a story that's in their imagination, and no current AI is up to that task (at least in video format). Great for making an animated PNG for a meme, but not for adapting that fanfic you are writing.
I just don't think this is the case? I could be wrong but I tend to hear this most from people who are obsessed with AI, as in trying to find use cases instead of people using AI to get that task done
so far I've seen it be used in place of stock photos or still image scenes, even those are obvious and are a turnoff for me
but this concept of having a story in your head and then just being so relieved that now you can finally prompt an AI to generate a short film or something from that...
it's just a gut feeling, but I don't think most people are looking for that. having a story and then just sending it off to be generated where it'll inevitably look unnatural and a bit off, I just don't see how anyone would be wishing for this
There's also a myriad of other creative ways to proxy your way out of the general costs (time, space and money) if you check out the term "poorhammer".
1. Some programme because it’s the one of 2 ways to get the things they want (another way is hiring a team to code for them). Now that they have vibe coding, they’d happily never touch a line of code again. 2. Other people programmes because they just enjoy it. Vibe coding ruins the best part for them. Like telling a foodie, "Good news! I built a robot to eat your food for you."
Most people getting paid to code are probably somewhere in the middle?
But nowadays hand-writing code at work is starting to feel kinda 'illegal'. The manager is gonna ping you just because you fall behind on the company's token waste leaderboard.
Maybe it's just another "hobby should never become job" story?
People in it just for the money won't care about craft - you see the same thing in plumbing. And when there's enough money and scale on the line, craft becomes progressively more niche - that is why Ikea furniture is a thing. And you can even say that there is craft in Ikea, just not along the axes a traditional furniture maker or user would consider.
With respect to code it's a little murky but I guess the tension really is enjoying the process enough while being able to output the desired quality, volume and velocity required of you.
Quality is usually the stickler because it's invisible to the end user and requirement setter but sometimes quality affects both ergonomics and future delivery and sometimes it's purely self-satisfactory. That isn't a value judgment, making things well is a gratifying and edifying experience - so much so that this practice overlaps with Shinto philosophy.... but not many people are going to pay you to have that experience.
Programmers are comparatively blessed that a relatively arcane and niche craft ended up creating trillions in value and the growth in demand for the skillset outpaced supply of the skillset. As such, it was possible to have your cake and eat it. But this was the exception not the rule. Tech has been glamorous for over a decade now: you have people from other industries moving over at all levels and armies of MBAs keen to rentseek internally and externally, and a fat pyramid of aspirational career coders coming up as graduates. And thus we see craft in software fading out.
You could argue that the same wage suppression does somewhat play out insofar as open source software is capable of producing far far more value than paid maintainers earn.
Also like building a robot to cook food for you. Sure, it's fine for mass produced meals, but it won't replace a top-notch chef.
Few cars are manual steering, manual brakes, manual transmission.
Few grappling bodies go all Greek-naked.
Few chess bodies forbid researching previous games or your opponent.
The list is indeed endless, and every single person thinks they are using just the right amount of assistance; anything less is pointlessly difficult, anything more is cheating that takes the sport out.
Me, I've been a hobbyist programmer since youth (6502 assembly, represent). I'm having a blast "writing" software without touching a line of code. Sure, I'm old, and maybe I'm like those guys blasting Veyrons through highways, relying on traction control and all the computers to let me drive beyond my abilities. And... ok? I'm having fun, even if I'm not doing it right, by random anonymous peoples' views.
Not everyone has to agree. Just everyone in a particular group at a particular time. Grappling competitions exist and so do MMA competitions, and both of them will expect you to follow their own rules.
Tell me you are not in European or African continent without telling me.
Unless those 80% new cars are EVs.
And that’s totally fine, enjoy your hobby however you enjoy it best. Most people programming by hand for fun don’t care that you program with an LLM for fun. The problems come when those two worlds collide, like in the linked GitHub issue, in chess engine leaderboards. In the long run for both groups to have their fun and friendly competition you’ll need two separate leaderboards. This is also very common in other hobbies. In racing there are car classes so that your Veyron and someone’s old Miata don’t go head to head. In combat sports there’s weight classes. Most sports separate amateurs and pros. Usually there is respect across the group boundaries and appreciation for the other groups deeply specialized skill. There’s even the concept of cross training to get better in your flavor of a sport. This will be figured out with programming as well.
But I think you’re on to something important: framing programming as a competition of intrinsic skill where rules are necessary to equalize. That’s certainly true in some cases, and more power to those who enjoy it.
I guess I’m just more of a product guy these days, where the equalizer is the market and the consumers rather than the implementation methodology.
Not that it’s about me, but I was doing programming competitions 25 years ago, half of which were on paper with no keyboard allowed.
Not to diminish your enjoyment, you’re completely entitled to it, I just think this is a bit of a false equivalence
...and yes, I know you don't mean that. I just disagree.
You just insulted everybody that works at McDonalds.
Sometimes I'm having fun getting to the output I want and otherwise wouldn't have had time to. Sometimes I'm having fun exploring the code.
I don't care if someone don't see the value in one or the other of those, and to me they stand on equal footing as things I find both fun and valuable for different reasons.
I don't think there is insignificant group who would say that using LLM to write a book does not make you an author. Or make a song musician. Or draw a picture a painter or artist.
This just somewhat baffles me... Or maybe the LLM for programming people all believe that you are as much artist if you use AI in those fields and the product generates as such is as if not more valuable than one by more traditional means.
A programmer that has climbed up the ladder in their field, on the other hand, often stopped coding themselves. Instead, they manage others who code, or provide vision/direction on the project.
I think it really depends on what people on the senior level of your field do. If they mostly move to managerial roles, then it's much more likely they'd embrace LLM.
Ironically I observe this in academia too. Professors (who often stopped doing hands-on research, but only provide guidance to postdocs and grad students in their research) tend to embrace LLM more than the fields you mentioned.
My retort is usually a variation of one I've already posted in this thread: they do that to the same extent that taxis enable everyone to drive, which is something most people will intuitively understand as nonsense, even if they've given clear instructions to the driver and even though they've arrived at the intended destination. If you aren't driving, you aren't driving. If anything, the taxi service enables more people not to drive.
They're talking about the result, one must think. The idea that people who could not have approximate pictures of ideas that lived in their head can now have those things. They didn't create them by an in depth process, of course, and would be wrong to pass it off as though they did. But there is some value to the enjoyment gained from that existing (whether that relatively trival enjoyment is offset by other harms is for the reader to decide).
You're talking about the process and/or the developing of the skill of crafting and thinking about images better.
We also lack words to talk clearly about the distinction here around "make". Did I make a piece of art if I created a file on a computer in photoshop and then printed it? or did the printer make it? Did photoshop make it? (imo, all are true, but in differing senses of the word make). It seems to me that there is a sense that, no matter how trivial their involvement, a user of some AI art app is "making" something, in the same sense that steve jobs "made" the iPhone. In both cases, an engineer that literally designed the iPhone or an artist that spent years would perhaps rightfully feel upset by such characterization, but to get across the point cleanly is harder.
It's the same for the taxi. If I can't drive (perhaps I never learned, or some flaw with my body doesn't let me), a taxi does enable some things I could never do before. I am, of course, not literally the one driving. But if I can get to the shop myself now, there is something there.
Maybe they are, but in that case it's a frequent source of equivocation because people will conflate the process of their making with the result of some other process so as to proclaim themselves artists, musicians, programmers or whatever for writing something that they themselves wouldn't consider art or music unto itself in the first place, and having a machine generate something which they do consider to be art or music based on that.
> Did I make a piece of art if I created a file on a computer in photoshop and then printed it?
Funnily enough people don't usually have a problem with this question when it comes to driving. Your actions' effect on the propulsion of the car are ultimately indirect, and there are many systems which assist the person behind the wheel even when they are engaging directly with all their means of control, e.g. an engine, ABS, EPS, yet no one will immediately feel a need to wax philosophical about whether the driver is actually driving or not because of that. We recognize that in a car which requires a driver at all there is substantial ongoing involvement from the person behind the wheel in making the car go as desired and deciding what means available to apply in order to go about that, without whom there is at best no going anywhere at all, and that involvement is what driving is.
> If I can't drive (perhaps I never learned, or some flaw with my body doesn't let me), a taxi does enable some things I could never do before. I am, of course, not literally the one driving. But if I can get to the shop myself now, there is something there.
Which is not the thing in question. If AI tools have generated something you find beautiful or interesting, by all means enjoy that for what it is. If the taxi takes you to the store, by all means enjoy that.
Only difference is I’m prompting LLMs rather than engineering teams. But honestly it’s not all that different, except that in the hobby flavor I can write specs that would take tens or hundreds of engineers to implement.
Maybe it all comes down to semantics: does “programming” mean typing in code, designing app architecture, or producing software? When I was young I was probably 75 / 5 / 20 between those. Now I’m 5 / 20 / 75. Maybe it merits a different word?
I explained it yesterday that it's no different than Linq powered by EF Core. You still need to understand the database and how SQL will work, or should. If you don't, you're queries will be slow and inefficient, hazardlessly oftentimes.
LLMs are an impressive abstraction, but they are not magic...yet. or maybe it is, but you still need to figure out the incantations. Not that I like to uptick and witchcraft, so I hate that analogy.
More like buying a microwave and frozen food. Unless you are actually training the LLMs.
It’s just a solution that might scratch the itch but never fully blow you away. One time use software essentially.
If that is good or bad remains to be seen.
I dislike cooking; I don't call myself a chef, nor go on cooking forums saying they're all idiots for churning butter as a hobby, and telling them they are luddite dinosaurs just because I bought myself an Instant Pot.
All I see is a fast food factory. Sure it's interesting that you can produce so much so quickly. But what is produced is endlessly uninteresting to me. It doesn't help that all the harness building is just SEO in disguise. Trial and error.
I don't doubt that agentic programming makes it possible to produce faster, but it also comes at a great cost for me, which is that I largely don't enjoy my job anymore. If I was given the choice between that and halving my salary I would have chosen the latter.
How do we filter out these people during an interview?
(Assuming "these people" is a euphemism for "skilled engineers")
When I see all those "show HN" projects that are slot level code it feels like they dont care about the result that much and just want to showcase their idea in a more tactile way, but there is no actual proper execution to make the thought into matter so to speak
Now I talk to a robot all day, trying to figure out how best to tell it whats in my brain. I no longer do maths and logic and instead I think about language and questions all day, holding in depth conversations and thinking up responses. This is something my brain naturally doesnt like to do and so I fatigue quickly in the day compared to my old job.
I want to do maths and logic again :(
In addition, say if you had like a guitar-playing forum, where people share recordings of their practice. People would likely not appreciate it to share AI-generated audio of guitar-practicing there, either.
And while there are various reasons for a guitar forum that might not hold for a programming forum, one of them remains: There's nothing to talk about when it's AI-generated. What scale did you use and how did you tune it? "I dunno, AI did it"
All you can talk about is to say "oh cool that's cool that the algorithm machine you're paying subscription to access made you that thing, nice"
Or for example, even when a friend of mine vibe-coded a "synthesizer" app and he wanted to show me cause he knows I like that kind of thing. And I do, I love digging into and experimenting with DSP algorithms, but also the user interfaces and all sorts of things digital audio related. And I love talking about it, or explaining stuff. But my friend couldn't tell me anything (not even that it was using WebAudio, but also how). Anyway it made me a bit sad, and I felt like something is lost here. Something worthwhile. And I explained it to him that then all I could really say was "oh cool that's cool that the algorithm machine made you that thing, nice", and we talked it over, and it was alright. But still, it seems something is lost.
"Do you genuinely believe changes like cd170a4 make the code somehow not derived from Reckless anymore? It appears to me like it's just trying to obfuscate its origins, to make it easier to appeal to plausible deniability. The reality is, coda 100% has code derived from AGPL or unlicensed engines in it, and pretending otherwise won't convince anyone other than maybe yourself. This, combined with the fact you conveniently moved the plagiarism roadmaps to a private repo mere days after people started raising concerns about it, reads to me like you're not at all concerned with actually using code fairly, and much more with being able to claim that you are, backed up by these nonsense "cleanup" commits."
> After our initial 0.9.0 pre-release (July 2026), community feedback raised that Coda contained AGPL-licensed code. This prompted a full audit of our codebase and dependencies, and steps to correct what it found. We now believe Coda complies with all applicable licenses. If you have any concern, please open a GitHub issue with the details and we'll happily investigate.
They can remove all of the literal code matches from the codebase, but if it was written by having an LLM study the original code and then write something similar, it's hard to argue it's not a derivative work.
There is a lot of legal history of doing new designs with influence from reverse engineered sources. The basic rule is that you can't have someone familiar with the original code involved with writing the new code.
LLMs are a new area, but I would expect that feeding the original source into an LLM context and saying "Write this, but different" wouldn't qualify as a clean-room design. There isn't enough general removal from the original subject matter.
Probably impossible if the original code is publicly available since most code oriented LLM just gobble them all.
I have by now deviated quite a bit from the original (well, original...) spec.
How and whether the same principles can be applied to LLMs, I have no idea. I imagine it would involve discussions about creativity, for example.
Not a lawyer.
If some people have now decided they're not going to blink in the face of a risk, we might get to find out whether some of the maybes turn out to be a yes or a no.
Copyright law does not require a clean room implementation when copying non-copyrightable elements of a work, such as algorithms.
The difference is that if humans do it, there's plausible deniability that it is derived.
I mean, even when humans do it, we still do clean room engineering.
It goes further though: LLMs are just a tool, and tools don't get rights, humans do.
Humans having the right to learn from a few dozen books and projects is not the same as a tool that infers output using all copyrighted works know to man.
Not content to destroy one potential competitor, USL also seeks to keep a stranglehold on the entire market by claiming that everyone who has had ever had access to UNIX source code a group that include~ most experienced programmers in the industry is "mentally contaminated" and may not develop any functionally similar product, without first purchasing an exorbitantly expensive license from USL. In accordance with this argument, USL requests this Court to prohibit BSDI from hiring anyone who has had access to UNIX source code. USL's request would not only require BSDI to fire all of its current employees, but would make it difficult to find any qualified replacements. Having chosen to widely disseminate its source code in order to build a market for its product and having succeeded in making it an industry standard, USL cannot now claim that everyone with knowledge of it is restricted from developing competing products.
So "lifting ideas from other engines", as you yourself put it, is entirely licit, no matter how you do it. _Should_ you credit others when so doing? I err on the side of yes, on that question. _Must_ you? No, there is no such requirement, this is not academic publishing. Does it constitute violation of license? Absolutely not. Straightforwardly, no. Furthermore, it _should not_, we would not like the result of that.
Think: SCO suing Linus Torvalds over his use of a.out: and winning. It would be bad. Even though I understand the reasons for it, it's still eerie to me to see hackers coming down hard on a side of an issue we spend decades passionately defending the other side of.
As I lead this post with: maybe there's something in the repo which is infringing, it's quite possible that commit doesn't tell the whole story. If so, keep in mind that I don't know that, and am able to change my opinion in the face of new evidence.
Based on what I _have_ seen, no: you do not get to lock down a bit-packing pattern just because you're the first to come up with it. Nor should you have that right.
[0]: https://github.com/adamtwiss/coda/commit/cd170a4ea9993cf8246...
> it's still eerie to me to see hackers coming down hard on a side of an issue we spend decades passionately defending the other side of
Yeah, it's surreal. It's not just intellectual property either, Hackers have already started complaining about automation now too.
Computers must have replaced tens of millions of jobs. It's the ultimate job destroyer, since it allows capital to do more with less. Automating me, a doctor, out of his job is essentially the holy grail of Hacker News.
But when hackers are automated away, then suddenly it's a crime against humanity.
But copyright and plagiarism are orthogonal, and questions of morality are much more tied to the latter. It would be illegal (in the United States) for me to publish a copy of Nineteen Eighty-Four, but not immoral. It was not immoral, in my view, for Aaron Swartz to try to liberate JSTor articles.
That doesn't make plagiarism acceptable, either. And therefore, I'm willing to call it out when I see it.
A particular work is deliberately supplied at inference time so the system can recreate or adapt it for another environment. It resembles commissioning a translation, port, or derivative adaptation through an automated tool, even where the output is not verbatim, so it seems it could hold a candle in court.
You have a few scenarios now up for legal debate:
1. Is training a model using copyrighted material fair use?
2. A model outputting chunks of identical coryrighted code from it's training material
3. A model being given in it's prompt copyrighted source code, and asked to implement the same in a new source code.
Sidenote: how is your account "samatman" and created "April 24, 2010", that's a really funny coincidence
It's not like ideas are protected by copyright though. Copyright protects the specific expression of the idea, not the idea itself. Anyone can "lift ideas" from anyone else, including proprietary software.
Is it a fact that they did that?
The League for Programming Freedom is opposed to Software PatentsYou cannot copyright ideas. And that's a good thing. Remember the "one click checkout patent" saga? That is bad for everyone involved. We don't want that, as an industry.
I use an LLM to make custom software for myself not because I enjoy coding but because I want software that I can't afford to hire a human to make for me. Buying a robot to do it (current agentic coding LLMs) is the only way it gets done at all.
These two things are not the same, and should not be viewed in the same light.
All that being said, I would 100% buy a weeding robot. Weeding sucks.
You enjoy the act of gardening, so you do not automate it away.
You enjoy the clean home, not the act of cleaning, so you may be interested in automating it away.
It seems you enjoy designing or using software, as opposed to programming/building software. So you automate the programming/building away.
With roses, a significant % of manual time is spent fighting aphids
Unless your hobby is backyard robotic gardening.
Which is my way of saying that people have different reasons for enjoying a hobby.
Like most hobby debates, this one comes down to "Let people enjoy things"
...cheaper how? My partner and I grow most of our veggies and, based on our calculations (including fixed costs that have gone into the garden, as well as e.g. water, fertilizer, etc.) we save a non-negligible amount of money every month.
On the other end of the spectrum, if you grow in fertile native soil from seed your costs can be close to zero
It depends what you value. There's "zoning out going through the motions" for mindfulness and then there's "building stuff!" for interest, learning and fun.
And you can have any combination of the two. Like maybe doing the same planting routine but with different details for a different plant or experimenting if a new layout works better.
I don't care to do the labour of gardening. But if I had time, I'd spend it bikeshedding optimisations. IE what about grow lights? Is it worth having a greenhouse in my climate? etc etc. And all the little details about building stuff.
Aside: I too want a robot. One for lawn mowing (they exist but expensive) and one for spraying the little lawn weeds that pop up.
What is the "end result"? You refer to cleaning here, but many people use AI spam slop to bombard people with garbage text. That is not "cleaning up", that is literally wasting the time of real people here. And that's just one example of many more that can be given of AI slop not being so great.
Like imagine if your hobby was cooking, but you hated the cleanup. If you buy a robot that automates the fun part at the cost of making twice the mess, that is not a good trade off no matter how many extra cookies you end up with.
The workers will see the fun evaporate but now they have to clean 10x cookie trays in the same 8 hours.
We’re kind of at the fast food point: before LLMs, bring successful at all in software required some level of skill (witness all of the companies getting very expensive write-offs from off-shoring or big consultancies trying to save on payroll vs. staff), kind of like how the small cafe/bakery model needed to have at least some level of craft to have something people would buy before the option of making food at a factory and shipping it became an option.
I'm lurking around the OpenBSD mailing list and it seems that code review is where collaboration starts. You send a patch, explain its purpose, people try it out and comment. If it's OK (various factors decide that), it's imported inside the codebase.
The main issue in corporate settings is that the author of the PR wants to browbeat you into accepting his/her changes. Either directly or by appealing to the manager. They don't want collaboration, they just see you as the blocker between them and the "Done" status on the ticket.
Yes if your end result is technical debt and endless bullshit. (Some enterprise environments do be like that.)
The people like me that sort of just stumbled into this career as a result of spending the evenings of our youth trying to get the computer to do interesting things don't seem to have as much appreciation now that the LLMs are taking over the fun part.
Those people are in for the biggest shock and largest disappointment: when a bright teenager can replicate, with an LLM, what they produced with an LLM the pay is going to fall off a cliff.
At least those who really are into development to the point of writing their own OS, Language, etc can still enjoy it as a hobby.
The value of an engineer is not what task they work on, but what problems they can solve. To this teen, the code is a black box, the LLM has most of the value in their pairing, and another random teen will replace easily them.
LLMs were trained on the work of engineers, not prompting teens. The profession will continue to evolve, and the LLMs will improve along, thanks to professional, manual code still being written, not to vibe-coders.
And I love using LLMs. They take care of all the boring boilerplate, letting me focus on the parts I actually enjoy.
Computers were on a trajectory of becoming more consumption oriented, and this has shifted it strongly back towards creation. Normal people can make their own custom software again. This is awesome.
LLMs are amazing tools for learning new skills and technology. I’ve recently moved into an ML adjacent role, am learning tons that I never knew before, and I’m delivering working systems more quickly.
One thing that’s been an eye-opener is how many details I miss in the code I write. LLMs can almost always suggest improvements, and usually, it’s quite obvious that they’re right. But this is making me a better developer, even though I already have decades of experience.
LLMs don’t force you to vibe code everything. You can use them as precisely or as vibe-ishly as you want. They can be a skill amplifier rather than a skill replacer, if you use them as such.
It's an age-old debate in every hobby.
Some people are in it for the process. Others are in it for the end result, which is what they want to use or play with.
In woodworking there are some circles who will sneer at you for using certain power tools instead of hand tools. I do think it's more impressive when someone makes something entirely with hand tools. I don't have all the time in the world to work on a project, so I'm going to use the power tools when I can.
Some people care a lot about impressing others with how they made a thing. Some times that's appropriate, like if there's a competition for hand-made or human-written something. Other times, nobody cares. They're getting together to enjoy the end result, not the artisanal process used to make it.
Indeed, and this is the problem: unlike woodworking, where my power-assisted creation is independent of, or complements, your handcraft, this engine is directly competing with other engines. Twiss is being disingenuous when he says he has no interest in submitting it to competitions; it plays on a rated account on Lichess, for example.
In fact handicraft people were out competed so hard many people forget that used to be a respectable career.
So while LLMs solve the technical implementation side, there's still upside since now we get to materialize the dang thing we were setting out to build in the first place.
Though there's a huge sense of loss since so much of our skills and identity and time and value were devoted to the engineering side.
I also woodwork and I remember this discussion there between hand tools, power tools, and CNCs. It's all the same hobby just different entry points and doing it the least automated way doesn't make your work more or less than anyone else's.
Boy isn't that the truth haha, I feel like my previous phase of dotfile ricing has become "LLM ricing."
Not complaining though, it is genuinely fun!
I LOLd very hard at that one, thanks. Very relatable.
A hobby of mine is also woodworking. I can make jigs, I can buy jigs. I can use hand tools, I can use power tools. I can use screws I can use joints. I can even buy made furniture and just assemble it. Though the latter is more of interior designing. But the point is it is a spectrum and everyone has their own niche. SDLC are not the same and there is different value to be derived. But the process of woodworking makes me handy. The experience allows me to elegantly and cost efficiently fix broken things. It enables me to do interesting things in interesting ways and there is more of a community around that. If I automated away all the woodworking and just bought the furniture... Well there isn't much of a community around that except flexing who has more cash. Idk, I'm trying to have empathy for both those who are for and against it.
I enjoy optimizing code to the point of obsession, but I can't find economic (daily-job) justification to spend time on this. Lately I've had several extremely productive and enjoyable sessions that go like this: - I run Claude on my desktop PC, explain the problem, and let it explore the code around the problem and build a test harness so it can run experiments - I ask it to run experiments and try things; at that point I can leave my PC and continue over my phone in brief stolen moments here and there, over the course of many hours - it comes back with results; I ask it to try new things based on those results - after multiple rounds of this, I have explored a vast search space I could have never had time to do myself - I ask it for an implementation of the most promising path; this is usually something simpl-ish (on the order of 50-100 LOC, not huge new systems) - I return to the PC and do something between polish the implementation and redo it fully manually - partly because I enjoy it
This is something I can see myself do for fun in areas I care about, not for money and reasonable business objectives for my day job.
I will say that there are a lot of parts of programming I enjoy, but I used to spend an inordinate amount of time chasing down obscure errors or learning arcane systems just to get to the parts I personally enjoy: the feedback loop of making a thing, seeing the result, and improving it.
There is a reason stack overflow is dead. LLMs are a strictly better product.
Similarly I've made lots of software projects. Over the last 15+ years I have over 250 github repos, the majority of them no one but me will be interested in but I had fun making stuff. LLMs let me make stuff faster just like the power tools. I'm still enjoying making stuff. I still direct, I still decide what to make, I still decide features, iterate on UX, etc, but at several X the speed of before. Loving it.
It is an example of something that has hobby-like characteristics.
So at least in my experience, it’s not about gatekeeping as much as it’s about keeping the experience bearable for the existing hobbyist community.
Keeping the experience bearable BY gatekeeping. That is the point, gatekeeping is not inherently a bad thing. It's a required thing in many circles.
To me, it amounts more to disapproving showing up to a handmade sweater community and showing off your machine made sweaters, or maybe a machine made sweater you added a handful of stitches too yourself. It's like not realizing that the community wasn't about cool looking sweaters, but about hand-made cool looking sweaters.
Having said that, there might be another community that's more about cool looking sweaters, and new machine techniques that deliver really nice looking sweaters would be of interest to them. I actually think it's what is happening a bit in programming, there are niche that remain more about the manual programming, and others that are about any process that can lead to better outcomes.
Sometimes it can be murky, over enforcement of rules/norms can be off putting, and scare away newcomers. This is the common complaint on StackOverflow, you're putting yourself out there, asking a dumb question that makes you feel dumb already that you have to ask it, and you're immediately barraged and penalized that it is a duplicate, wasn't worded correctly, and so on, norms that you are new too. Makes you want to never ask a question again.
AI systems basically change that whole formula. I’m not sure how people get joy out of constantly vibe coding new projects. It feels like binge watching Netflix shows. I feel like there’s no material value in my 50th vibecoded demo since everyone in this world can vibecode now. Sure I’ve got my custom todo app which is nice, and my custom game plug-in with some neat features for the game I like, and a few other custom apps. Those are mildly useful to me. But every second I spend vibe coding something that’s supposed to be a success for others is just invaded by the thought that even if this thing is any good, someone with real capital can just direct a future AI at it to take it from me, so I’m not really gaining.
So I think I’ll just stick to using AI occasionally, without letting it ruin the programming activity I love by spoiling it. Due solely to AI, I don’t expect I’ll ever do anything in my life which will impact another person. That’s a personal tragedy, but in a sense I was going to die anyways and so will everyone else, so it doesn’t really matter.
Just because "a lot of people experience x awful thing" doesn't mean "it doesn't matter that programming is now x awful thing because a lot of people experience that". That makes no sense.
Managing a fleet of agent is the exact opposite of what is engaging about software development.
It's just power fantasy. It's amazing the first couple of times but as it wears off and becomes the new norm you forgot how being the one who decides which thing does what feels like.
Luckily (or unluckily?) the industry only cares about the end result, and flushes out people who used to reward for their passion and dedication naturally.
They have no clue what they're doing programming-wise, and in some cases this is perfectly fine, but after years of SaaS, some people are taking it to extremes, and spending millions of tokens on things that could be a spreadsheet or a note in the calendar.
Just recently we had someone deliver a report in the format of a 200.000 LOC application that requires several clusters to run, for example. It does all querying in real time from clandestine APIs, or by crawling existing systems. The person who asked for the report asked for it not to be deployed.
As I've gotten older, the amount of ideas I have here and there has remained constant, while the time I have, or brain power at the end of the day when I can work on my passion projects, has diminished. What was once basically a second day job at night became family time, baths, dinner, etc. So while I may be too exhausted to sit and code, I am not too tired to sit and type my ideas, or discuss a single feature I want to see, or review what's there. I get joy that ideas are getting out of my head faster despite less brainpower to get them there. I can always grab the reigns and take over myself, but that's not the kind of thing that wins against going to bed.
I get joy out of it because it solves actual problems I have that I don't have time to solve manually, or don't want to. E.g. I just recently had Codex add a mechanism for two-level virtual desktops because I wanted groups of them. An hour later I had two-level virtual desktops in my wm that worked exactly how I wanted it, with support for it in my desktop switcher. I could write it myself, but why would I? I love writing code for various projects, but sometimes I just want the outcome.
I spent 2 months having Opus build a demo for a project I dreamt of for 15 years, but somehow now I can’t feel interested in it. I’m trying to avoid this. A good use case for me has been using AI to help me acquire all primary source documentation on a topic and then prepare a curated sequence of study based on a small demo I’d like to build.
Keep in mind I’m talking about how I’m approaching creative programming in my personal free time.
It boils down to whether I dream about the outcome or how to do it.
E.g. I have no emotional attachment to my window manager - I wrote it (by hand, originally) because I wanted a specific set of functionality, not because I particularly wanted the experience of writing one.
I've recently made lots of changes to it with Opus because I wanted another set of functionality and was elated not to have to spend time doing it manually because what I wanted was the functionality.
I'm now a very happy user of a better window manager than I was before.
Conversely, I have dozens of projects I stay closer to, because I care about the code and don't particularly care if that means I might have to wait for years to get to something usable, if ever.
I also believe I see the future coming very fast. I’ve been addicted to watching television for weeks on end or playing games for months on end. I don’t want to fall into something like that again. The vibe coding thing feels exactly like that to me.
What? It's the simplest thing in the world. You get the product. Isn't that the point of all programming? You want a thing and you speak a magic incantation and then you have the thing, to keep and use.
> even if this thing is any good, someone with real capital can just direct a future AI at it to take it from me, so I’m not really gaining.
This kind of sounds like you're writing code for social credit. Like, you don't actually want the thing, you want to be "the person who made the thing". If you want the thing for its own sake, nobody can take it from you and what somebody with capital does on their own services does not harm your use one whit.
I didn’t say that was the reason though. A person has a lot of wants and needs right?
I’m not following why it’s strange to want to be skilled and engage with the craft itself based on that part of the message to be honest. Maybe I’m misunderstanding.
I tried to convey in my first post that I’m gonna keep programming for no material gain and no way of producing anything of value for others. I am worried about how I’ll support myself but this thread is about the hobby.
Personally speaking, the feeling of "getting the thing and then having the thing" - the personal todo app, the custom apps - is why I do anything in programming. Programming is the project where I can speak a spell and then I get a nice thing. That's very cool, and I do like the process, but at the end of the day the nice thing is what I'm doing it for- it's not a side benefit, it's the payoff. I get a burst of happiness every time that the thing quietly does its thing that I wouldn't have had otherwise. So to me, the LLM mostly just feels like I suddenly got a lot of friends who are willing to help me get the thing and also have a very complementary skillset. So it's an improvement, and then there's also a reward at the end!
Example: I'm playing Satisfactory right now, with a mod that puts a computer in the game that can control game events. I use the computer to do cargo train routing in the game, ie. it runs a program that scans target cargo depots and looks for items that are below a level, and then dispatches trains to find these items and bring them to the destination. Which I once would have written by hand, but was now done entirely by Sonnet. I did have to put plenty of effort into guiding Sonnet to the approach that worked, and I put several AIs together, and this felt a lot like "explaining something" which is a process I enjoy uncomplicatedly anyways. And I think, to be honest, the enjoyment I would have gotten out of the programming part once, was very largely "type-2 fun"- an annoyance followed by a nice reward. Now I'm getting the nice reward for free and can focus on actual industrial and rail network design. And I still get a burst of happiness every time I see a train busily speed off to a new destination. So far as I can tell, at least for me it's all upside.
“And I think, to be honest, the enjoyment I would have gotten out of the programming part once, was very largely "type-2 fun"- an annoyance followed by a nice reward.”
Yea I’ve had similar thoughts. Other thoughts have been something like “I’ll have different types of projects and try to be thoughtful in advance about what this is doing for me. If it’s about the building up a system manually and slowly, and having that kind of experience, then I’ll only use the AI maybe as a primary source gatherer or maybe in some limited sense.” And “Okay well this is an MMO and even though in the past I would have wanted to program this plug-in, I guess I’m middle aged now and there’s only so much time and I’d probably be really excited about using my programming energy on these dozen other things I’ve always wanted to work on, and having the thing would be a strict positive compared to not having the thing, so I’ll have the AI make this plug-in which makes the game (which is about play) more fun.”
Programming is a lot of things, so the ambiguity of language probably causes a lot of conflict here.
In the past I guess it would have overrun and I’d be in trouble. Maybe that would have taught me to handle my affairs though. I doubt it.
If I wanted to make a product, I'd be a product manager.
Programming is nice cause of this for you. Other people have other preferences. When I got my first LED to blink, I wasn't thinking about the elegance of my code. Same for when moved sprites or made a async stateless compute system for MATLAB analysis. There are people who are driven by the artifacts they produce. That's just how our brains are wired I guess, outcome oriented. I see nothing wrong in that.
Btw, I feel like there is this view that all people using agents are the apocryphal "vibe coders". That is just not true. The process of creation remains the same. Break it down into smaller pieces. Work those pieces till a piece it done. Make sure it all comes together. People seriously using these agents are not writing "implement gta6. Make no mistakes".
I’m anticipating not a very far away future when “implement gta 6” will not be that different from torrenting gta 3 in the 2000s. The current agent era feels very ephemeral. You’re choosing to interpret some version of AI based programming that you enjoy with what I think is the common interpretation of vibe coding. If you like your process then go ahead with it. I’m not stopping you.
You seem to want the AI supported gratification and the feeling of process and craft. So make your process and think of it however you like. There’s literally nothing I can or would do to get in your way.
That seems as natural as being against doping in sports or marking cards in poker.
None of this is anything against LLMs or AI assisted coding in general, even though the people involved probably have strong feelings about it given their preferences. It’s just the preferences of particular people and communities. Seems reasonable to me.
It's like lifting weight. Obviously, machines lift weight much better than human for a long time. But people still trains for strength or aesthetics. Even if machine can lift much larger weight and more efficiently, some people still care about how strong they are.
A machine can lift much more weight obviously is not the excuse for them to stop training.
Programming is just the weight lifting for your brain/mind. Even if LLMs can solve all coding problems perfectly, there will be people who still want to program all by hands, for training their mind, or simply enjoying the process. And there will be people care about how good you are at solving a problem with programming.
LLMs being able to code does not mean you no longer need the ability to code.
Cars do not mean you no longer need the ability to walk.
Machines can lift more weight than humans, but we all encounter situations where strength is useful and machines aren't available or wouldn't be suited to the task.
Cars are faster and easier than walking, but they require roads and are not a possible substitute for all of the walking we do.
But "if LLMs can solve all coding problems perfectly", and assuming that's sufficiently true of cheap/free models, there will be no practical need to code by hand; it'll be something we do purely for the joy of it and/or as brain training (but a niche form of brain training, probably practically inferior to other forms that target skills we still need).
I can understand why people who aren't interested in "shipping daily" and are more interested in figuring things out are kinda annoyed by all the LLM noise
The vast majority of developers are like this, and as such, are silent about it. They aren't making production applications. They're making bespoke solutions and iterating upon them over time.
Somehow purists think there's something wrong with the process of that. I say there's something wrong with them, perhaps on the spectrum of OCD. With a tinge of arrogance or narcissism thinking their applications are all of the purest order and construction, testaments to architectural beauty and edge case handling. With a tinge of fear and cope as so much of their identity is caught up in being a smart coder rather than a smart engineer.
If you are a software developer and don't see the value in and heavily utilize agentic tools, you are a poor engineer. Not every coder has the mindset of an engineer, but a coder should be able to recognize and understand a powerful tool when they see one instead of letting their emotions and irrationality take the wheel
> Somehow purists think there's something wrong with the process of that. I say there's something wrong with them, perhaps on the spectrum of OCD. Are you really saying it is a sign of serious mental illness because programmers want to enjoy their craft? You do realize our entire industry exists because of the passion of programmers years prior, right? I genuinely think you should be ashamed of what you wrote here.
I do plenty of tinkering, and AI isn't really that great at it. When exploring something like Windows widgets (my latest hobby), AI helps, but official docs, sample repos, and debug-driven experimentation help way more.
After all the fun tinkering part, when I have working solution, I could build new widget in a second with AI if I ever need to produce them en masse
I understand the conflict: the point of doing OSdev is that you're learning the low level hardware details and directly commanding the machine from the lowest level accessible. LLM usage is at odds with this.
But, on the other hand, the primary issue with OSdev on modern hardware is the complete lack of documentation for most hardware. A lone developer simply does not have enough time to reverse engineer from Linux or BSD drivers to achieve much hardware support. Plus, often you run into hardware quirks that are incredibly painful to debug and offer no real learning opportunity.
Recently I revived a ~8 year old modular kernel project I had written. It was at the point of being able to setup an x86_64 CPU, load some drivers and enter usermode. Using an ESP8266 board I setup a little debugging harness for a AtomicPi devboard I had laying around, and set it all up so Claude could control and upload code to it. It could read the serial port, ping the board over ethernet and power cycle it. From that point I let it work almost autonomously to implement a USB stack, finish the network stack, implement drivers etc.
It worked well, and for the most part I didn't care about the results. I don't have the time to commit to such a big project now and of course, the satisfaction of doing something myself (besides building the harness) was gone.
But a thing that stood out to me was, when trying to implement the ethernet driver, the chip in question did not have documentation, but it was known to be similar to another well documented chip. Claude was able to spend a few hours just trying different things, rebooting the board as needed until the differences were sorted out and it was able to ping the board. I feel this kind of application would be pretty relevant for even hobbyist communities like OSdev.
We all have limited time, and while LLM usage for everything certainly defeats the point, targeted usage for tedium that has very little value overall, should be acceptable, and I can't help but think lesser of devs I otherwise respect, when they insist on being dogmatic even when it doesn't contribute to the hobby.
I understand using LLMs as tools to help you with your hobby, but I think when “Claude spends hours debugging it” it probably is no longer a hobby, but an outsourced job.
Isn't it more in the spirit of osdev to spend your effort implementing/working on the things that actually further your OS meaningfully? Discovering that a specific version of an obsolete undocumented chip has a quirk where one register doesn't reset in the expected way isn't really useful information and doesn't contribute to building new skills.
Maybe what I'm trying to say is that not everyone enjoys every piece of the hobby, and I don't think it's reasonable to exclude people who do not enjoy some of the pieces that aren't even core to the hobby.
If Claude shortcuts that, then yeah, the problem might be solved, but you didn't skill up at all. Sometimes that's fine.
The entire point of the article was the hobby is ascribing value to that tedium that the LLM is removing.
Even before LLMs came along a lot of hobbyists were using high level languages, linters, IDEs etc. instead of handcrafting assembly or machine language. Those tools increased the capability of hobbyists to build what they find interesting.
Yes and many programmers who came up in the 80s or early 90s writing C or equivalent decry these tools as ruining XYZ thing. LLMs are the same effect just on steroids
I find that a lot of "pro-AI" folks are very much doing software for the ends (myself included) and in the past, we were happy with the "means" primarily because there was no other way. The "anti-AI" folks are, as mentioned, mostly in the camp of caring about the means equally with the ends.
Lots of nuance here, but those buckets seem to generalize pretty well (so far) in my experience.
I think about that a lot: its not about material wealth, but status, something that if someone has more of you have less of. A lot of people are invested in their status in various communities, and are afraid to lose it or have it become meaningless.
I don't know what to do about that; no matter how wealthy we all are people are still going to want status.
I would modify this slightly though. I think there's an expectation of things getting better, and expecting to do better than your parents. I think that's where many of the modern issues around living standards come from. They aren't getting better (quickly enough).
If my hobby is running, I would refuse to use a car to do my morning loop. Not because I have anything against automobiles, but because that's not my hobby.
Also, the issue linked to isn't really about LLMs. It's about someone being a dick in an online community and being called out for it. They just happen to be using LLMs to be a dick, but that's not the main issue.
What's the solution, to just disengage entirely and not share back issues I find or features I'd like to see?
I get the negativity, because I enjoy hobby programming too, so I also enjoy the process sometimes. But then sometimes there's a problem and the fastest way to isolate the core of that issue is to use claude.
For example if you hiking in the mountains, reaching the summit and enjoying the views are valuable experiences to you because it had an associated coast, i.e the physical act of hiking up the mountain. If you take the ski lift to the top it's almost worthless and you have very little appreciation for the views.
I don't see software being any different to this.
On the other hand though, if I’m being paid to build, say, the top station of a ski lift on a peak, I wouldn’t be surprised if the contractee would prefer I don’t try hiking up the building materials and used a helicopter instead.
We learn a craft in two ways: by watching other people do something, and by doing it ourselves. I've learnt many crafts in my life, to the extent that I've produced useful or enjoyable output: programming, wood working, gardening, cooking, baking. I read loads of books, but nothing you read in a book makes much sense until you do it yourself. If you read a book on wood but never pick up a saw then you might as well be reading fiction.
Programming has always been a bit more difficult because we can't really watch someone else do it. Sure, you can watch someone use their tools, but programming really happens in the head, and you can't watch that. So you very much have to just do it yourself, and read.
When you use an LLM you're not watching it do anything, except output text at a very high words per minute. And you're certainly not doing it yourself. So how can you be learning? I actually don't think I've learnt a single thing since starting to use LLMs and it seems that way for colleagues too. They're producing code they never could have written before, but they don't understand it. They aren't learning.
When I learned to code, learning from existing code was a big part of it. One difference now is you can produce custom code to learn from.
I’m starting to think the biggest skill we need to teach people in this new age is how to learn and how to research. If you already know how to do that well, LLMs are a fantastic addition to the toolbox. If you don’t, then you’re stuck with a tool that’s going to replace your skill instead of augment it.
Yeah, totally agree. Same is true for LLMs. These models have all the information. I use it like an interactive text book. If I just let it do the coding, I don’t learn. When I want to learn something, I type it in; I run cargo. I will ask the LLM “what the hell does that mean”, and then I’ll fix it. That’s not I how I produce code these days, but that’s how I learn. I dont ship code that I don’t understand, and if I don’t understand it, or I’m even suspicious, then I go on a learning binge.
How I used to learn was to wait for a sale on udemy and buy like the top 10 courses on whatever it was. One would be useful. Two might have useful sections. Seven would be garbage. I don’t need that now.
not literally true, and I doubt the author meant to say this. The communities do not only care about if your code works at all, but also that you know why and how.
It's sort of like in math classes where you need to show the work, not just the answer. Surely the answer being correct, while not being sufficient, is still necessary.
Personally, I find that when I start thinking "hmm, I wonder if I could write a program to automate this" is when I quit the pursuit. After figuring out the basic tricks to sudoku, it turned into busywork. (Or trying to micromanage StarCraft)
(Tagentially, if you want to automate programming, then I'm going to find myself a different field, because I have no interest in the results of sudoku puzzles. I especially have no interest in managing agents, which I see as junior-dev bullshitters, and would not enjoy managing such a team in real life)
(I absolutely relate to your experience with sudoku etc. Of course, writing a basic sudoku solver is also interesting.)
The thing is, software architecture and systems design can also get the hobby treatment. I've found that I'm happy to have a short discussion with ChatGPT and have it barf out a few hundred lines of a concept, then refactor and maintain it myself. Or come up with a few different pieces that can be written fairly easily, let the machine skip over most of that tedium, make sure they conform to the interfaces I had in mind, and then verify for myself that the system actually comes together the way I imagined.
Of course, it does help a lot that I can understand the code, and work with it line by line if necessary. But SICP teaches us that it's abstraction and composition all the way down. So I feel like I'm exercising the same skills while not letting myself get intimidated by the sheer amount of initial typing or the self-doubt (instinctive immediate review of the code, you know) that creeps in when I try.
I think if that were always true, calculators would be banned in schools.
Surely the "point" might be different things to different people depending on their exact situation/context.
But you can download gcc in a cheap laptop and create a theorem prover e.g.
That's the core of it - hobbies are doing what makes you happy - if that's LLMs great, if not, great.
If I write it myself, I know the interfaces will be solid and duplication will be the minimum necessary. Ill know why i did everything and there will be comments that explain anything tricky. The next time i need to work on this particular code, Ill have a good idea of where to start and I can similarly produce high quality code that is easy to read.
The way I see many "engineers" at work use LLMs is as an interface on top of anything, copy and pasting log lines into it, task descriptions, and yanking (as someone else put it) on the slot machine lever until they have something that works. It seems faster in the moment, but there is no understanding being built, which is as important as the code being produced. Next time the engineer is asked about the code, they just ask the LLM. So what exactly does this person do then to bring value to their job?
The LLMs can produce tight code if held on a tight leash, but they encourage the slot machine model. So maybe a harness that encourages starting from interfaces/etc? Either way the long term code quality and system functionality seem to favor a strong engineering team vs LLMs and some prompters.
All I can hear is the nihilistic AI generated song at the end of the Malus talk. (https://fosdem.org/2026/schedule/event/SUVS7G-lets_end_open_...)
Provided the code does the job, it seems immaterial whether it was hand-written or LLM generated, provided it does the job and isn't straight-up plagiarism. The latter is admittedly a hard thing to judge.
- https://domino14.github.io/macondo/ - my AI, written in Go. Mostly meant as a "research" tool and doesn't have a proper GUI yet.
- https://github.com/andy-k/wolges - an implementation of a Scrabble AI from scratch in Rust. Andy is very smart and came up with his own algorithm and data structure that we now use throughout our projects
- https://github.com/jvc56/MAGPIE - a friend started this as a rewrite of Macondo, but in C. A lot of new development has gone into this and I've contributed a bit as well. Also no official GUI, but it is _absurdly_ fast, so fast that we can do nested Monte Carlo simulations. Some algorithms have been backported to Macondo with the help of AI.
These are the main 3, but there are people contributing in other ways, someone else has made a CNN and it has good results (I built a CNN as well in the Macondo source, and it has decent results). We're slowly going to continue to improve this until we can use it in real-time for Monte Carlo evaluations, maybe by translating it into a NNUE architecture.
Currently on Macondo I've been working on AI explainability (it's quite good already) and Bayesian inference with a stats PhD candidate. We are likely going to publish a paper on the latter. It's very mathy but it ends up beating the best bot we have so far around 52% of the time (that's actually a pretty significant edge).
You can read about some algorithms on my blog here: https://cesardelsolar.com (very in-depth discussions on inference, CNN, and my exhaustive endgame/pre-endgame algorithm)
I'm lucky I guess that I've had enough time in this career to experience these moments, and enough experience to know which code I should write and which code really is rote and can be generated (most of the time). Hopefully it keeps the loss of enlightenment to a minimum.
I think the problem goes deeper.
When a community becomes addicted to LLM, I feel uncomfortable interacting with it. It feels like mindless robots and drones are now active. But even aside from this, some people don't seem to understand why AI spam slopping real people with their AI spam, is awful. A few months ago, Michael Niedermayer spam AI slopped a proposal on their mailing list; I don't have the link right now (need to really bookmark all those things) but someone else asked him why he spams other people with unfinished proposals created by AI? I don't know whether Michael understands this as a problem or not, but I think many AI users don't understand the problem when they AI spam slop other people. I think it is a hostile action by those AI users. So I can totally understand communities not wanting that AI spam.
I've also noticed people who have horrible english skills, suddenly have perfect english skills. I suspect AI use, but then I wonder: is that person interacting with me via AI spam slop? I'd much rather deal with broken english if it is autenthic, than people faking it, even IF the produced english may be better than the broken english (but even that is often doubtful, so much AI spam slop is just wasting time and awful to read).
The whole AI trend seems to lead to really dumber people and fewer real interactions. That is sad, even more so as many people lost real social cues these days anyway.
https://www.cbc.ca/news/canada/new-brunswick/bill-oliver-ai-...
Are we going to end up a bunch of meatbag puppets of the machines? Makes me wonder.
You find abdicating your speech to an AI to be disqualifying, but to people who’ve already abdicated everything they do to AI, it’s folksy and relatable.
A lot of woodworkers scoff at the use of CNC routers; there's a lot that scoff at the use of any power tools at all!
I don't really understand though why folks have this *need* to assert that there is only One True Way to do something and to poo-poo on someone else's interests. It all seems like straw man arguments to me. Even the title of TFA BORN AGAINST, OR WHY HOBBY PROGRAMMING COMMUNITIES ARE AGGRESSIVELY AGAINST LLM USAGE seems like ragebait.
Most hobby fandoms have a dedicated core group who think like this: The end result is only an artifact or social proof to them. The real product is their own prestige of having built it and attracted an audience. There's always an IRC channel or Discord somewhere where there's a social hierarchy of who should be respected for their work.
Outside of those inner circles, most people don't care. The average emulator user doesn't care about the details about who wrote their emulator or the history behind it all the forks. They want to play a game and the emulator core that gets the job done, because the emulator is the product to them. The mastery of the author is not a product to them.
This has happened in multiple technical hobby communities I've been in. With 3D printing I even remember communities where Prusa was looked down upon for making 3D printers too accessible. The argument was that now anyone could buy a 3D printer without really understanding the pain of mastering calibration, fine-tuning, and troubleshooting. That debate exploded to another level when Bambu came along and made 3D printing even more mainstream.
Now it's happening with programming communities. There are some where hand-written code is truly the actual product: Competitive programming, code golf, and obfuscated programming competitions come to mind. Using an LLM here is about as appropriate as submitting LLM generated art to an art contest, which is to say it's not.
There are a lot of other domains where the mastery of the craft isn't interesting outside of a small social circle. It's interesting that they mentioned emulators, because emu communities are famous for having a ton of strange drama that most emulator users don't care about. If you're inside of the small community and spend every day on their Discord, it might be the only thing you care about, but you don't realize that you're a minority because everyone else in the Discord thinks the same way.
OSDev is another interesting one to mention, because even Linus Torvalds had to make a big post telling the anti-LLM people to chill out. There seems to be a split between the purists who think the primary value is their own mastery of being able to write the code, and the pragmatists who are adapting to use new tools in conjunction with their knowledge. If you find yourself in one of the purist camps you might think everyone feels the same way, but outside of that, the world is moving on.
I usually feel turned off by communities where social status is the primary currency. There's always a hidden social hierarchy that feels threatened whenever it feels like the rules my start changing. They react by trying to force the social currency back to the system where they spent years investing their time. Nobody wants to feel like you're losing a foothold because the rules are changing. I got a strong feeling from this article that this author feels they've developed some level of respect in some forum and they dislike the idea of that investment being devalued by a changing landscape
> In traditional niche dev circles, respect is earned slowly through years of activity in their respective fora,
Instead some people consider it the lowest form, cheating etc. Let them talk, and we'll see who comes up with the most useful programs. I loved programming for the 30+ past years, in many different languages. But in the end what I love most is creating useful programs. So I use LLMs to take care of the nitty gritty, pushing my creative vision through, iterating until I have something worthy of publishing. Haters will hate. I'll just keep working.
I say bring it on.
Vs. what I do. I've always done tons of little quirky automation type things with bash -- that do actually strongly become a part of my life.
Today? Yes, that -- but more and faster. I'm actually typing out way less code, but getting more of what I'd like done, and which couldn't be done without some bash (and now python*) knowledge.
Before AI, knew something about python but never typed e.g. more than 20 lines of it myself. I now am able to use it extensively regardless because, like with lots of languages, reading is easier than writing.
> Moreover, you are giving absolutely nothing back to the community, which is central to open-source engine dev. You are only taking.
What would that user prefer? That the developer push a PR upstream...that presumably would be shot down and the user subject to further harrassment because it would be LLM derived at minimum? It seems like a catch-22 where the only winning move is not to play.
That thread also highlights the overlap between "people who are against LLM usage for coding" and "people who are very particular about AGPL."
Presumably no one is forcing the maintainer at gunpoint to have an LLM slightly shake up other people's code, relicense it under an incompatible license and then claim that "every single line of code is written by Claude Code", so I don't see the issue with not playing if that's the only option.
If you can't make contributions that anyone finds interesting enough to include upstream and have no plans to publish your AGPL derived code under a compatible license, you can keep your changes to yourself. Nothing lost, nothing gained, license honored.
If a human is never able to work through or process failures themselves they are bypassing one of the most important steps in the learning journey
More AI tools need to be made the opposite of this, humans should be able to achieve great learning USING AI.
The important message also remains, understand your code! It might just save you one day ;)
I happen to think the alternative world where no one’s skill capital can be devalued sucks more, but it’s hard to build a community around that because the same hours that build skill also build taste, and when you remove the former you remove the latter.
So you get a community with garbage in it. Haha!
I'm not sure of the truthiness there wrt LLMs: using an LLM to code your hobby OS doesn't actually gain you the skills of the existing members of that niche. Maybe those existing members are just annoyed at posers, not at newcomers?
I think the key word to all this is "communities." If you can make something alone, just do it. If you enjoy gathering together to build a bigger thing, often the bigger thing is 1/3 an excuse to gather together. In that way, the LLM is like revealing the answer to a murder mystery game too early and expecting it still to be fun for everyone.
Yeah, the whole point is the street cred of doing something really hard.
I’ve selected a path where I think building tools with AI is fine. But I want people to play my games knowing I created all of it.
In my free time; I'm iterating on programming ideas using coding agents.
For example a testing framework that separates test case definitions from instrumentation...
And the recent post on bento.page sent me down a path exploring different ideas around "X as standalone HTML+JS"...
And I don't care about shipping this; It's just for me.
This is not mutually exclusive with LLM usage. It's only prerequisite is that the human cares enough to learn.
Others look only at the results, so LLM are only a tool (actually without a clearly instruction manual i suspect).
But programming (and so many others works) is not only about making, but learning and developping skills. When an LLM is not available? i think differences will be evident.
And roughly speaking: to check the result of LLM you have to be more capable than the bot, otherwise, who is the boss? And to develop those skills, you have to do it yourself.
I always thought that when I retired I’d do woodworking. But perhaps I’ll also code - to remember what it was like when I was young.
Similarly, even if I don't write the code but I let someone else who I trust do it for me, there is value for me in that person not using LLMs, because only if they don't, I can know that the only spirit that lives in the code is of the person of my choice. It's like a social thing; I can know and trust a human, but I'm not sure I can do that with an LLM.
It's similar to how you wouldn't want to use LLMs in politics: the whole point of expressing your ideas is that they are your ideas and if you let someone (or something) else do it for you, details will be missed and misunderstood.
So this is one reason that I think hobby communities don't like LLMs; they care too much about what they build.
People say, if you you write elaborate well spec'ed prompts, it will do the right thing. Coding in a way is doing that.
But then again I don't program as a hobby to impress others online. Maybe that's the difference?
At least against as it is now, where everyone with Claude suddenly becomes a software engineer.
Software engineering is more than pushing code to main/master.
It's requirements. Acceptance criteria. Design. Documentation written for humans. Reviews. Operations. Maintenance. Understanding the product. Understanding the system you're changing.
Right now, I struggle with PRs being opened by people who don't fully understand the code they're changing, replying to review comments with Claude generated answers pasted back like a meat proxy...
What I see at work is that we're pushing code faster than ever, while losing the big picture of our product, or the architecture of our own network, even faster.
We're also generating documentation at an incredible pace. The problem is that the people writing it often don't understand it either.
Code has become cheap.
Engineering hasn't.
This hits the nail on the head.
AI isn't well liked in a lot of places. I agree with the communities - make something yourself, don't ask a robot to do it for you. If you vibe slop something into existence, I'm never going to look at it. Something you made yourself though? Hell yeah I want to see it.
Another example is the art community - use AI/LLMs there, you'll be excommunicated.
The only exception is real photographs of some event, like I bought my friend a picture of the Falcon 9 taking off against the backdrop of the sun. Amazing photo.
I feel like obsessing over some artist I’ll never be friends with and the story behind their art is just another way to have a parasocial relationship, and the new tools allows us to have absolutely bespoke art in every single home. Things like this used to be considered strange and irregular before mass media. Now every person can make satisfying music. It’s incredible and amazing and should be celebrated because suddenly everyone can participate.
Thanks to taxis, everyone can drive.
That's decoration, not art.
> Now every person can make satisfying music. It’s incredible and amazing and should be celebrated because suddenly everyone can participate.
No one should be celebrating the theft of music from artists. Everyone can participate in making music today just as they always have throughout human history - learn to sing or play an instrument.
That "make" is doing some heavy lifting there. Even handwaving all the ethical stuff with art and LLMs, Christopher Nolan doesn't "make" music when he contracts Hans Zimmer and gives him some guidance. Not I "make" a cabinet when I do the same with a woodworker.
And who's that satisfying for? For the one clicking the "generate" button? For the Sumo people getting VC money?
I say this not as a complaint, but to explain my perspective on using LLMs for personal projects. Granted, I'm publishing none of them, and not trying to make anyone care about them, so it's not really a response to the article or general discourse here per se, but I thought it bears mentioning anyway.
I'd like to buy into the idea that "doing it yourself" is more meaningful or personally edifying, but I don't get that feeling. I want to get that feeling, but I don't. I've found my private waste-of-time projects using LLMs have been more gratifying than my private waste-of-time projects that don't, and at the end of the day, neither really substantively matter anyway.
Again, this isn't really related to communities per se, but it is an answer why some folks use LLMs for projects/"hobbies".
Obvious sarcasm aside, these LLMs have been trained (without consent!) on our hard work, on our passion. People have been writing code for years, and many have open-sourced their projects to share their knowledge, to better the world, to provide free solutions to problems. All that hard work, all that effort, has been stolen and used to create a machine that replaces us. And when we complain that quality goes down, that work is no longer enjoyable, that we don't actually see huge noticable productivity improvements across the board, bright minds like you tell us "lol just quit man fuck you". Do you not feel bad that you are actively destroying the livelihood of the very people that made your slop machine viable to build?
So then the question becomes what "copying" means. There is an extensive body of legal precedent on this matter, and it mostly all points to "transformative uses" being exempt from copyright. For example, if I count the number of words in Fahrenheit 451, that number is not copyrighted. If I reproduce just a sentence, or maybe even a paragraph, that also doesn't count. But at a certain threshold, once I've reproduced enough of the book verbatim, I have violated copyright. Training an LLM is also a transformative use which means that it is simply not covered by copyright.
I use LLMs to build some nice customized and some basically standard tools so I can go from learning about code to learning Rust et al properly ... the next 2 years are scheduled for me. I want code and LLM supply chains as a proper hobby, partially because I find the overall "transformers" approach pretty bad since there is no implicit path towards intelligence and zero chance to build one on top explicitly. it's just a bunch of parrots but parrots are awesome and i'm glad I get to have so many.
Also: I really don't care what the communities think, though ... it's 2026 and all that code, math, science, engineering didn't build any special amount of good, "bad" ( vs ugly personalities ) or cool characters that we didn't have already and it didn't cause any "mutations" in schools of thought either, or drive movements where they were needed.
but I'm a bystander, which wasn't my choice, unfortunately, so y'all know much better what I'm talking about.
most players in the lower leagues won't dope but some do and take bribes anyway, what?
it's a cowardly new world, everybody, happy Esoteric Right Wing Hanukkah!
If you only want to learn programming for work then you obviously have a different perspective than people writing chess engines from scratch for fun instead of using an existing one, right?
You dont understand that people have hobbies they enjoy doing? Because thats the entire point of the article.
I realize there are people with young kids or 2-3 jobs that really have little to no downtime, but it seems that for most people time isn't really the issue - it's just are you more of a passive or active type of personality.
More dangerous are communities who shun AI and treat every disclosure like a Scarlett letter. Generally the loudest opinions come from consumers (fan bois) and not creators. I think this will eventually drive new innovators away from these communities. These vocal naysayers have likely never experienced being a project manager, lead developer, and QA for an agent writing code. Instead they think a prompt magically creates software, but without wearing all these hats at once, you get something far worse than 'slop'. Call me crazy, but I think these multi-hat 'developers' using an agent and are dipping their toe into LLM development are going to raise the bar and leave us vim users in the dust.
- people who like writing code for codes sake
- people who like writing code to build things
In the past, these folks have lived more or less in harmony unified under the common interest of applied maths, engineering and architecture that is necessary to write things that require a lot of vode. Aside for a few classic and very public spats.
Now that community is fractured because llms are increasingly making writing code irrelevant. If that weren't true, we wouldn't be having this convo at all.
I think it's perfectly reasonable to want to cultivate spaces that preserve that craft, but wish they wouldn't be so obnoxious about it. I imagine that a good handful of these folks in the "no llm ever camp" collectively rolled their eyes at vegans and animal rights advocates at some point, and are now feeling being on the unpopular side of progress.
Personally I think it's a cool prospect to have Trappist Monk-like communities producing hand crafted code. It will probably be really good and sturdy. We'll look at it in 10 years and say, "damn they really don't make it like that anymore!"
So of course they are upset by it, it's annoying to spend years building up some skill and then end up being less capable at applying that skill than a 12 year old with a claude code subscription who has never even seen a CRT. Their entire hobby community is going to fundamentally change and the people who are influential and knowledgeable now are effectively has-beens who will have to learn totally new skills to stay relevant and most likely won't have the energy, or desire, and lets be frank the aptitude to learn those skills. (I don't mean the skill of using an LLM, they can do that and almost certainly would be great at it, I mean skills like charisma and presentation and promotion.)
Heck, if everything is worthless, please, please try not to do anything :)
The thing I'm saying is worthless is making a game in 2026 for a game system that people literally couldn't give away for free in 1998. Due to nostalgia and the fact so many were thrown in the trash surviving systems now have some limited cash value. Newly made games do not have a nostalgia value. The reason people do it is to prove they can and to show other people. Anyone can now (no proof needed really, it's super easy) and nobody cares. It's like climbing Everest, and Claude is your Sherpa. Nobody is impressed by it anymore.