We've seen this cycle before with high-level languages or IDEs, where the bar for entry drops but the underlying complexity of building reliable systems remains the same.
I write code in my work, and enjoy it. I also write less directly because of llm's but get to read more (which I luckily also enjoy, as long as I understand what is going on). I think you made more of a snide comment and deep down you actually enjoy it ;-). I mean, who doesn't like to code. It's awesome. If not, i hope you find something fulfilling sooner rather than later.
>It's not the end of programming, in the same way as the invention of the printing press wasn't "the end of drawing by hand".
The printing press let you scale copies of something already drawn. That is nothing like what AI is doing here which is making drawing easier. It's more like people now draw by hand using pens and pencils instead of scraping small rocks against bigger rocks.
>It's not the end of programming, in the same way as the birth of big tech wasn't the beginning of programming.
Considering big tech brought programming to the masses arguable big tech did begin programming as a thing people do.
Every other time that programming has gotten easier we got more (often worse) code, though some folks who made their livings exploiting the inefficiencies had to adapt. This will be the same.
This isn't the end of programming, it's the beginning of your broccoli haired nephews ability to program. It's also opening a whole new world of opportunity to those with real skill.
A world filled with mountains of terrible code is a world that will reward the very programers who are trying to fight that future today.
> This isn't the end of programming, it's the beginning of your broccoli haired nephews ability to program
Which is funny because that was the case 25+ years ago. We (the nephews) would create websites using PHP. It was common for 15, 16 years old teenagers to know more about tech than the adults. That's not the case anymore. So this time it might be different.
Yeah right, like I want to take over the shit broccoli haired nephews created... Even for 3x paycheck.. The one who takes it is a looser, being exploited for pennies to fix all the shit. And I bet at the end, you will be fired anyway, because you are not of their kind..
Winners are managers (who hired nephews) and nephews themselfs, they did barely any usefull job and got paid a lot. They should do usefull job instead, like drivers, warehouse jobs, services, farming and other less paid but very importand and usefull jobs.
> Winners are managers (who hired nephews) and nephews themselfs, they did barely any usefull job and got paid a lot.
That is what the hype is about. They created their kingdom/empire with software and data. And they’ve been looking for any solution that promises to replace programmers. LLM is the best mirage ao far and have grown to cult level.
If someone comes with a pretty decent idea and execution, it doesn’t really matter if code was written by human or llm, does it? Like if the next torvalds ships something great in the open, who cares how was it written.
> it doesn’t really matter if code was written by human or llm
In theory no, but for a successful execution you must have people that understand what's going on, otherwise the competitor that understands will eat you.
> who cares how was it written.
In the pre AI era a lot of sw was written by interns, juniors, some bigcon consultants, some offshore indian or east-european dudes, it didn't even matter before.
> Like if the next torvalds ships something great in the open, who cares how was it written.
I was reading a biography of Tovalds and one of the core things that mark the birth of Linux was the drive to learn. Same when reading “Coders at Work”, the will to learn and make things happen is a common recurrence.
I’ve never seen LLM coding that is not “I can’t bother to learn about this”. It’s always about magicking answers out of thin air, not about motivation to explore some fields.
Many interesting metaphors to find, "it's like the printing press", "it's like IDE's", "it's like going from hardwired programming to assembly language". At the same time it's unlike anything we've seen before. Interesting to see where the journey goes, I am along for the ride.
People keep giving examples like printed press, or whatever, but honestly the printed press did kill writing/drawing by hand. When was the last time you bought a hand written book or a hand painted picture?
Sure, they still exist, but they are priced and marketed towards a specific class of people, and not the mass. And with software it’s even harder as software was never targeted at the mass. Software was always a by product of a car, a tv, or your fridge. No ordinary folk ever said “I’m going to go a buy WinZip now”. They simply had a variant of WinZip available with whatever computer they bought. There is simply no category of people who are willing to pay for “high quality hand crafted software”, except for perhaps a few ideologically motivated people.
Software development “lost” because it was seen as “something that comes for free with a computer”, and not something that people have to actively buy, like they buy TVs, fridges, or paintings.
I’m confused by your example. Prior to the printing press, hand written books (or any reading material) would not have been accessible to the masses; hence the low literacy rates.
Furthermore, while no one was saying “I’m going to go buy WinZip now”, many were lining up to buy WinRAR?
Finally, the issue with “high quality hand crafted software” is that software and similar projects are made possible through engineering not through artistry.
Simply put:
"Engineering is achieving function while avoiding failure." — Henry Petroski
> When was the last time you bought a hand written book or a hand painted picture
So programming "ended" if it's no longer such a dominant profession, even if plenty of people still do it as an art/craft/hobby and some of them even get paid to do so?
Doesn't sound like the "end of programming" to me, more a reduction in the number of programming jobs. One that if anything is overdue - many developers today get paid to do jobs that shouldn't need to exist even without AI.
> And with software it’s even harder as software was never targeted at the mass.
Do products like Adobe Photoshop or Microsoft Office Suite not count as “targeted at mass?” People—normal people—absolutely used to pay for software before SaaS took hold.
I don’t think the problem is that normal people don’t want to buy ““artisan software”” but that people are so cynical about software companies that giving them any money at all is asking to have the rug pulled out from under you.
Printing press did not kill writing at all, only the job of scribes. One still had to write their ideas down and work on a manuscript and some one had to set the press up (not trivial at all). All that is to say that I think the ability to write code from scratch is still necessary while I don't really think the printing press metaphor is apt I think being able to write scripts and write small programs to do stuff you want by hand is still useful even though it's true that LLM's can do this sometimes you want to understand the thing fully and to work exactly how you want it.
And seem to hold in contempt wanting to understand and think. So many just want to parrot what they hear wealthy people say and the idea of human though activity and research and contemplation being obsolete feels them with a twisted sense of joy.
The printing press killed the work of scribes, people who copied books page by page. It was a form of art and craftsmanship. It didn't disappear completely, but 99% of production moved to the printing press.
Programming will follow the same path. Typing code by hand will be relegated to 1%, while 99% will be generated by assistants and agents.
I do not think programming is dead, but because I don't think LLMs can really do the job unassisted. I agree with you that the printing press is a dumb argument, basically tantamount to saying programming is dead except for bespoke artisanal programs to buy at ren faires and hacker conferences.
It accepts the inevitability premise so completely that I wonder if the author is secretly an AI booster trying to make the other side look bad.
Typing code by hand is dead I agree. But calling it the end of programming was bit weird though.
I use to write a lot of code everyday for more than 2 decades, but now with LLMs, I can get most of my code written by LLMs, and I make sure they write the code that I want to. Same as, when I had to manage a team of juniors.
I find it interesting just how different the responses to this new tech are.
I wonder if there is an inverse correlation between "usage of local models" and "despair", or if I am just making that up.
It would make sense, I suppose, given that local models still come with quite the upfront cost. So a lot of people are experiencing LLMs purely through the user-hostile cloud.
the technology isnt the problem at all. it's just being used as an thin pretext to devalue and treat devs like shit by the followers of a new executive religion.
Script kiddies aren't really a problem, but there's a place where technically competent CTO types can sooner/inexpensively solve problems with AI rather than documenting what they want done.
I think AI brings in many problems. It also brings in new functionality for people unable or unwilling to write working programs on their own.
My number #1 gripe of AI is that you become dependent on external companies. I am no loner willing to go that route in general, but even less so when it comes to greedy US corporations that constantly abuse mankind.
My number #2 gripe is that AI is ... dumb. I am noticing this on youtube. So many fake-videos appearing, and while I in general do not watch them for the most part, they are incredibly deceitful. And ... dumb. It's just a waste of time to watch them, even more so than it is already a waste of time watching videos on youtube (I often have them just for background music in the browser, for music at the least).
AI also makes projects less interesting to contribute to. If 99% of the contributions come by AI, why would I want to be bothered to "get involved"? Ruby has this problem right now. Granted, AI is a helper tool, but boy, what a fake it is - Hiroshi Shibata suddenly has perfect english skills, but it is AI slop generated. So I am no longer "interacting" with humans but with the AI slop. That's not worth my time.
AI also is very ruthless against programming languages being bad in certain areas. Case in point: documentation. If AI can slop AI spam generate documentation, even if it is not perfect, whereas I need to manually search for ruby documentation while Google has crippled its search engine to become useless, then this is now a huge disadvantage I am at with. Searching for stuff takes time. Ruby is just an example here; you can extend this to many other projects, including non-programming languages. We need better documentation, but oddly enough the documentation is constantly getting worse in the last +5 years or so. AI skynet is cannibalizing us - people don't seem to realise or care. I somewhat gave up on trying to nudge people to seriously improve their documentation, they just don't seem to want to (not all, but most).
I feel like the article in question was actually saying some like others are expressing here that writing code by hand is over not the other aspects despite the title.
I also think the metaphor of drawing is not perfect but a quite good in this comment in that drawing is talking about something that is missed on lots of analysis, drawing is a free form kind of sketching and one can quickly draw something and work things out. I think being able to quickly code something by hand is still valuable and though you can get an llm to write a script sometimes you just want to write it yourself so you know the ins and out of your 100 cloc script.
I think that coding by hand is largely not going to occur anymore. What that does to the learning of new developers + the stagnation of old developers? I do not know.
I believe that what makes a good developer (a great developer, really) is the ability to do something at the smallest level "perfectly" or close to it. Chaining those tiny building blocks together. Understanding the entire flow of data + logically. Making something as fast as it can be. When you put more and more of those tiny blocks together, things get a little more imperfect - but a fantastic developer can keep them all perfect for a long time.
We're jumping way above that level. That small, perfect logic process, that visibility is now completely hidden from us, some of us wouldn't even know how to do it anymore. And juniors of today will never learn to.
And maybe they don't need to. Perhaps we will just build "bigger" software so that the level of imperfection becomes less noticeable. Larger systems, more functionality, more security.
I think the brain atrophy is going to hit us HARD though. It feels too fast to rip away the one thing that taught us all of this. Building something piece by piece was the way most of us learned. Then you can jump into bigger and bigger projects and your brain creates those pathways. Previous generations had months / years to learn new tooling - and now we threw ALL of the tooling away and said "use this instead". I think good developers will always find a way to learn and grow, but man is it going to take some time, we completely shifted the way the entire industry produces the machine in the matter of 12 months.
I still believe that code is the best way to represent a sufficiently complex specification, natural human language feels very ambiguous sometimes, like extra lossy compression (whatever that is worth).
It's really interesting to the vagueness of natural language and the fact that it's actually impossible to express a single concrete meaning in natural language. That is to say their is always a set of possible interpretations one can never simple refer to one single thing and a program in a programming language is always only one thing (I know that you could argue that compiling to different machine code means their is vagueness since it might produce a x86-64 binary or a powerpc binary and the issue of UB but you see my point).
I do think that if one simply 'vibe codes' something they can never really understand the program even if they prompted it unless they study the source in detail. In natural language you can only ever point to a set of things that containing a subset of thing(s) you want.
The responses about code quality, wisdom that comes with experience, etc. are missing the big picture: no one will ever care about any of this as long as things work as explained in the prompt.
Eventual shortcomings will be fixed not by looking into the code, but by typing in another prompt, asking another agent.
Doing it manually will not bring any value to anyone.
I saw it already with my own eyes: a friend of mine started an online psychology counseling site. A year ago they would hire a developer. Now, he only needed an LLM. And it’s the whole thing: a booking system, video calls, a website. What’s underneath: he doesn’t know. The only thing that matters to him is that it works.
Gotchas aside: I’m not saying the end of programming is a good thing. I’m just saying: it’s happening. Unless the LLM tech will suddenly hit any sort of ceiling.
66 comments
[ 0.31 ms ] story [ 47.8 ms ] threadI write code in my work, and enjoy it. I also write less directly because of llm's but get to read more (which I luckily also enjoy, as long as I understand what is going on). I think you made more of a snide comment and deep down you actually enjoy it ;-). I mean, who doesn't like to code. It's awesome. If not, i hope you find something fulfilling sooner rather than later.
The printing press let you scale copies of something already drawn. That is nothing like what AI is doing here which is making drawing easier. It's more like people now draw by hand using pens and pencils instead of scraping small rocks against bigger rocks.
>It's not the end of programming, in the same way as the birth of big tech wasn't the beginning of programming.
Considering big tech brought programming to the masses arguable big tech did begin programming as a thing people do.
This isn't the end of programming, it's the beginning of your broccoli haired nephews ability to program. It's also opening a whole new world of opportunity to those with real skill.
A world filled with mountains of terrible code is a world that will reward the very programers who are trying to fight that future today.
Which is funny because that was the case 25+ years ago. We (the nephews) would create websites using PHP. It was common for 15, 16 years old teenagers to know more about tech than the adults. That's not the case anymore. So this time it might be different.
Winners are managers (who hired nephews) and nephews themselfs, they did barely any usefull job and got paid a lot. They should do usefull job instead, like drivers, warehouse jobs, services, farming and other less paid but very importand and usefull jobs.
That is what the hype is about. They created their kingdom/empire with software and data. And they’ve been looking for any solution that promises to replace programmers. LLM is the best mirage ao far and have grown to cult level.
In theory no, but for a successful execution you must have people that understand what's going on, otherwise the competitor that understands will eat you.
> who cares how was it written.
In the pre AI era a lot of sw was written by interns, juniors, some bigcon consultants, some offshore indian or east-european dudes, it didn't even matter before.
Companies used to release broken software on purpose since a very long time.
nope. but, vibe coded slop is all we're actually seeing.
I was reading a biography of Tovalds and one of the core things that mark the birth of Linux was the drive to learn. Same when reading “Coders at Work”, the will to learn and make things happen is a common recurrence.
I’ve never seen LLM coding that is not “I can’t bother to learn about this”. It’s always about magicking answers out of thin air, not about motivation to explore some fields.
…when was that?
Sure, they still exist, but they are priced and marketed towards a specific class of people, and not the mass. And with software it’s even harder as software was never targeted at the mass. Software was always a by product of a car, a tv, or your fridge. No ordinary folk ever said “I’m going to go a buy WinZip now”. They simply had a variant of WinZip available with whatever computer they bought. There is simply no category of people who are willing to pay for “high quality hand crafted software”, except for perhaps a few ideologically motivated people.
Software development “lost” because it was seen as “something that comes for free with a computer”, and not something that people have to actively buy, like they buy TVs, fridges, or paintings.
Furthermore, while no one was saying “I’m going to go buy WinZip now”, many were lining up to buy WinRAR?
Finally, the issue with “high quality hand crafted software” is that software and similar projects are made possible through engineering not through artistry.
Simply put:
"Engineering is achieving function while avoiding failure." — Henry Petroski
So programming "ended" if it's no longer such a dominant profession, even if plenty of people still do it as an art/craft/hobby and some of them even get paid to do so?
Doesn't sound like the "end of programming" to me, more a reduction in the number of programming jobs. One that if anything is overdue - many developers today get paid to do jobs that shouldn't need to exist even without AI.
Do products like Adobe Photoshop or Microsoft Office Suite not count as “targeted at mass?” People—normal people—absolutely used to pay for software before SaaS took hold.
I don’t think the problem is that normal people don’t want to buy ““artisan software”” but that people are so cynical about software companies that giving them any money at all is asking to have the rug pulled out from under you.
Programming will follow the same path. Typing code by hand will be relegated to 1%, while 99% will be generated by assistants and agents.
The 1% will be 100x bigger than the 100% of 10 years ago.
It accepts the inevitability premise so completely that I wonder if the author is secretly an AI booster trying to make the other side look bad.
I wonder if there is an inverse correlation between "usage of local models" and "despair", or if I am just making that up.
It would make sense, I suppose, given that local models still come with quite the upfront cost. So a lot of people are experiencing LLMs purely through the user-hostile cloud.
Getting a really really artificial vibe about the content being pushed to this website :/
Making me feel less motivated to even come here to be honest
My number #1 gripe of AI is that you become dependent on external companies. I am no loner willing to go that route in general, but even less so when it comes to greedy US corporations that constantly abuse mankind.
My number #2 gripe is that AI is ... dumb. I am noticing this on youtube. So many fake-videos appearing, and while I in general do not watch them for the most part, they are incredibly deceitful. And ... dumb. It's just a waste of time to watch them, even more so than it is already a waste of time watching videos on youtube (I often have them just for background music in the browser, for music at the least).
AI also makes projects less interesting to contribute to. If 99% of the contributions come by AI, why would I want to be bothered to "get involved"? Ruby has this problem right now. Granted, AI is a helper tool, but boy, what a fake it is - Hiroshi Shibata suddenly has perfect english skills, but it is AI slop generated. So I am no longer "interacting" with humans but with the AI slop. That's not worth my time.
AI also is very ruthless against programming languages being bad in certain areas. Case in point: documentation. If AI can slop AI spam generate documentation, even if it is not perfect, whereas I need to manually search for ruby documentation while Google has crippled its search engine to become useless, then this is now a huge disadvantage I am at with. Searching for stuff takes time. Ruby is just an example here; you can extend this to many other projects, including non-programming languages. We need better documentation, but oddly enough the documentation is constantly getting worse in the last +5 years or so. AI skynet is cannibalizing us - people don't seem to realise or care. I somewhat gave up on trying to nudge people to seriously improve their documentation, they just don't seem to want to (not all, but most).
Software engineering will remain.
I also think the metaphor of drawing is not perfect but a quite good in this comment in that drawing is talking about something that is missed on lots of analysis, drawing is a free form kind of sketching and one can quickly draw something and work things out. I think being able to quickly code something by hand is still valuable and though you can get an llm to write a script sometimes you just want to write it yourself so you know the ins and out of your 100 cloc script.
I believe that what makes a good developer (a great developer, really) is the ability to do something at the smallest level "perfectly" or close to it. Chaining those tiny building blocks together. Understanding the entire flow of data + logically. Making something as fast as it can be. When you put more and more of those tiny blocks together, things get a little more imperfect - but a fantastic developer can keep them all perfect for a long time.
We're jumping way above that level. That small, perfect logic process, that visibility is now completely hidden from us, some of us wouldn't even know how to do it anymore. And juniors of today will never learn to.
And maybe they don't need to. Perhaps we will just build "bigger" software so that the level of imperfection becomes less noticeable. Larger systems, more functionality, more security.
I think the brain atrophy is going to hit us HARD though. It feels too fast to rip away the one thing that taught us all of this. Building something piece by piece was the way most of us learned. Then you can jump into bigger and bigger projects and your brain creates those pathways. Previous generations had months / years to learn new tooling - and now we threw ALL of the tooling away and said "use this instead". I think good developers will always find a way to learn and grow, but man is it going to take some time, we completely shifted the way the entire industry produces the machine in the matter of 12 months.
I still believe that code is the best way to represent a sufficiently complex specification, natural human language feels very ambiguous sometimes, like extra lossy compression (whatever that is worth).
I do think that if one simply 'vibe codes' something they can never really understand the program even if they prompted it unless they study the source in detail. In natural language you can only ever point to a set of things that containing a subset of thing(s) you want.
Eventual shortcomings will be fixed not by looking into the code, but by typing in another prompt, asking another agent.
Doing it manually will not bring any value to anyone.
I saw it already with my own eyes: a friend of mine started an online psychology counseling site. A year ago they would hire a developer. Now, he only needed an LLM. And it’s the whole thing: a booking system, video calls, a website. What’s underneath: he doesn’t know. The only thing that matters to him is that it works.
That to me looks like the end of programming.
So, you'll rely on the tech that created buggy code to somehow suddenly create bug-free code?
Gotchas aside: I’m not saying the end of programming is a good thing. I’m just saying: it’s happening. Unless the LLM tech will suddenly hit any sort of ceiling.