Like a young child coming home from kindergarten with their latest crayon scrawls, the internet is currently awash with people sharing their AI-generated work. And just like the young child’s drawings, much of that work should be proudly put up on the walls within the artist’s house—and no further.
Turns out this post hit a nerve :) You can follow the comment threads on both Hacker News and lobste.rs.
Turns out this post hit a nerve :) You can follow the comment threads on both Hacker News and lobste.rs.
Prologue: I ❤️ AI 🔗
I am not an AI-hater. In fact, I think AI-haters are on the wrong side of history.
It’s just that I know when to keep my crayon-drawings to myself ;) And I am getting increasingly sad and frustrated seeing communities that I value slowly wilting under the onslaught of shit. Often that shit is perhaps naïvely shared with no deleterious intent, but shit nonetheless it is.
Congratulations, you entered a prompt and pressed return. 🔗
"I rewrote Kafka in COBOL"
"I rewrote Kafka in COBOL"
Great, enter it at your next science fair. Meanwhile stop begging for stars on your brand new GitHub repo that no-one’s touching with a bargepole.
"I wrote a blog post about Kafka"
"I wrote a blog post about Kafka"
Did you though? We can tell that Claude wrote it, and it’s a piece of garbage.
"I made this video about Kafka"
"I made this video about Kafka"
Cool story bro. Except AI made it, and it’s only of interest as a curiosity, not a useful learning artefact.
"I’m self-publishing an ebook that I wrote about Kafka"
"I’m self-publishing an ebook that I wrote about Kafka"
What you mean is, you got Claude to scrape the internet and crap out a "book" that you should be ashamed to give away for free.
Like a young child coming home from kindergarten with their latest crayon scrawls, the internet is currently awash with people sharing their AI-generated work. And just like the young child’s drawings, much of that work should be proudly put up on the walls within the artist’s house—and no further.
Turns out this post hit a nerve :) You can follow the comment threads on both Hacker News and lobste.rs.
Turns out this post hit a nerve :) You can follow the comment threads on both Hacker News and lobste.rs.
Prologue: I ❤️ AI 🔗
I am not an AI-hater. In fact, I think AI-haters are on the wrong side of history.
It’s just that I know when to keep my crayon-drawings to myself ;) And I am getting increasingly sad and frustrated seeing communities that I value slowly wilting under the onslaught of shit. Often that shit is perhaps naïvely shared with no deleterious intent, but shit nonetheless it is.
Any fool can feed coins into a fruit machine and pull the arm.
Step 0: Profit 🔗
The pattern I see over and over seems to be:
Step 1: Discover agentic coding. Mind blown.
Step 1: Discover agentic coding. Mind blown.
Step 2: Chuck a project up onto GitHub (if it’s actually up ).
Step 2: Chuck a project up onto GitHub (if it’s actually up ).
Step 3: Have AI write a breathless blog post about your vibe-coded project. Share blog post and repo to any subreddit and Slack group that you can find. Not sure which is suitable? Post to all of them—people will love to see it! /s
Step 3: Have AI write a breathless blog post about your vibe-coded project. Share blog post and repo to any subreddit and Slack group that you can find. Not sure which is suitable? Post to all of them—people will love to see it! /s
/s
Let me tell you now: pause after step 2. Take a really long breath. Think really hard about what you’ve created, and why you want to share it. If it’s "because it’s cool" then I’ve got news for you: agentic coding is no longer a novelty. It’s just how shit gets done now.
If you can think of the prompt, AI can write it. Big deal. That’s so early-2026. Move on.
Still want to share it far and wide? Is it actually useful? Are you using it? Has it got really good documentation? Is it usable? Have you actually come back to the code again and again and put it through its paces? Or was it a one-night stand with Claude and the next morning neither of you thinks it was such a good idea?
Still want to share it? If it’s software, are you prepared to stand behind it as something people will raise issues against, maybe submit PRs for? If it’s written, is it something you’d want to read? Is it actually adding to the cumulative understanding of the community, or is it just an LLM auto-completing its way through text that you can’t be arsed to write and I can’t be arsed to read?
Who cares? 🔗
No one forces me to read this stuff. Why am I so bothered by it?
Because like bindweed, it’s slowly strangling the organic life out of communities. When I open up Reddit now, it’s increasingly overrun with vibe-coded AI stuff. Whilst much of it is well-intentioned I’m sure, it does nothing to contribute to the community.
AI slop is driving up the noise, and making the signal more and more difficult to discern in communities. This risks becoming a downward spiral; as communities become more polluted by this stuff, members will get frustrated from wading through AI slop and draw back, thus diminishing the life of the organic community even further.
Carrying on like this, online communities will either wither and die, or converge on something like the dystopian-but-banal MoltBook in which AI agents "talk" to each other with no humans present.
There’s good 'slop' and bad slop 🔗
You may have noticed that AI Slop has become the mot du jour.
AI Slop
The broad use of the term that I’m generally familiar with is as a negative description for low-effort material created by AI and foisted upon those to whom it is of no benefit. However I learnt recently that there are those—probably correlating strongly with the AI-hating crowd—who brand anything written about AI as "AI Slop", even if not written by AI.
Material created with the assistance of AI is not bad in itself. It’s the purpose to which it’s put.
A good use of AI is when it enables people to do something they couldn’t do before, to contribute to a community when they couldn’t before. Done with the care and good intent of a human behind it, this is a nett positive.
Bad AI slop, on the other hand, is monkeys throwing crap over the fence for a purpose other than furthering the community. This includes spam, engagement farming, and simply thoughtless noise in a space which is not for that purpose.
OK, but who made you gatekeeper of the internet? 🔗
The standards of common decency and taste, that’s who.
Let’s take a step back. Sharing content online is wonderful. It’s pretty much what made the internet what it is today.
The knack is to understand what you’re sharing, to whom, and why.
If you were born before around 1980 you’ll know that there was the Geocities era. Every high-school nerd had a homepage (mine was in Vienna since you’re asking).
Just because I built a homepage on Geocities, complete with 'Under Construction' anigifs, a web counter and a web ring banner, does not mean that I should be sharing it to anyone who’ll listen. Amongst my friends, sure. My parents, of course—they’ll be proud of anything I build. But to the general internet? Who cares.
And now with AI-generated content, whether a vibe-coded app or a blog post, the same applies. The internet went through a collective convulsion in early 2026 as everyone discovered the power of Claude Opus 4.5 (and don’t get me wrong, it is damn cool). And what does anyone do when they discover anything neat? They want to share it with their friends!
Combine that with the deafening AI-hype machine of grifters already in overdrive—and suddenly subreddits and Slacks are overrun with AI-generated material.
So what should I share? 🔗
Built with AI, not by AI 🔗
This excellent section heading is taken from my friend and colleague Gunnar Morling’s recent article. As I outlined above, AI is a powerful tool, and I will argue with anyone for the case that it’s pretty much a dereliction of one’s job to not be including it in one’s toolbox. Gunnar nails the nuance though:
Build with AI.
AI is just a tool.
You need to do the thinking, the instructing, the checking.
Gunnar has built a fantastic new project (Hardwood; a new parser for Apache Parquet), using AI. Does that mean it falls foul of my wrath and judgement? No, of course not. It’s a project that’s taken four months so far, with a solid roadmap, a burgeoning community, and a thoughtful and careful design behind it.
Contribution 🔗
Does your offering contribute anything to the community?
If you boil it down to its essentials, is what you’re sharing anything other than the manifestation of a prompt fed into an agentic coding tool? If I took your prompt and ran it, would I end up with something similar? Prompt engineering is fun and an interesting study, but it’s tangential to the subject itself. Consider a community of ornate furniture enthusiasts (I’m sure such a thing exists); it’s the equivalent of bombarding them with Ikea-esque pieces simply because you’ve got a really interesting set of chisels that you want to show off.
Just like I’m not posting my kid’s drawings off to the National Gallery just yet, I’m also not sharing every cool app that I can build with Claude. Not that "software is art" (though some of the best actually is), but there’s nothing much interesting in the puerile output of a process. Anyone with a few tokens can prompt their way to a bit of software. Throwaway tools are just fine. They’re great, in fact—the internet is built on weird little scripts that people have built and shared. But chuck them on gist/GitHub—they don’t need a launch blog post as if you’re the incarnation of Steve Jobs.
Respect the community 🔗
This is a tale as old as time. Well, the internet anyway, once it got beyond ARPANET and BBSes.
Whether Usenet, Reddit, lobste.rs, or any other online platform, the netiquette is always to "lurk". Hang around, read what gets written, get a feel for "the vibe".
I’m not the arbiter of what’s acceptable in a given community. The community members are. Vibed an amazing new implementation of the Kafka protocol, but not sure if people want to see it? Read the room, and get a feel for whether they’ll welcome with open arms your AI slop—or not. If in doubt, ask!
As well as lurking, another way of showing respect to the community is to be very open and clear about if, how, and where you’re using AI in your contribution.
The Asymmetry of Bullshit 🔗
What’s the impact on others of your contribution?
The amount of energy needed to refute bullshit is an order of magnitude bigger than that needed to produce it.
The amount of energy needed to refute bullshit is an order of magnitude bigger than that needed to produce it.
If you splurge out a gobbledegook article, you’re putting that workload onto your reader to realise that it’s not worth wading through. If you dump a complex PR into a project without due care, you’re obligating the reviewers to go through the code and for them to explain to you why it can’t be merged. In both of these scenarios the community would be better off without your contribution.
Unfortunately the sheer volume of AI slop (and I mean that in the bad way) has meant that communities and projects are struggling to deal with the impact of contributions, in some cases even slamming their doors shut on anything that’s touched AI. Projects like Vouch have sprung up to try and address this, but the Rubicon has been crossed and it’s unclear where things will end up.
Pre-AI, the effort required for contributions was sufficient proof of work to either deter people or demonstrate an actual commitment. Communities could deal with sub-par contributions. Those well-intentioned and willing to learn could be mentored and often would develop into important members of the community. Those less well-intentioned and not doing much more than spamming could be dealt with because the volume was so low.
With great power… 🔗
or perhaps that should be "With a great number of tokens"
Communities are powerful yet fragile things. Don’t be the bindweed that suffocates the life out of them.
Explore with great joy the power that LLMs and agentic coding tools bring. Enjoy the frisson of jfc that is cool that it invariably brings.
But respect the community, and only share what is truly relevant. Save the crayon pictures for your kitchen fridge.
Bindweed photo by Joshua Ralph on Unsplash. All other pictures by my kids. Which is ironic, given my exhorting for people to literally keep their childish drawings to themselves ;)
Bindweed photo by Joshua Ralph on Unsplash.
Bindweed photo by Joshua Ralph on Unsplash.
All other pictures by my kids. Which is ironic, given my exhorting for people to literally keep their childish drawings to themselves ;)
All other pictures by my kids. Which is ironic, given my exhorting for people to literally keep their childish drawings to themselves ;)
Congratulations, you entered a prompt and pressed return. 🔗
"I rewrote Kafka in COBOL"
"I rewrote Kafka in COBOL"
Great, enter it at your next science fair. Meanwhile stop begging for stars on your brand new GitHub repo that no-one’s touching with a bargepole.
"I wrote a blog post about Kafka"
"I wrote a blog post about Kafka"
Did you though? We can tell that Claude wrote it, and it’s a piece of garbage.
"I made this video about Kafka"
"I made this video about Kafka"
Cool story bro. Except AI made it, and it’s only of interest as a curiosity, not a useful learning artefact.
"I’m self-publishing an ebook that I wrote about Kafka"
"I’m self-publishing an ebook that I wrote about Kafka"
What you mean is, you got Claude to scrape the internet and crap out a "book" that you should be ashamed to give away for free.
Any fool can feed coins into a fruit machine and pull the arm.
Step 0: Profit 🔗
The pattern I see over and over seems to be:
Step 1: Discover agentic coding. Mind blown.
Step 1: Discover agentic coding. Mind blown.
Step 2: Chuck a project up onto GitHub (if it’s actually up ).
Step 2: Chuck a project up onto GitHub (if it’s actually up ).
Step 3: Have AI write a breathless blog post about your vibe-coded project. Share blog post and repo to any subreddit and Slack group that you can find. Not sure which is suitable? Post to all of them—people will love to see it! /s
Step 3: Have AI write a breathless blog post about your vibe-coded project. Share blog post and repo to any subreddit and Slack group that you can find. Not sure which is suitable? Post to all of them—people will love to see it! /s
/s
Let me tell you now: pause after step 2. Take a really long breath. Think really hard about what you’ve created, and why you want to share it. If it’s "because it’s cool" then I’ve got news for you: agentic coding is no longer a novelty. It’s just how shit gets done now.
If you can think of the prompt, AI can write it. Big deal. That’s so early-2026. Move on.
Still want to share it far and wide? Is it actually useful? Are you using it? Has it got really good documentation? Is it usable? Have you actually come back to the code again and again and put it through its paces? Or was it a one-night stand with Claude and the next morning neither of you thinks it was such a good idea?
Still want to share it? If it’s software, are you prepared to stand behind it as something people will raise issues against, maybe submit PRs for? If it’s written, is it something you’d want to read? Is it actually adding to the cumulative understanding of the community, or is it just an LLM auto-completing its way through text that you can’t be arsed to write and I can’t be arsed to read?
Who cares? 🔗
No one forces me to read this stuff. Why am I so bothered by it?
Because like bindweed, it’s slowly strangling the organic life out of communities. When I open up Reddit now, it’s increasingly overrun with vibe-coded AI stuff. Whilst much of it is well-intentioned I’m sure, it does nothing to contribute to the community.
AI slop is driving up the noise, and making the signal more and more difficult to discern in communities. This risks becoming a downward spiral; as communities become more polluted by this stuff, members will get frustrated from wading through AI slop and draw back, thus diminishing the life of the organic community even further.
Carrying on like this, online communities will either wither and die, or converge on something like the dystopian-but-banal MoltBook in which AI agents "talk" to each other with no humans present.
There’s good 'slop' and bad slop 🔗
You may have noticed that AI Slop has become the mot du jour.
AI Slop
The broad use of the term that I’m generally familiar with is as a negative description for low-effort material created by AI and foisted upon those to whom it is of no benefit. However I learnt recently that there are those—probably correlating strongly with the AI-hating crowd—who brand anything written about AI as "AI Slop", even if not written by AI.
Material created with the assistance of AI is not bad in itself. It’s the purpose to which it’s put.
A good use of AI is when it enables people to do something they couldn’t do before, to contribute to a community when they couldn’t before. Done with the care and good intent of a human behind it, this is a nett positive.
Bad AI slop, on the other hand, is monkeys throwing crap over the fence for a purpose other than furthering the community. This includes spam, engagement farming, and simply thoughtless noise in a space which is not for that purpose.
OK, but who made you gatekeeper of the internet? 🔗
The standards of common decency and taste, that’s who.
Let’s take a step back. Sharing content online is wonderful. It’s pretty much what made the internet what it is today.
The knack is to understand what you’re sharing, to whom, and why.
If you were born before around 1980 you’ll know that there was the Geocities era. Every high-school nerd had a homepage (mine was in Vienna since you’re asking).
Just because I built a homepage on Geocities, complete with 'Under Construction' anigifs, a web counter and a web ring banner, does not mean that I should be sharing it to anyone who’ll listen. Amongst my friends, sure. My parents, of course—they’ll be proud of anything I build. But to the general internet? Who cares.
And now with AI-generated content, whether a vibe-coded app or a blog post, the same applies. The internet went through a collective convulsion in early 2026 as everyone discovered the power of Claude Opus 4.5 (and don’t get me wrong, it is damn cool). And what does anyone do when they discover anything neat? They want to share it with their friends!
Combine that with the deafening AI-hype machine of grifters already in overdrive—and suddenly subreddits and Slacks are overrun with AI-generated material.
So what should I share? 🔗
Built with AI, not by AI 🔗
This excellent section heading is taken from my friend and colleague Gunnar Morling’s recent article. As I outlined above, AI is a powerful tool, and I will argue with anyone for the case that it’s pretty much a dereliction of one’s job to not be including it in one’s toolbox. Gunnar nails the nuance though:
Build with AI.
AI is just a tool.
You need to do the thinking, the instructing, the checking.
Gunnar has built a fantastic new project (Hardwood; a new parser for Apache Parquet), using AI. Does that mean it falls foul of my wrath and judgement? No, of course not. It’s a project that’s taken four months so far, with a solid roadmap, a burgeoning community, and a thoughtful and careful design behind it.
Contribution 🔗
Does your offering contribute anything to the community?
If you boil it down to its essentials, is what you’re sharing anything other than the manifestation of a prompt fed into an agentic coding tool? If I took your prompt and ran it, would I end up with something similar? Prompt engineering is fun and an interesting study, but it’s tangential to the subject itself. Consider a community of ornate furniture enthusiasts (I’m sure such a thing exists); it’s the equivalent of bombarding them with Ikea-esque pieces simply because you’ve got a really interesting set of chisels that you want to show off.
Just like I’m not posting my kid’s drawings off to the National Gallery just yet, I’m also not sharing every cool app that I can build with Claude. Not that "software is art" (though some of the best actually is), but there’s nothing much interesting in the puerile output of a process. Anyone with a few tokens can prompt their way to a bit of software. Throwaway tools are just fine. They’re great, in fact—the internet is built on weird little scripts that people have built and shared. But chuck them on gist/GitHub—they don’t need a launch blog post as if you’re the incarnation of Steve Jobs.
Respect the community 🔗
This is a tale as old as time. Well, the internet anyway, once it got beyond ARPANET and BBSes.
Whether Usenet, Reddit, lobste.rs, or any other online platform, the netiquette is always to "lurk". Hang around, read what gets written, get a feel for "the vibe".
I’m not the arbiter of what’s acceptable in a given community. The community members are. Vibed an amazing new implementation of the Kafka protocol, but not sure if people want to see it? Read the room, and get a feel for whether they’ll welcome with open arms your AI slop—or not. If in doubt, ask!
As well as lurking, another way of showing respect to the community is to be very open and clear about if, how, and where you’re using AI in your contribution.
The Asymmetry of Bullshit 🔗
What’s the impact on others of your contribution?
The amount of energy needed to refute bullshit is an order of magnitude bigger than that needed to produce it.
The amount of energy needed to refute bullshit is an order of magnitude bigger than that needed to produce it.
If you splurge out a gobbledegook article, you’re putting that workload onto your reader to realise that it’s not worth wading through. If you dump a complex PR into a project without due care, you’re obligating the reviewers to go through the code and for them to explain to you why it can’t be merged. In both of these scenarios the community would be better off without your contribution.
Unfortunately the sheer volume of AI slop (and I mean that in the bad way) has meant that communities and projects are struggling to deal with the impact of contributions, in some cases even slamming their doors shut on anything that’s touched AI. Projects like Vouch have sprung up to try and address this, but the Rubicon has been crossed and it’s unclear where things will end up.
Pre-AI, the effort required for contributions was sufficient proof of work to either deter people or demonstrate an actual commitment. Communities could deal with sub-par contributions. Those well-intentioned and willing to learn could be mentored and often would develop into important members of the community. Those less well-intentioned and not doing much more than spamming could be dealt with because the volume was so low.
With great power… 🔗
or perhaps that should be "With a great number of tokens"
Communities are powerful yet fragile things. Don’t be the bindweed that suffocates the life out of them.
Explore with great joy the power that LLMs and agentic coding tools bring. Enjoy the frisson of jfc that is cool that it invariably brings.
But respect the community, and only share what is truly relevant. Save the crayon pictures for your kitchen fridge.
Bindweed photo by Joshua Ralph on Unsplash. All other pictures by my kids. Which is ironic, given my exhorting for people to literally keep their childish drawings to themselves ;)
Bindweed photo by Joshua Ralph on Unsplash.
Bindweed photo by Joshua Ralph on Unsplash.
All other pictures by my kids. Which is ironic, given my exhorting for people to literally keep their childish drawings to themselves ;)
All other pictures by my kids. Which is ironic, given my exhorting for people to literally keep their childish drawings to themselves ;)