"Humans are now writing code in strict specification language so that AI agents have completely context and don't mistakes. This specification language is called C' and has led to a whopping 20% reduction of code. 1000 of C++ code can be expressed in no more than 800 lines of specification C' code written by humans"
WTF is this?! Sattire? AI generated propaganda? I honestly don't get it. Can OP elaborate why it's a good content worthy of people’s time? Thanks in advance.
> “Autonomous Proxies for Execration, or APEs,” Pluto said.
> “By typing in a few simple commands, I can spawn an arbitrary number of APEs in the cloud,” Pluto said.
> “I have hand-tuned the inner loops to the point where a single APE can generate over a megaBraden of wide-spectrum defamation. The number would be much larger, of course, if I didn’t have to pursue a range of strategies to evade spam filters, CAPTCHAs, and other defenses.”
“Have you tried this out yet?” Corvallis asked.
“Not against a real subject,” Pluto said. “I invented a fictitious subject and deployed some APEs against it, just to see how it worked in the wild. The fictitious subject has already attracted thousands of death threats,” he added with a note of pride.
“You mean, from people who saw the defamatory posts seeded by the APEs and got really mad at this person who doesn’t even exist.”
Make a fictitious subject with all the traits of the person you really want to attack (Subject X). Have your social media bots attack Subject X. Anger spillover on social media will begin attacking your true target by trait association. The real target will have a difficult to impossible time coming at you via legal channels as there is no direct association.
I am ape writing this post after ape cooking breakfast, and then I'll go for an ape walk. In the future, maybe by Thursday, I can have agents do all of that and relax.
Why has nobody mentioned yet how dangerous this really is? Have we all forgotten the great Datacenter burnings of 2031? The APEs are one step away from becoming fully fledged Luddite terrorists. Artisanal software is unamerican just like President Barron said the other day on his Twitch stream.
I really like to understand the practice of software engineering by analogy to research mathematics (like, no one ever asks mathematicians to estimate how long it will take to prove something…).
Something I think software engineers can take from math right now: years of everyone’s math education is spent doing things that computers have always been able to do trivially—arithmetic, solving simple equations, writing proofs that would just be `simp` in Lean—and no one wrings their hands over it. It’s an accepted part of the learning process.
It's not ape coding. It's skill coding. People who don't have the skill to do math and logic ask others to do it for them.
The reason we have programming languages is the same reason we have musical notation or math notation. It is a far more concise and precise way of communicating than using natural languages.
We could write music using natural language, but no one does because a single page of music would require dozens of pages of natural language to describe the same thing.
"Aping in" in crypto means (meant?) buying crypto without doing any research.
I know it's not what the thought piece is about, but it's equally accurate to say engineers are "aping in" on AI coding without doing any research. Very much the same vibe, my anti-AI friends suddenly flipped their tune to shill slopped together apps.
I expect it to go about as well as it did in crypto.
I call it Tradcoding. Not using AI for anything. (You just copy-paste from StackOverflow, as our forefathers once did ;)
I also have two levels "beneath" vibe coding:
- Power Coding: Like power armor, you describe chunks of code in English and it's built. Here you outsource syntax and stdlib, but remain in control of architecture and data flow.
- Backseat Coding: Like vibe coding but you keep peeking at the code and complaining ;)
I dunno; I think Tradcoding would go beyond regular modern coding, and rather imply some kind of regressive Nara Smith "first grind and sift the flour in your kitchen"-style programming.
No Internet connection, no cache of ecosystem packages, no digitized searchable reference docs; you sit in a room with a computer and a bookshelf of printed SDK manuals, and you make it work. I.e. the 1970s IBM mainframe coding experience!
I always thought that ape coding is what we call vibe-coding nowadays. Maybe the write of the article (maybe an ai generated blog?) misunderstood the terms.
It's pretty strange to me that we imagine a world where AI can handle every problem but we still talk about code. It's like how the Jetson's had bulky TVs.
You don't talk about all the assembly high level languages make, or at least it's no longer how people view things. We don't say "look at this assembly I compiled." Instead the entire concept fades to the back.
> You don't talk about all the assembly high level languages make, or at least it's no longer how people view things.
Speak for yourself. I routinely look at assembly when worrying about performance, and occasionally drop into assembly for certain things. Compilers are a tool, not a magic wand, and tools have limits.
Much like LLMs. My experience with Claude Code is that it gets significantly worse the further you push it from the mean of its training set. Giving it guidance or writing critical “key frame” sections by hand keep it on track.
People who think this is the end of looking at or writing code clearly work on very different problems than I do.
The issue is you're measuring this statistic incorrectly.
If you look at the per capita number of people talking about assembly when looking at all the people on the planet it's highly likely there are more people looking at assembly now then whenever your back then was. Programmers simply where a tiny part of the population back then.
Each time we make coding easier and more high level we invite more programmers into the total pool.
I would probably just call it hand coding, as we say we use hand tools in wood working. Many do this for fun, but knowing the hand tools also makes you a better woodworker.
It's an interesting question: Will coding turn out to be more like landscaping, where (referring to the practice specifically of cutting grass) no one uses hand tools (to a first approximation)? Or it will it be more like woodworking, where everyone at least knows where a Stanley hand plane is in their work shop?
I would call it code-plumber. It's like a plumber who are today socio-economocally very distinct from architects, civil and structural engineers.
They will have very narrow to zero understanding — don't need it to fix — of shear forces, navier stokes.
They will command high rates if labor is limited(a plumber in Indonesia will commande lower ppp adjusted hourly rates than America). CS education become a subset of applied math since graduate hiring of code-plumber will require a narrower certificate to fix an AI system — which works very much like how plumber working to fix a building leak is different from a person fixing a water pipe burst under a road.
A few AI systems will become dominant, That should be a mix of Anthropics and your Googles. They will hire code plumbers to plumb together all the things they provide.
You don't have to use much brain at all as a code-plumber. You become a remote journeyman logging in and plumbing with given tools, making sure there is low back pressure(a term where load on future plumbers interacting/fixing with ai decreases) and the like.
61 comments
[ 2.9 ms ] story [ 53.4 ms ] thread“Have you tried this out yet?” Corvallis asked.
“Not against a real subject,” Pluto said. “I invented a fictitious subject and deployed some APEs against it, just to see how it worked in the wild. The fictitious subject has already attracted thousands of death threats,” he added with a note of pride.
“You mean, from people who saw the defamatory posts seeded by the APEs and got really mad at this person who doesn’t even exist.”
Make a fictitious subject with all the traits of the person you really want to attack (Subject X). Have your social media bots attack Subject X. Anger spillover on social media will begin attacking your true target by trait association. The real target will have a difficult to impossible time coming at you via legal channels as there is no direct association.
I really like to understand the practice of software engineering by analogy to research mathematics (like, no one ever asks mathematicians to estimate how long it will take to prove something…).
Something I think software engineers can take from math right now: years of everyone’s math education is spent doing things that computers have always been able to do trivially—arithmetic, solving simple equations, writing proofs that would just be `simp` in Lean—and no one wrings their hands over it. It’s an accepted part of the learning process.
It's so great to be alive in this time of of dehumanizing AI.
The reason we have programming languages is the same reason we have musical notation or math notation. It is a far more concise and precise way of communicating than using natural languages.
We could write music using natural language, but no one does because a single page of music would require dozens of pages of natural language to describe the same thing.
I know it's not what the thought piece is about, but it's equally accurate to say engineers are "aping in" on AI coding without doing any research. Very much the same vibe, my anti-AI friends suddenly flipped their tune to shill slopped together apps.
I expect it to go about as well as it did in crypto.
I also have two levels "beneath" vibe coding:
- Power Coding: Like power armor, you describe chunks of code in English and it's built. Here you outsource syntax and stdlib, but remain in control of architecture and data flow.
- Backseat Coding: Like vibe coding but you keep peeking at the code and complaining ;)
- Vibe Coding: Total yolo mode. What's a code?
No Internet connection, no cache of ecosystem packages, no digitized searchable reference docs; you sit in a room with a computer and a bookshelf of printed SDK manuals, and you make it work. I.e. the 1970s IBM mainframe coding experience!
You don't talk about all the assembly high level languages make, or at least it's no longer how people view things. We don't say "look at this assembly I compiled." Instead the entire concept fades to the back.
Speak for yourself. I routinely look at assembly when worrying about performance, and occasionally drop into assembly for certain things. Compilers are a tool, not a magic wand, and tools have limits.
Much like LLMs. My experience with Claude Code is that it gets significantly worse the further you push it from the mean of its training set. Giving it guidance or writing critical “key frame” sections by hand keep it on track.
People who think this is the end of looking at or writing code clearly work on very different problems than I do.
If you look at the per capita number of people talking about assembly when looking at all the people on the planet it's highly likely there are more people looking at assembly now then whenever your back then was. Programmers simply where a tiny part of the population back then.
Each time we make coding easier and more high level we invite more programmers into the total pool.
It's an interesting question: Will coding turn out to be more like landscaping, where (referring to the practice specifically of cutting grass) no one uses hand tools (to a first approximation)? Or it will it be more like woodworking, where everyone at least knows where a Stanley hand plane is in their work shop?
They will have very narrow to zero understanding — don't need it to fix — of shear forces, navier stokes.
They will command high rates if labor is limited(a plumber in Indonesia will commande lower ppp adjusted hourly rates than America). CS education become a subset of applied math since graduate hiring of code-plumber will require a narrower certificate to fix an AI system — which works very much like how plumber working to fix a building leak is different from a person fixing a water pipe burst under a road.
A few AI systems will become dominant, That should be a mix of Anthropics and your Googles. They will hire code plumbers to plumb together all the things they provide.
You don't have to use much brain at all as a code-plumber. You become a remote journeyman logging in and plumbing with given tools, making sure there is low back pressure(a term where load on future plumbers interacting/fixing with ai decreases) and the like.