I agree with the author that -- right now -- we're still in the part of the AI adoption / product development curve that it's an extreme force multiplier.
I like to think of it as a normal distribution, the further away a programmer is to the right of the mean, the more their benefit. It's almost like it's their standard deviation squared (σ²). So someone like Matt Perry (as OP mentioned), who is a >99.99% programmer for argument's sake and is therefore four standard deviations away from the mean... Matt gets a (4×4) 16x multiplying effect on their productivity.
Someone who is a slightly above average programmer might see a 2 or 3x boost on their productivity, which is huge(!) and might also make them fear for their job. Which tracks with the level of moral panic we are seeing and experiencing. This math kinda still holds up for "bad programmers" too (i.e. left of the mean), as in they still see a boost to their productivity (negative squared is a positive number)... but there's something iffy about their results. The technical debt is unmaintainable and because they don't _understand_ the systems that they're operating in, they end up in the "3 hour" prompt loops that the OP refers to.
> Similarly, if Matt Perry handed me the keys to the Motion repository and told me to take over, I wouldn’t have the same results even though I have access to the same set of LLM tools.
The question is -- how long is this multiplier going to exist for? Some people would wager "for the foreseeable long-term future"; some people think it will widen further; and some people think it will diminish or god forbid even collapse. It feels like most arguments at the moment (like this article's) are that the humans who "know what they are doing" will be able to baton the hatches and avoid being usurped by ever-capable models. I saw it in a café yesterday: someone was using a coding agent to build a marketing website for their project, getting more and more frustrated by not getting the outcome they wanted. Their friend typed a couple of sentences on their keyboard and got a "Dude! How did you do that? That was sick!" a minute or so later. "I used to build websites" the friend said. -- The friend 'knew what they were doing'.
How much longer is knowing what you're doing going to be a moat?
> How much longer is knowing what you're doing going to be a moat?
For a looooonnnnngggg time, unless there's massive progress in AI research.
Fundamentally, next token prediction is limited. Granted, I'm pretty amazed at how well it's done, but if you can't activate the right parts of the models (with your prompts), then you're not going to get good results.
And to be fair, for lots of things this doesn't matter. Steve in Finance or Mindy in Marketing can create dashboards that actually help them, and the code quality mostly doesn't matter.
For stuff that needs to be shipped, monitored and maintained you still need to know what you're doing.
We also need to consider the price. At some point the price will need to go up (assuming cost of producing each token doesn’t drop dramatically) to generate enough revenues to cover not only operating expenses and taxes (once the nol carry forward’s are used up) but reinvestment. OAI and Anthropic are burning through their cash balances. OAI has also stated some very ambitious plans to develop models beyond just programming… I will be very intrigued to see how they are going to generate enough revenue to fund all this in the future.
> How much longer is knowing what you're doing going to be a moat?
To me, I don't see how this will ever not be an advantage. All software requires constraints. Some of those constraints might be objective (scale, performance, etc.) but a lot of them are subjective and require active decision making (architecture, UI, readability).
So if there was only one way to do something or only one desired output, then yes I think models would surpass humans. But like art, I don't think there is a objective truth to software and because of that, humans get the opportunity to play an important role.
Now whether that is valued from a business/industry perspective is a question that I think we all know the answer to unfortunately.
100% agree with this. I think takes like OP's would be much more interesting if they staked out a position in the future. I think it's pretty uncontroversial to say that someone with a great deal of technical expertise is going to be a hugely more effective LLM user today.
The question that really matters is whether that will continue to be the case. My guess is that technical expertise matters less over time, and the ability to specify the desired outcome is eventually the only thing that becomes important. But I could be wrong! The direction this all goes is pretty fuzzy in my mind.
> My guess is that technical expertise matters less over time, and the ability to specify the desired outcome is eventually the only thing that becomes important
if you look at LLMs based coding as another step up in programming abstraction then it's clear this is the case. Think about the progression of programming languages. Over time, we go further and further from the hardware and closer and closer to specifying the desired outcome. The terminology, structure, and completeness of a user story that guides a codingagent to the desired output, and only the desired output, is the new programming language.
> if you look at LLMs based coding as another step up in programming abstraction then it's clear this is the case. Think about the progression of programming languages. Over time, we go further and further from the hardware and closer and closer to specifying the desired outcome. The terminology, structure, and completeness of a user story that guides a codingagent to the desired output, and only the desired output, is the new programming language.
But that entire narrative follows from one, single, very big "If". It is not a given that AIs are a step up in abstraction.
Like, copying the answers in a test isn't considered an abstraction, I don't consider copy-pasting AI into your codebase an abstraction.
in the case of tools like claudecode there's no copy/pasting. Claudecode updates files directly, runs tests, starts/stops server, everything else on its own (with your permission).
I guess to take it a step further, you can lay your requirements in order with guidance in a markdown file called 'myprogram.md'. Then tell ClaudeCode to read that file and do what it says. In that way, myprogram.md, actually your requirements doc, is the programming language being turned into the 1s and 0s the computer understands.
I think this is one of the key takes right now. I too have similar experience.
Which way is it going to go?
i) “Seniors” also get superseded by even more capable models that can do all of the things which currently require experience.
ii) Linguistics become the new higher order abstraction (English is the new high-level programming language) _but_ there are different / orthogonal ways of approaching software development than the way we do things now — which “juniors” become more adept at more quickly.
There's also iii) people realize that if the LLM needs that much babysitting, it doesn't actually add value. So they don't use it very much because it is too limited as a tool.
Exactly, much better with just "we do what we do, and the consequences are what they are" without (potentially mis-)attributing it to things like military service or whatever their past experience may be.
In 2026 it is probably not theoretical physics. Around 2000 or so the dam broke and there have been numerous routes to quantum gravity which are plausible from a calculational point of view... but not a lot of experiments to rule them out except for those that are the most blatantly Lorenz violating. Until we see sparticles, proton decay, and the like, we can't say that much about the whole world of GUT and Strings that we're stuck in right now.
It is a pity that you can't make an experimental commit on an experimental branch without igniting a fire of delirium through some people who -- if they were able to put their emotional response aside for a minute and could weigh this up on the basis of merit -- would probably agree with the motivations for researching this approach.
> if/how hard it’d be to get it to pass Bun’s test suite and be maintainable
Every month brings new opportunities to completely abstract the process of porting code with agents, all using linguistics. What an exciting time.
For those looking for a similarly interesting (and interestingly similar) example, see Cloudflare's port of Next.js[0], "vinext", from a couple of months ago. It had some teething problems at the start but I'm using it in a few production projects now with minimal issues.
I'm mostly on board with what you're saying, but under such an interpretation of "forcing", people are never truly forced into anything. That puts it in fundamental conflict with the very existence of the word, i.e. renders it meaningless.
That said, I did also walk away from most mainstream platforms already, so it's not like I disapprove of the message necessarily. I did find it regrettable that the calculus worked out that way though, and I don't find it reasonable to deny that there is / was a calculus. You do give up on things that are not just the assholes. I'd definitely classify that as a force.
But maybe I'm just missing that this was supposed to be inspirational rather than literal, and mistook your words. I don't know.
I am a topic starter, and I had no emotional response, was just being curious. Never expected it will land at HN #1. I specifically posted the link to the first commit and not to the whole branch, because currently the prompt is the most interesting part.
The branch name is "claude/phase-a-port", there was zero indication this was an experiment until Jarred commented. The more accurate title might have simply been "there is a branch in the official repo of bun describing a port to rust from zig". No amount of soft titles would have prevented the discussion. People have their opinions about Bun, about Zig, about Rust and it's all going to come out in a discussion board.
Can’t every branch be considered an experiment? I have a ton of experimental branches that I don’t label «experimental». One of the reasons you use git…
Sure, but then how does it change anything around the discussion? You are still running an experiment to port to Rust, it still gets posted, the Rust-heads and Zig-heads still make their comments.
> there was zero indication this was an experiment
The goal of Phase A is a **draft** `.rs` next to the `.zig`
that captures the logic faithfully — it does **not** need to compile. Phase B
makes it compile crate-by-crate.
I mean, it would be hard to spell it out any clearer than that! Code that fails to compile is just not very useful for real work.
Phase B clearly says compilation is the next goal. The first goal is to get a like for like logic, the second goal is to get it to compile. Can you guess what the third goal will be? Throw out the code?
Yes, but that would require people to read past the title. You can't get a proper knee-jerk first post in if you do that! Completely unfair to expect people to make that sacrifice/effort.
[there was some sarcasm there, BTW, if anyone has a faulty detector that didn't pick up on it]
I couldn't use that title because I didn't know if it an experiment at the moment. Even now the correct title would be "Bun author says that he is entertaining the idea of porting it from Zig to Rust, creates an experimental branch".
An original topic starter? I'm pretty sure that this was originally posted on X by someone else, as I commented there, and minutes after, it was copied and put here on HN with the twisted title; the original was more of a "question, surprise tone"
This topic starter. I saw a post on Twitter in "for you" feed, verified it, found an interesting bit (rewriting prompt) and started a topic on HN. Like I said, I never expected it to hit #1.
It’s annoying for the team members I suppose, but to be fair, if you’re working on a high-profile open source project, owned by one of the most hyped companies in the world, and your branches are public, it’s probably a good idea to be clear in the branch naming and supplemental files if you’re just “experimenting”.
By working in public on a popular open source project, you are communicating intent and purpose to your users and the general public through your commit messages, branch names, and documentation. You’ll save yourself a lot of grief if you act accordingly.
The fact someone who works on Bun is willing to create and even push a branch generated by a stochastic parrot is very telling of the direction the project is going.
Doesn't matter if it's "experimental", it's a dumb experiment that shouldn't exist.
Why are you treating branches as if they are holy? This is all OSS, people work on this in their free time, git is got and people can use branches as they like to experiment and share their experiments with others. If you don't like the code, don't use it you damn leech.
Underplaying AI, overselling what an experimental branch is, and suggesting it's representative of the entire project, all while suggesting people shouldn't even consider new tools and methodologies. Where to start.
I think that was a very constructive comment about the unconstructive way people are shoe-horning other concerns about bun into this thread abut a specific aspect which itself turns out to be just an experiment that someone knee-jerk reacted to, despite several active threads already discussing those matters one of which only just fell off the front page.
While the concerns many have about Bun's potential future direction are valid IMO, of the posts on this thread the one you are criticising is one of the more constructive.
I've been working on an adjacent problem (extracting website branding data from a URL) for the past year, and previously had to rely on procedural techniques such as these -- props to the author!
However, models are now getting to the point where we are starting to learn the bitter lesson[0] even with stuff like color-palette generation. Nano Banana 2 [gemini-3.1-flash-image-preview] especially is adept at performing arbitrary operations on images. Before then, you would have to use a model such as Gemini Flash to perform segmentation[1] and then post-analyze those segments.
Here's a prompt I used with Nano Banana 2 in AI Studio
> Derive a coherent, designer's color palette from this image alone.
> Provide 5 distinct HEX color codes as your response.
[Attachment == the picture of the car, first in the author's article]
[Settings: Output .. images & text; Thinking level .. minimal]
Response:
> I have extracted five distinct hex color codes directly from the key elements in this image, representing the colorful facade and the vintage car:
> #FF96C5 (The main pink wall)
> #38C6F1 (The light blue car)
> #AEF6A5 (The green wall)
> #E51988 (The dark pink trim and railing)
> #5F432B (The dark wood of the door and windows)
And they all pretty-much check out. Not hyper-accurate, but really not far off anymore. I didn't even have to try!
The bakery example is interesting, because it's presented as "both sides have been working on this thing and think they should get 50%"... and then the _solution_ is "A path back to 50% for Daniel" -- who gets an objectively worse deal than his disputant.
It's definitely an interesting application of LLMs, the output text to me reads very GPT-ey, with the punctuated and concise phrasing.
I like to think of it as a normal distribution, the further away a programmer is to the right of the mean, the more their benefit. It's almost like it's their standard deviation squared (σ²). So someone like Matt Perry (as OP mentioned), who is a >99.99% programmer for argument's sake and is therefore four standard deviations away from the mean... Matt gets a (4×4) 16x multiplying effect on their productivity.
Someone who is a slightly above average programmer might see a 2 or 3x boost on their productivity, which is huge(!) and might also make them fear for their job. Which tracks with the level of moral panic we are seeing and experiencing. This math kinda still holds up for "bad programmers" too (i.e. left of the mean), as in they still see a boost to their productivity (negative squared is a positive number)... but there's something iffy about their results. The technical debt is unmaintainable and because they don't _understand_ the systems that they're operating in, they end up in the "3 hour" prompt loops that the OP refers to.
> Similarly, if Matt Perry handed me the keys to the Motion repository and told me to take over, I wouldn’t have the same results even though I have access to the same set of LLM tools.
The question is -- how long is this multiplier going to exist for? Some people would wager "for the foreseeable long-term future"; some people think it will widen further; and some people think it will diminish or god forbid even collapse. It feels like most arguments at the moment (like this article's) are that the humans who "know what they are doing" will be able to baton the hatches and avoid being usurped by ever-capable models. I saw it in a café yesterday: someone was using a coding agent to build a marketing website for their project, getting more and more frustrated by not getting the outcome they wanted. Their friend typed a couple of sentences on their keyboard and got a "Dude! How did you do that? That was sick!" a minute or so later. "I used to build websites" the friend said. -- The friend 'knew what they were doing'.
How much longer is knowing what you're doing going to be a moat?