Post cover image
Ryan Dahl makes an impactful statement on the future of SWE.

January 19, 2026

The creator of Node.js says the era of writing code is over

What Ryan Dahl’s declaration means for the rest of us who still type for a living

By JP Caparas

7 min read

At 5am this morning (down under, that is), Ryan Dahl posted a tweet that's already racked up hundreds of thousands of views:

"This has been said a thousand times before, but allow me to add my own voice: the era of humans writing code is over. Disturbing for those of us who identify as SWEs, but no less true. That's not to say SWEs don't have work to do, but writing syntax directly is not it."

If this came from some random tech influencer, you could dismiss it. Scroll past.

But Ryan Dahl isn't some random tech influencer. He's the person who created Node.js.

The guy who received a standing ovation at JSConf EU in 2009 when he first demonstrated that JavaScript could run on the server. Nine years later, he gave another talk at the same conference titled "10 Things I Regret About Node.js" and then built an entirely new runtime called Deno to fix those mistakes.

None

When Ryan Dahl says something is over, it's worth paying attention.

The man who keeps starting over

To understand why this tweet matters, you need to understand who's saying it.

In 2009, Dahl was a freelance developer who'd become obsessed with a problem: web servers were terrible at handling many simultaneous connections. Most servers used threads, which were expensive. He'd been watching nginx handle thousands of connections with an event-driven architecture and wondered if the same approach could work for application code.

His solution was Node.js. At that first JSConf demo, he showed an IRC server written in JavaScript. The audience gave him a standing ovation. Just like that, web development changed forever.

But here's the thing about Dahl that makes him different from most tech celebrities: he's willing to admit when he got things wrong. Publicly. In 2018, he returned to JSConf EU with a talk that opened with a confession. "I have mass with Node," he said, "because I see bugs that I introduced." He listed his regrets one by one: the way he handled Promises, the security model he chose, the build system, the package.json design. All the things that had calcified into the foundation of one of the most widely-used runtimes in the world.

Then he announced Deno (which I use for some stuff, I'm a Bun dude for the most part), built from the ground up to address those mistakes. Security by default. TypeScript as a first-class citizen. No node_modules folder. He didn't defend his past decisions. He burned them down and started again.

This is someone who's demonstrated, repeatedly, that he cares more about being right than about being consistent.

What he's actually saying

Let's read the tweet carefully. Dahl isn't saying software engineers are obsolete. He's saying writing syntax directly is no longer the job.

There's a crucial distinction here. The work of building software still exists. Understanding problems, designing systems, making architectural decisions, verifying that things actually work. All of that remains. What's changing, Dahl argues, is the mechanical act of translating those decisions into lines of code.

This isn't a new position for him. In June 2025, he published a blog post titled "We're Still Underestimating What AI Really Means" where he wrote:

"We're living through what may be one of the most significant moments in history: the emergence of a new non-biological form __of intellig__ent life. And yet, it doesn't feel like it. There's no cinematic score, no blinking AGI warning light. Just Slack threads, blog posts, and conference panels."

He compared it to witnessing childbirth: "profoundly transformative, with some shocking moments, but also lots of mundane time waiting around."

And he's in the trenches with these tools. In another tweet from June 2025, he complained: "Really tired of hearing 'You're absolutely right!' from Claude." That's not the observation of someone theorising from the sidelines. That's someone using AI coding assistants daily and getting annoyed by their quirks.

Dahl isn't a utopian or a doomsayer. He's a working programmer who's watched the tools change and is calling what he sees.

The titans weigh in

Dahl isn't alone in this view. And he isn't unopposed either.

Andrej Karpathy, the former Tesla AI director, coined the term "vibe coding" in February 2025 to describe exactly what Dahl's talking about. "There's a new kind of coding I call 'vibe coding'," he wrote, "where you fully give in to the vibes, embrace exponentials, and forget that the code even exists." His punchline: "The hottest new programming language is English."

George Hotz, the hacker behind comma.ai and tinygrad, frames it as a change in role rather than a loss of one. "I think the mass of programmers are going to be replaced by AI… The 10x programmers are going to become 100x programmers." he's said. His metaphor: we're becoming directors, not cameramen. The 10x developer, in his view, becomes the 100x developer with AI tools.

Source: Lex Fridman Podcast #387

But not everyone's convinced:

DHH, the creator of Ruby on Rails, has been sharp in his pushback (which has since changed this month). "The pitch is that developers will become 'orchestrators' of AI systems," he wrote. "I've heard this song before. It was sung to mainframe operators in the 80s." His view: fundamentals matter more than ever. "If you can't evaluate AI-generated code, you're not a programmer, you're a prompt typist."

None

… but just this month:

"Yes, I'm ready to give the current crop of AI agents a promotion. They're no longer just here to help me learn, answer my questions, or check my work. They're fully capable of producing production-grade contributions to real-life code bases."

Source: https://world.hey.com/dhh/promoting-ai-agents-3ee04945

None

How the turntables…

Bjarne Stroustrup, creator of C++, echoes the concern about understanding. "Yes, I do have serious concerns about AI. It has a tendency of guiding people towards things which everybody used to do. Furthermore, I fear that people lose the ability to detect problems because they are so used to having it done for them."

Source: DevClass interview, May 2025

And Kent Beck, the pioneer of Test-Driven Development, has taken perhaps the most interesting position. In a post "The va_lue of 90% of my skills just dro__pped to $0. The leverage for the remaining 10% went up 1000x. I need to recalibrate._" he acknowledged that most of what he knew how to do can now be done by AI. But his response wasn't despair. It was adaptation. "The skill is shifting from 'writing code' to 'evaluating code' and 'asking good questions.'"

Source: https://tidyfirst.substack.com/p/90-of-my-skills-are-now-worth-0

Then comes John Carmack, lead programmer of Doom:

"Coding was never the source of value, and people shouldn't get overly attached to it. Problem solving is the core skill."

Carmack views this as natural progression, not a crisis.

Source: https://x.com/ID_AA_Carmack/status/1762110222321975442

The paradox: orchestrators still need fundamentals

Here's where it gets uncomfortable. And honestly, a bit ironic.

If the future is orchestration (telling AI what to build rather than building it yourself) then you might think the fundamentals matter less. Why learn algorithms if Claude can implement them? Why understand memory management if you're just describing what you want?

But nearly every experienced voice, even the optimists, lands on the opposite conclusion.

ThePrimeagen, the Netflix engineer and content creator, is even more direct. His documented position: AI creates "learned helplessness" where juniors ask AI to "fix this line" without understanding why. His worry is a generation of "incompetent programmers" who can't handle complex, bespoke problems — the "Junior Forever" trap.

Source: https://www.youtube.com/watch?v=z-U1r4k5r2s

Jonathan Blow, Creator of Braid & The Witness, says:

Anyone extoling the virtues of AI code generation, who actually knows how to program, will do so with the context of low quality in mind. Anyone running around saying 'the AI just generates all this code and it's great!!' either: (a) Does not understand code and should not be trusted with making decisions about code; (b) If representing themselves as people who understand code, are obvious frauds, or else have so much Dunning-Kruger they don't know how bad they are.

Source: https://x.com/Jonathan_Blow/status/1977480106588410278

What this actually means

So where does that leave us? The working developers who still type for a living?

I think the honest answer is: somewhere uncomfortable.

We're not obsolete. The demand for people who can think clearly about software problems isn't going away. But the specific activity that many of us trained for, the act of translating intent into syntax, is becoming less valuable by the month.

Kent Beck's framing might be the most useful: the skill is shifting. From writing to evaluating. From implementing to verifying. From typing to asking better questions.

Martin Fowler, the software architecture author, makes a related point:

"I'd imagine the AI persona with these traits: eager to help, stubborn, very well-read but inexperienced (high intelligence, low wisdom), won't admit when it doesn't 'know' something."

Source: https://martinfowler.com/articles/exploring-gen-ai/08-how-to-tackle-unreliability.html

That "inexperienced"/"confidently wrong" is the thing. Hell, it might be the thing. AI tools don't know when they don't know. They'll generate plausible-looking code that subtly misunderstands your problem. If you can't see the mistake, you'll ship it.

This is why the fundamentals crowd isn't just being nostalgic. They're being practical. The ability to read code critically, to understand systems deeply, to spot when something is "confidently wrong," these are the skills that let you use AI tools without being led astray by them.

The man who admits his mistakes

What strikes me about Ryan Dahl's tweet isn't just the claim. It's the acknowledgment: "Disturbing for those of us who identify as SWEs."

He's including himself. He's not standing above the disruption, pointing down. He's in it.

That's consistent with who he's been. The person who stood on stage and listed his regrets about Node.js. The person who started over with Deno because he saw the problems he'd created. The person who, in his AI essay, wrote about the Turing test being passed without anyone really noticing: "For most of my life, that milestone felt impossibly far off. When we crossed it, there was no headline. Just another Hacker News thread."

Dahl has been right about things before. He saw that JavaScript could be a server language before most people took the idea seriously. He saw the flaws in his own creation clearly enough to build something better.

Maybe he's right about this too. I suspect he is.

But even if he is, even if writing code by hand becomes as quaint as writing machine code by hand, the underlying work remains. Understanding problems. Designing systems. Verifying that things work. Thinking clearly.

The era of humans writing code might be ending. The era of humans doing the hard thinking that makes software useful? That's not going anywhere.

The question is whether we're ready to adapt. Whether we can let go of identifying as "people who write code" and become "people who build software" instead, regardless of how the building gets done.

Ryan Dahl, at least, seems to think we have to. And he's been willing to start over before.

References