Dec 26, 2025 · 37m · latent-space
Steve Yegge's Vibe Coding Manifesto: Why Claude Code Isn't It & What Comes After the IDE
0:00 / 0:00
▶ Watch on YouTube →
gold bands on the timeline = statements, start to end. Hover to read, click to jump. CC turns on captions
Live from the AI Engineer Summit, veteran engineer Steve Yegge explains how vibe coding, multi-agent architectures, and disposable code are transforming software development and making traditional manual programming obsolete.
How this conversation actually went
Every chapter scored 0–10 on four independent dynamics. Hover any point for the reasoning behind the score. How this is scored →
The hosts as informed peer 3.7 Guest teaching 3.9 Guest disagreement 4.1 The hosts pushing back 3.0
speaking balance: gold is the hosts, purple is the guest (3 minute bins)
Sharpest disagreement ▶ 3:22 IDEs make you a bad engineer
Steve aggressively dismisses traditional development workflows, setting a strict deadline of January 1st before labeling any engineer still using an IDE as bad and out-of-date.
Hardest push from the hosts ▶ 32:08 Saturating curves challenge model convergenceThe host refuses Steve's optimistic framing of rapid open-source convergence by arguing that progress appears faster only because lower-tier models are catching up against an asymptotic ceiling.
Biggest teaching moment ▶ 4:15 The 2,000-hour trust thresholdSteve educates the host on research showing engineers must log over 2,000 hours with agents to accurately anticipate failures like hallucinations and amnesia before they can truly trust them.
The host holds their own ▶ 20:24 Explaining stacked diffs and merge queuesWhen Steve declares that nobody has a solution for high-volume agent merge conflicts, the host counters with industry solutions around stacked diffs from Facebook and recent discussions with GitHub leadership.
the scores for every segment, with the reasoning behind each
| Chapter | Topic | The hosts as informed peer | Guest teaching | Guest disagreement | The hosts pushing back | Why |
|---|---|---|---|---|---|---|
| The Vibe Coding Movement and Senior Engineer Backlash | 3 | 4 | 6 | 3 | Steve aggressively mocks senior engineers with 12-15 years experience who resist vibe coding, recounting an online spat. The host mildly tempers the hot take by noting that companies will inevitably have to manage a mixed workforce of AI adopters and holdouts. | |
| Dropping the IDE and Building Trust in Agents | 2 | 5 | 5 | 3 | Steve delivers strong hot takes declaring developers who still use IDEs 'bad engineers' and 'dinosaurs'. He educates the host on the empirical need for 2,000 hours of agent interaction to develop predictable intuition, while the host questions if veteran engineers will truly become interns. | |
| Why Claude Code Isn't It and Orchestration Dashboards | 3 | 4 | 5 | 3 | The host challenges Steve's controversial view that Claude Code isn't it given its widespread popularity. Steve details why terminal-based diff reading is unsustainable and explains the architectural necessity of orchestration dashboards. | |
| Multi-Agent Systems, Beads, and the Code Merge Wall | 5 | 3 | 4 | 4 | When Steve claims the code merge wall is an unsolved crisis for high-velocity multi-agent coding, the host steps in with domain knowledge on stacked diffs and merge queues used at Meta and GitHub, causing Steve to admit he is unfamiliar with the concept. | |
| The Factory Farming Era and Throwaway Code Philosophy | 3 | 4 | 4 | 2 | Steve introduces the 'factory farming' analogy for code generation and explains why Joel Spolsky's classic rule against rewriting code is dead. The host largely listens and affirms Steve's broad industry experience. | |
| Frontier AI Lab Execution and Open Source Scaling | 5 | 4 | 3 | 4 | Steve provides inside perspectives on execution bottlenecks across Google, Anthropic, and OpenAI. When Steve predicts open-source models will rapidly achieve frontier parity, the host pushes back with an asymptotic saturation counterargument. | |
| The AI Engineer Mindset and the Future of Coding | 5 | 3 | 2 | 2 | The host and guest explore the definition of an AI engineer, with the host adding specific technical context about MCP tool calling discoveries by Cloudflare and Anthropic, framing the role as analogous to high-earning F1 drivers. |
Statements from this episode (17)
Yegge: Senior engineers with 12-15 years experience hate vibe coding most
“It's not junior engineers. It's not non-engineers. They're all vibe coding. It's senior engineers, senior leaders people who have, so basically you can narrow it down to. 12 to 15 years of experience. They hate vibe coding and they hate AI and they're online g…”
Yegge: OpenAI sees a 10x productivity gap between AI adopters and non-adopters
“Anecdotally, they're sharing that performance, the performance differences like 10 X By any way that you measure it. So lines of code, commits, business impact, whatever. And it's so stark and pronounced that the people who aren't adopting it are now 10 times …”
Yegge: Engineers still using an IDE by January 1st are bad engineers
“If you're still using an IDE to develop code by January first. You're a bad engineer.”
Yegge: Developers need 2,000 hours with AI before trusting it
“Jean just pulled up a study that showed that you actually have to spend a year or 2000 hours with AI before you trust it. And what does trust mean? Trust in this case specifically means before you as a user can predict what it's going to do.”
Yegge: Top engineers avoiding AI will operate like interns within a year
“Look, I have friends Who are much better engineers than I am, ok, I mean, world class, maybe some of the best in the whole world, ok, have built technologies that you've heard of, and they're not using AI yet, except the occasional, I'll ask Cursor a chat ques…”
Yegge: IDEs should only run in background for AI agents
“Yeah, so you, all you do is leave IntelliJ running, but you shouldn't look in it. It's a tool for the AI now, right?”
Yegge: Claude Code is too hard for most engineers
“And the answer is it's too hard. It's too hard. You have to be able to read, man, most engineers, honestly, like to them, five paragraphs is an essay. Okay. And with cloud code, you've got to read waterfalls of not just information, but also code and diffs, ri…”
Yegge: Developer tooling will become agent orchestration dashboards
“So what it's going to be is it's going to be your agent orchestration dashboard. It's you're going to walk in in the morning and be like, yo, so how's it going to do, right? It's like, oh, that one's still running. That one's running a tool. That one needs my …”
Yegge: Graphite is best poised to solve the AI code merge wall
“Merging is the, it's the wall that everyone is hitting right now. I think the company that's best poised to solve it is graphite.”
Yegge: 10x AI productivity turns merging into architectural rewrites
“As soon as you get to the point where like every developer is 10 times as productive, merging their code becomes this incredibly complicated problem because I, you and I work at the same time for two or three hours. We make, you know, 30,000 line change each. …”
Yegge: Automated factory farming of code will arrive by summer
“We will still get to factory farming code. With today's model's capabilities, and we'll get there fast, we'll get there by summer, but the models are getting smarter so fast.”
Yegge: Rewriting code from scratch with LLMs beats trying to fix it
“And now we've discovered that it is for a larger and larger and larger class of Piece of bodies of code. It is better to just start over and rewrite it from scratch than it is to try to fix it. The LLM will do a better job.”
Yegge: Google improved execution by finally holding its engineers accountable
“Google has turned it around on, on execution. They finally did the thing that they should have done, you know, 15 years ago, which is hold people accountable. And it's not just engineers do whatever they want all the time, which is what it was for 20 years.”
Yegge: Google, Anthropic, and OpenAI are unbelievably chaotic internally
“All three of those companies, Google, Anthropic, and OpenAI are unbelievably chaotic internally right now.”
Yegge: Anthropic is hiring over 100 people for Claude Code
“They're hiring like a hundred plus people for cloud code in the next, I don't know, month. I mean, like they're going wild and that's just cloud code.”
Yegge: Open source models will match Gemini 3 by next summer
“From what I've heard, they, they're seven months behind, and that, that gap is gradually narrowing. The frontier models, which means OSS models will be as good as Gemini three next summer.”
Yegge: Agents handle MCP better by writing code than direct tool calls
“The agents can't call MCP very effectively because they don't have any training on tool calls, but they have plenty of training on writing code, so you tell them don't call the tool, write code to call the tool, and they do way better with it, right?”