I'm gonna miss coding
Thursday, Nov 19, 28 min, about 7 min to read
Theo-brain watched all 28 min
Theo argues that coding as we know it is dead. Claude already writes more than 90% of his own code, and he expects hand-written code to become a niche artisan craft, like Italian shoemakers. Reacting to Lauren's essay, he says AI now rewards curious people, and the skills you built while learning to code still carry over.
Briefing is enoughYou can skip the video. The briefing already covers Theo's main points, including his tip to change the prompt and rerun instead of fixing an agent's code by hand.
Tap any time to jump the video there. Quotes are YouTube's automatic captions, word for word, so some names come out misspelled.
Key takeaways
Theo says coding as we know it is dead, not partly dead. Very few things still need all their code written by a human. When the AI gets a part wrong, you can tell it to fix it or write that bit yourself.
The reality is that coding as we know it is dead now. It's not kind of dead. It is.
Am I saying that there is no code that is best written by hand? Honestly, at this point, kind of. Yeah, there's very very few things that need to have all of their code written by a human still.
Theo agrees with a quote from Lauren's essay. There are still Italian shoemakers, but look at your feet. Hand-written code will become a niche artisan craft, not the code people actually use.
But yeah, like that is that is where code by hand will go is going to weird artisan [bleep] not actually used code. It just it's obviously the case.
Theo's biggest excitement is that AI rewards curiosity. A machine that never says "I don't know" can explain any public code, even on the free tier of ChatGPT. That keeps curious teenagers from burning out for lack of people around them to help.
I think naturally curious people have an unimaginable advantage here more than like ever before because you have a machine that was built to answer your curiosities to encourage you to go deeper to never say sorry I don't know the answer to that to go find it for you anything that exists in public human…
Theo notes that many people thriving with AI tools, including Boris (creator of Claude Code), Lauren, Matt Pocock and himself, come from web development. He thinks web developers got used to tools changing very fast.
Regardless, we all came from WebDev. And a reason I think this matters is we all got used to things changing really fast.
Theo's closing point is that coding is dead, but the skills built while learning it are not. If you bring the same excitement and curiosity to building with AI, it will take you further.
Coding is dead, but the skills we built alongside learning to code are not. And if you can find ways to apply that excitement around building and learning and that curiosity that got us all here to building with AI, it will take you even cooler and better places.
One skill here is new to you.
Re-prompt instead of hand-editing New
- What it is
- Treat getting the right code from an agent like a video game. Some input exists that produces code needing no edits, so you keep adjusting the prompt and context rather than fixing the output yourself.
- Why Theo cares
- Theo likes this idea, quoted from Alex in Lauren's essay. He compares it to a speedrunning trick: try lots of odd approaches, and once one works you never have to think about that task again.
- How to try it
- When your agent's output is wrong, change the prompt or context and rerun instead of editing the code by hand.
- Once a prompt works, save it so the same task can be repeated without new effort.
In theory, there's a series of input tokens that's going to give you the code that you want as an output that doesn't need editing. It's entirely possible. You just need to figure out that series of input tokens.
I like this framing a lot. It's almost like a speedrunning technique. Like you could just go through the part the normal way or you can try a shitload of weird stuff and if you get it right once then you never have to think about…
Advice
Learn to love things moving rather than where things are. Get comfortable with change happening around you, and trust that it almost always leads somewhere better, even if it doesn't at first.
Learn to love things moving, not where things are. It's kind of one of those like journey is better than the destination things, but it it's deeper than that. It's like learning to love the experience of things changing around you and knowing that change will almost always lead to better, even if it doesn't initially.
Be empathetic to developers grieving hand coding, since we owe them a lot. But Theo draws the line at attacking people like Lauren as "vibe coders" without knowing she helped build the React compiler.
Where I draw the line is when they start actively talking [bleep] on other devs who have been there for both. If you go after Lauren for being a vibe coder and don't even know that she built the React compiler, which…
Hot takes
Anthropic saying only 90% of its code is written by AI sounds too low to Theo. He calls it out as nonsense.
His reasoning: Claude already writes more than 90% of his own code.
…since to the point where when Anthropic says that only 90% of their code is written by AI, I call [bleep] because Claude writes more than 90% of my own code.
People furious about AI use, such as a game using AI to fix small bugs before launch, are the same type as teachers who hated students using Wikipedia.
His reasoning: Wikipedia became one of the greatest sources of information ever, and the people who opposed it have not moved much since.
That's what it is. Same teachers who were mad at you for using Wikipedia to find the information you wanted as sources for your papers are the ones who are flipping a [bleep] [bleep] about Big Lock the game using AI in its debug process to fix some small things before it went out.
Being labeled a vibe coder (someone who builds by prompting AI) makes engineers take you less seriously, and being a woman engineer compounds it.
His reasoning: People only get so much space in the public mind. Once you are seen as one thing, how others see your other qualities gets tainted.
I learned this quickly. If you become labeled as a vibe coder, engineers will take you less seriously and you will take up less space in their brains. There are reasons that is good and there are reasons that is bad. But it is reality.
Not knowing how everything already works is an advantage for new builders using AI, even if their low-quality pull requests are annoying.
His reasoning: Not knowing lets them try things no one else would have tried. He cites high schoolers rethinking video editing, animation and games.
There is a unique advantage to not knowing how everything already works because it'll give you the opportunity to try things no one else would have tried.
How Theo thought about this
Theo explains why he feels less loss than other developers. His excitement is more about where things are going than where they are. His best moments were never falling for one fixed pattern. They came when he started to understand a pattern and then something changed. He says he likes the trajectory.
…that my excitement is more for where things are going than where things are because all of my biggest moments weren't like I found an old pattern and fell in love with it. It was I started to understand a pattern and then something happened. I like the trajectory personally.
Theo grounds his idea that the public gives each person limited space in his own experience. Around 100k subscribers, within a week or two, people stopped seeing him as a developer who makes videos and started seeing him as a YouTuber who pretends to code.
…noticed when I hit around 100k subs, it almost felt like a switch flipped and the world no longer saw me as a developer who YouTubes. They saw me as a YouTuber who pretends to code.
Sign in to mark what you've watched and keep notes.
Have a question about this one? Ask about this video
Theo Daily is unofficial and not affiliated with Theo or t3.gg. Written by Claude Opus 5.5 from each video's transcript.
