Aider was also some of the most amazing prior art.
After switching from the early copilot (crazy to think they were first to market with this stuff?) it was an amazing TUI experience.
CC won because it offered token hungry devs a fixed price for almost infinite usage in the early days with stacks of VC cash to burn.
Nauseating on so many levels. What are we celebrating, even? Authors themselves say that Claude does their work, so what is their input? It's been barely a year in production for what can be best described as a 'buggy mess'.
To be fair, Pi only had to be useful. Claude Code had to solve the difficult unsolved problem of making a terminal print text, so naturally they built a small game engine [1]
Yes, and to solve the equally tough problem of actually writing the game engine's screen updates to the terminal, they used the log-update library which cleared the screen area it was going to update before writing the text, resulting in horrendous screen flickering.
Now, a mere human might have fixed this by removing the flicker-inducing clear-before-write, but Anthropic had a powerful AI-hammer at hand that was cleverly (after months of deep thought) able to turn this tough problem into a nail-like band-aid which they duly applied... leave the root-cause clear-before-write bug in place, and have the terminal cover it up by buffering and combining the clear+update !
My own experience: much less context bloat, and hidden system prompts. Can tune it to your own workflow, instead of dealing with whatever hidden system prompt Claude Code forces on you.
I didnt personally used OpenCode for heavily, but i remember when i ran it on the server it's TUI was more expencive to run, comparing to PI, so it lagged af. Meanwhile PI doing the same (or even better) job, has very minimalistic and fast TUI, it is like as minimal as possible agent CLI tool.
Claude itself (Opus) gave me the following conclusion:
> The article frames Claude Code as the latest step in a lineage from punch cards → ed → modern editors → AI agents. Multiple team members say they no longer write any code by hand. The overall tone is "we built the right primitive (read/edit/bash), bet on future model improvements, and it paid off."
Honestly I don't want to endure an awkward terminal interface and/or questionable design to just arrive at this conclusion.
“The autonomous software engineering agent vision is more or less coming to fruition.”
Oddly enough no evidence of this is provided. I’ll admit it’s possible to have Claude write 100% of your code - however this is very distinct from autonomy. If I iterate on the code with Claude - telling it what to change, checking for errors and reporting them back to Claude, etc., I can build an app without “writing code” - but that is totally different from autonomy. And it’s not even clear how much of a time saver this really is if I know the language myself.
Incredibly cringe. Literally every single medium mode problem for a team is framed as a groundbreaking achievement. These people snort their own farts.
Please be respectful. I find it incredibly dismissive that you would use such language to describe such amazing individuals. They revolutionized the industry. They built a game engine to render text for cripes sake! So please get it right.
> @dawn-drain — Research Engineer at Anthropic
My main project for my first three years at Anthropic, beginning in 2021, was to try and make a model that was as good at coding as possible— at least as good as me.
Perhaps the most telling part of this transcript is Tristan Hume, a prolific and talented programmer, and presumably an expert performance engineer, is repeatedly saying "this thing just doesn't work that well yet" or "it's just not there yet" and everyone else is kind of just fawning over it.
"It actually made something that worked. But when I tried using it, I found I didn’t like it. I need to wait for a Claude that has the taste ..."
Aider was also some of the most amazing prior art. After switching from the early copilot (crazy to think they were first to market with this stuff?) it was an amazing TUI experience. CC won because it offered token hungry devs a fixed price for almost infinite usage in the early days with stacks of VC cash to burn.
Nauseating on so many levels. What are we celebrating, even? Authors themselves say that Claude does their work, so what is their input? It's been barely a year in production for what can be best described as a 'buggy mess'.
The self-mythologising is strong with this one...
Pi was written by like one guy and it's better than this
To be fair, Pi only had to be useful. Claude Code had to solve the difficult unsolved problem of making a terminal print text, so naturally they built a small game engine [1]
[1] https://x.com/trq212/status/2014051501786931427
Yes, and to solve the equally tough problem of actually writing the game engine's screen updates to the terminal, they used the log-update library which cleared the screen area it was going to update before writing the text, resulting in horrendous screen flickering.
Now, a mere human might have fixed this by removing the flicker-inducing clear-before-write, but Anthropic had a powerful AI-hammer at hand that was cleverly (after months of deep thought) able to turn this tough problem into a nail-like band-aid which they duly applied... leave the root-cause clear-before-write bug in place, and have the terminal cover it up by buffering and combining the clear+update !
Amazing times we live in!
Hall of fame tweet. Can't believe it hasn't even been 6 months
Can you be more specific about how one is better than the other?
My own experience: much less context bloat, and hidden system prompts. Can tune it to your own workflow, instead of dealing with whatever hidden system prompt Claude Code forces on you.
Thanks! What about OpenCode? I'm trying it out right now and would love to learn your thoughts or feedback.
I didnt personally used OpenCode for heavily, but i remember when i ran it on the server it's TUI was more expencive to run, comparing to PI, so it lagged af. Meanwhile PI doing the same (or even better) job, has very minimalistic and fast TUI, it is like as minimal as possible agent CLI tool.
Claude itself (Opus) gave me the following conclusion:
> The article frames Claude Code as the latest step in a lineage from punch cards → ed → modern editors → AI agents. Multiple team members say they no longer write any code by hand. The overall tone is "we built the right primitive (read/edit/bash), bet on future model improvements, and it paid off."
Honestly I don't want to endure an awkward terminal interface and/or questionable design to just arrive at this conclusion.
“The autonomous software engineering agent vision is more or less coming to fruition.”
Oddly enough no evidence of this is provided. I’ll admit it’s possible to have Claude write 100% of your code - however this is very distinct from autonomy. If I iterate on the code with Claude - telling it what to change, checking for errors and reporting them back to Claude, etc., I can build an app without “writing code” - but that is totally different from autonomy. And it’s not even clear how much of a time saver this really is if I know the language myself.
Once you are popular, whatever said is automatically true
Incredibly cringe. Literally every single medium mode problem for a team is framed as a groundbreaking achievement. These people snort their own farts.
Please be respectful. I find it incredibly dismissive that you would use such language to describe such amazing individuals. They revolutionized the industry. They built a game engine to render text for cripes sake! So please get it right.
Huff. They huff their own farts.
You’re totally right! You’re not building a CLI tool, you’re building a game engine!
Guess I have to throw it in here… “/s” on my original comment
Ah your mind hasn’t been fine tuned to the ChatGPT response, I applaud you.
I bet most readers will get angry and reply before reading the full comment :-)
> They built a game engine to render text for cripes sake!
Nope they didn't. I recommend taking a look at a real game engine (Unreal or Unity) and compare.
I personally wrote two game engines used for commercial games. Which included rendering text. It is not hard.
I support this comment, at the expense of wanton karma forfeiture.
I think the fart snorting assertion can be proven. There is evidence.
new product : claude cringe
Yeah for sure lmao.
> @dawn-drain — Research Engineer at Anthropic My main project for my first three years at Anthropic, beginning in 2021, was to try and make a model that was as good at coding as possible— at least as good as me.
> I had written a whole pull request by hand, and then Adam rejected it. But he said, “Actually, you should use clide for this.”
Interesting
Sonnet 3.5 and Cursor proving PMF for agentic coding should've at least got a mention...
Why do I have to select if i want to read it in terminal or in the browser? I clicked the link, show me the text.
You are not being clever, you are annoying.
You are absolutely right and I apologize. Would you like me to summarize the article in five short bullet points? Just give me the word!
Nice; Pretty interesting historical narration of Claude Code.
Perhaps the most telling part of this transcript is Tristan Hume, a prolific and talented programmer, and presumably an expert performance engineer, is repeatedly saying "this thing just doesn't work that well yet" or "it's just not there yet" and everyone else is kind of just fawning over it.
"It actually made something that worked. But when I tried using it, I found I didn’t like it. I need to wait for a Claude that has the taste ..."
Yes, I am familiar with that experience.