Do you write perfect code? I was in denial for a long time too, but the reality is this is how software engineering is now. It can handle pretty much any codebase and vastly faster than you ever will be able to. With enough context, it writes good code and it's only getting better.

> Do you write perfect code?

For simple things, occasionally!

For more complex things, usually not, but I write working code. LLMs will sometimes write working code, sometimes not.

If you haven't tried any of the newer models from the last ~6 months, you're out of touch. A year ago, LLMs were essentially just a clever search engine for me, I found they fell apart pretty quickly on bigger codebases and more complex projects.

It doesn't always write perfect code on a first pass but nor do you, and it's amazing at understanding any issues it may encounter, even very low level ones. I don't think there's anything I can do that it can't now. There is a learning curve to working well with LLMs, but once you get there you'll be producing high quality code faster than you can imagine.

Part of me wishes they weren't so good. I enjoy(ed?) the process of working through problems and it does mean you might no longer have a perfect mental model of your code. But I do also enjoy being able to build things well and fast.

Opus 5.1 was just so darn verbose and overtuned. I look forward to trying out the new Luna/Sol and recent Claude fixes.

Yeah to be honest I still find Claude too verbose with Fable 5.1, I much prefer the output style of OpenAI's models. With strict guidelines it can be improved but it tends to drift in longer conversations.

This is easily proven incorrect unless you assume overly draconian definitions of "perfect".

```python

print("hello world!")

```

runs exceptionally well. Bullet-proof in all use cases I've needed it.

Nobody is earning a living writing hello world programs, and in real programs perfection doesn't exist. Perfection is at best a widely-held opinion.

Two thoughts:

1. Since when do people write code solely for the purpose of earning a living?

2. Yes. I agree that with complex programs, perfection doesn't exist -- mostly a composition paradox issue though, IMO!

This is also easily proven incorrect when you start to analyse how “print” works and wether it stands to be improved in any way.

No. No code is “perfect”.

Aye, a draconian view of the "perfect."

As opposed to… the lenient view of “perfect”? Perfection is a binary state, either something is or it isn’t. What is /defined/ as perfect in a given context is subject to change, but the meaning of the word is absolute.

I love that we are debating the platonic perfect form of perfect.

I may not write perfect code the first time, but I can, on my own, reliably figure out where my code is imperfect and fix it. And I get reliably better at that over time. Without requiring companies with massive power and concerning agendas to spend astronomical amounts of money and energy to power that improvement.

> Still waiting for my LLM to write perfect code.

I am not writing perfect code either.

How many companies depend on perfect code?

If you had enough common sense and ethical integrity not to participate in the blockchain bubble, and really believe what you claim, then why are you participating in the AI bubble?

...or did you?

> then why are you participating in the AI bubble?

I think you know that most software engineers don't have a choice if they value their employment. There was no industry-wide push to "participate in blockchain or be fired."

I disagree with parent but your comment just reeks of “ha! gotcha!” vibes, which… yeah… no.

I maintain the notion that LLMs are just what all NFT grifters moved to after the NFT fad died.

I really don't get this. Sure there is hype. But LLM's have already changed our industry, and it will never be the same.

I don't care whether the LLM can solve this or that mathematical previously thought unsolvable theorem. What I do care about is can it write good, maintainable code. And every new release of frontier models - they get better at it.

Good output depends on good input (prompt), and a good set of available tools the model can use to verify their work. If given this, nowadays really you need to try to get the model to output garbage.

Anthropic is now running at $100B in annual revenue. Based on a quick Google, that's about 3 OOMs greater than the biggest NFT company.

Remember debugging that last weird edge case? These guys are almost there on a whole N64 game. Incredible effort!

Pretty neat how these abstract origins still shape our world, especially in CS. Kinda makes you appreciate the shoulders we stand on.

Sonic Pi is fantastic for live coding sessions. Used it once for a small demo, really intuitive and fun to noodle with.

Saw him explain Fourier transforms once; it finally clicked. Probably another masterclass in clarity.

14x commit growth in one year is brutal for any infrastructure. Scaling isn't just adding hardware.

GitHub is unstable since around 2020