It's my experience that opus 4, and then, particularly, 4.5, in Claude code, are head and shoulders above the competition.
I wrote an agentic coder years ago and it yielded trash. (Tried to make it do then what kiro does today).
The models are better. Now, caveat - I don't use anything but opus for coding - Sonnet doesn't do the trick. My experience with Codex and Gemini is that their top models are as good as Sonnet for coding...
I was trying to do something yestesrday and Claude was keep messing it up, after like an hour i realized the model somehow switched to sonet, opus 4.6 is crazy good. It’s very obvious in practice.
Although I feel like for chasing bugs and big systems codex is even better
I am experimenting at a very early stage with using Verus in Rust to generate proveably correct Rust. I let the AI bang on the proof and trust the proof assistant to confirm it.
There is another route with Lean where Rust generates the Lean and there is proof done there but I haven't chased that down fully.
I think formal verification is a big win in the LLM era.
> As presented, Gollum is badly off, I reckon - missing the books textual description. The flowers are out of line.
This is addressed in the article. "Paul Gravett writes in his new book about Tove Jansson: ‘Her Gollum towered monstrously large, to the surprise of Tolkien himself, who realized that he had never clarified Gollum’s size and so amended the second edition to describe him as ‘a small, slimy creature’."
We have Jansson to thank for the clarification, it seems!
Tolkien made significant changes to the Gollum chapter. In the first edition Gollum gives up the ring willingly. The ring was not yet the Ring, and Gollum was not yet a Hobbit.
The man took retcons as an intellectual challenge. Sometimes the retcon itself spun off a whole new story. But it makes The Hobbit really incompatible with its own sequel, even after his changes. (You have to read it as having a very unreliable narrator.)
> Now it is a curious fact that this is not the story as Bilbo first told it to his companions. To them his account was that Gollum had promised to give him a present, if he won the game; but when Gollum went to fetch it from his island he found the treasure was gone: a magic ring, which had been given to him long ago on his birthday. Bilbo guessed that this was the very ring that he had found, and as he had won the game, it was already his by right. But being in a tight place, he said nothing about it, and made Gollum show him the way out, as a reward instead of a present. This account Bilbo set down in his memoirs, and he seems never to have altered it himself, not even after the Council of Elrond. Evidently it still appeared in the original Red Book, as it did in several of the copies and abstracts. But many copies contain the true account (as an alternative), derived no doubt from notes by Frodo or Samwise, both of whom learned the truth, though they seem to have been unwilling to delete anything actually written by the old hobbit himself.
Indeed: the intro to The Lord of the Rings explains that previous editions of The Hobbit, where the ring was a gift rather, were Bilbo's original lie to cover up the theft. Perhaps that was all the influence of the Ring itself.
That's how I've always imagined it - The Hobbit as published is the story told as if intended for children (hobbit or otherwise), but the 'actual' in-universe events were just as dark and realistic as the tone of The Lord of the Rings.
Clarifying question -- what do you mean Gollum was not yet a Hobbit? I don't think he ever was - but a river folk before the ring deprived him wasn't he? I never read first edition so I suspect there are some differences as you allude. (ring not being the ring).
Actually - in the creative process did he kick off the Hobbit then expand into the world building as an after thought and turn the one ring into this wild expansive creative endeavor? I always assumed it had been pre-built in his mind then spilled out in ink (As a sequence of events).
In the original version, there is minimal physical description of Gollum (it was dark after all) and the ring was simply a magic ring that granted invisibility. Gollum lost it and IIRC he just let Bilbo go. They whole idea of him being some hobbit-like creature corrupted by the One Ring was not present at all. It was one of a series of fairy-tale adventures no more important than the trolls turning to stone. Bilbo needed a way to sneak around Smaug, so he found a magic ring.
It's doubtless still possible to find that version, I read it in an old country library that had it on the shelf since the 1950s.
> There are of course other very minor changes. For instance, Gandalf tells Bilbo to bring out the chicken and tomatoes in the unrevised edition vs. the chicken and pickles in the revised edition. But I'll skip over these inconsequential changes.
Bit of medievalism there (tomatoes being a Colombian Exchange thing).
He knew about the umbrella and kept it in. It’s more likely Tolkien was concerned with the date and whether fresh tomatoes would be available than medievalism.
There's something so fundamentally European about potatoes, despite being (comparatively) new there. And British in particular. They're stodgy, bland, filling, comforting, and really tasty in a dull way. They displaced things like turnips, which had too much flavor and a dispiriting watery texture.
He could so easily have written that passage for turnips, but it would have been less comforting. (It is often read as if the word "potato" was used to translate some extinct native-European root vegetable, but I find that doubtful.)
Thanks - sounds like it was some reworking after building out the original story line to make the rest of it work. I enjoy the storyline that doesn't have to be tied into the main arc (ie trolls or fairy-tale adventure component).
I'm looking forward to when I can run a tolerably useful model locally. Next time I buy a desktop one of its core purposes will be to run models for 24/7 work.
Define useful I guess. I think the agentic coding loop we can achieve with hosted frontier models today is a really long way away from consumer desktops for now.
a lot of AI assisted development goes into project management and system design.
I have been tolerably successful. However, I have almost 30 years of coding experience, and have the judgement on how big a component should be - when I push that myself _or_ with AI, things go hairy.
> 30 years ago, it was safe to go to vendor XY page and download his latest version and it was more or less waterproof.
You _are_ joking, right? I distinctly remember all sorts of dubious freewarez sites with slightly modified installers. 1997-2000 era. And anti-virus was a thing in MS-DOS even.
I am working on a little project in my offhours, and asked a non-hacker (but competent programmer) friend to take a run at exploiting it. Great success: my project was successfully exploited.
The industrialization of exploit generation is here IMO.
I am _much_ more interested in i. building cool software for other things and ii. understanding underlying underlying models than building "better claude code".
It's my experience that opus 4, and then, particularly, 4.5, in Claude code, are head and shoulders above the competition.
I wrote an agentic coder years ago and it yielded trash. (Tried to make it do then what kiro does today).
The models are better. Now, caveat - I don't use anything but opus for coding - Sonnet doesn't do the trick. My experience with Codex and Gemini is that their top models are as good as Sonnet for coding...
reply