> 3. Regarding me specifically, I work on the LessWrong codebase which is technically open-source. I feel like calling myself an "open-source developer" has the wrong connotations, and makes it more sound like I contribute to a highly-used Python library or something as an upper-tier developer which I'm not
That’s very interesting! This kinda matches what I see at work:
- low performers love it. it really does make them output more (which includes bugs, etc. it’s causing some contention that’s yet to be resolved)
- some high performers love it. these were guys who are more into greenfield stuff and ok with 90% good. very smart, but just not interested in anything outside of going fast
- everyone else seems to be finding use out of it, but reviews are painful
> Anthropic claims they solely use agents to code and don't modify any code manually.
Have you used CC? It shows. They did not make their fortune off this, and it’s at least lost me a customer because of how sloppy it is. The model is good, and it’s why they have to gate access to it. I’d much rather use a different harness.
I do think you’re on to something though. As societal wealth further concentrates among the few, we’re going to get more and more slop for the rest of us because we have no money (relatively speaking). Agentic coding is here to stay because we as a society are forced more and more slop. It’s already rampant, this is just automating it.
...uh, I think Claude Code is great, actually. A lot of that is indeed just the strength of the underlying model, but the local client is great too. Plan mode, checkpoints, subagents... I've been using Claude Code for a year now, and I feel like Anthropic has steadily been eliminating pain points.
I love Claude Code and use it all day, every day for work. I would self identify as an unofficial Claude Code evangelist amongst my coworkers and friends.
But Claude Code is buggy as hell. Flicker is still present. Plugin/skill configuration is an absolute shitshow. The docs are (very) outdated/incomplete. The docs are also poorly organized, embarrassingly so. I know Claude Code's feature set quite well, and I still have a hard time navigating their docs to find a particular thing sometimes. Did you know Claude Code supports "rules" (similar to the original Cursor Rules)? Find where they are documented, and tell me that's intuitive and discoverable. I'm sorry, but with an unlimited token (and I assume, by now, personnel) budget, there is no excuse for the literal inventors of Claude Code to have documentation this bad.
I seriously wish they would spend some more cycles on quality rather than continuing to push so many new features. I love new features, but when I can't even install a plugin properly (without manual file system manipulation) because the configuration system is so bugged, inscrutable, and incompletely documented, I think it's obvious that a rebalancing is needed. But then again, why bother if you're winning anyway?
Side note: comparing it to Gemini CLI is simply cruel. No one should ever have to use or think about Gemini CLI.
I've been using deletated Claude agents in vscode and it crashes so much it's insane... I switched to copilot Claude local agents and it works much better.
Idk about this whole vibe coding thing though... Well see what happens
I don't think they will lose on inference because that assumes that compute becomes cheap for all evenly.
Their spending today has secured their compute for the near future.
If every GPU, stick for RAM and SSD is already paid for. Who can afford to sell cheap inference?
Z.ai is trying to deal with this by using domestic (basically Huwawei silicon not Nvidia). And with their state subsidy they will do well.
Anthropic has a 50bn USD plan to build data centres for 2026.
OpenAI similarly has secured extraordinary amounts of other people's money for data centres.
All these will be sunk costs and "other people's money" while money is easy to get hold off. But will be a moat when R&D ends.
Once all the models become basically the same who you go with will be who you're already with (mostly OpenAI), and who you end up with (say people who use Gemini because they have a Google 2TB account).
Some upstart can put themselves into the ground borrowing compute and selling at a loss but the moment they catch up and need to raise prices everyone will simply leave.
ChatGPT is what is most likely to remain a sustained frontier model. Maybe Claude jumps ahead further a few times, Gemini will have its moment. But it'll all be a wash with ChatGPT tittering along as rarely the best. But never the worst.
> Once all the models become basically the same who you go with will be who you're already with (mostly OpenAI)
Imho, people are undervaluing the last mile connection to the customer.
The last Western megacorp to bootstrap its way there was Facebook, and control over cloud identity and data was much less centralized circa-late-00s.
The real clock OpenAI is running against is creating a durable consumer last-mile connection (killer app, device, etc).
"Easy to use chat app / coding tool" doesn't even begin to approach the durability of Microsoft, Apple, Google, or Meta. And without it, OpenAI risks any one of them pulling an Apple Maps at any time.
Unless it continually plows money into R&D to maintain the lead and doesn't pull an Intel and miss a beat.
Maybe they do, but that's a lot of coin flips that need to continually come up heads, in perpetuity.
> As people say - something changed around Dec/Jan
Yes, Anthropic decided they wanted to IPO and got the hype machine in full swing.
Don’t get me wrong LLMs are here to stay but how we’re currently using them is likely going to change a lot. Stuff like this:
> in general everybody has to do bottom up cleanup and documentation of all their projects, setup skills and whatnot and that's assuming their corp is ok with it, not blocking it
Is not needed to get a lot out of AI, and is mostly snake oil. Integrating them with actionable feedback is, but that takes a lot of time and rethinking of some existing systems.
I don’t like the Internet analogy cause that’s like producing a new raw material, but AI is gonna be like Excel eventually (one of the most important pieces of software in the world).
It's not Dec/Jan is not just anthropic, tooling clicked (ie. vscode copilot became great, no need for CC really but CC for VSCode also stopped sucking), great models were released with really good tool use, better coding and sucking less and people had holiday break at scale to actually have time to play with it – it all clicked for a lot of people around that period.
Only a personal anecdote, but the humans I know that have used it are all aware of how buggy it is. It feels like it was made in 2 weeks.
Which gets back to the outsourcing argument: it’s always been cheap to make buggy code. If we were able to solve this, outsourcing would have been ubiquitous. Maybe LLMs change the calculus here too?
Yeah, it’s odd watching the outsourcing debate play out again. The results are gonna be the same.
Which is a shame, cause I think LLMs have a lot more use for software dev than writing code. And that’s really what’s going to shift the industry - not just the part willing to cut on quality.
AI has been a lifesaver for my low performing coworkers. They’re still heavily reliant on reviews, but their output is up. One of the lowest output guys I ever worked with is a massive LinkedIn LLM promoter.
Not sure how long it’ll last though. With the time I spend on reviews I could have done it myself, so if they don’t start learning…
That’s very interesting! This kinda matches what I see at work:
- low performers love it. it really does make them output more (which includes bugs, etc. it’s causing some contention that’s yet to be resolved)
- some high performers love it. these were guys who are more into greenfield stuff and ok with 90% good. very smart, but just not interested in anything outside of going fast
- everyone else seems to be finding use out of it, but reviews are painful