[nubg]

rules of thumb for when to take blog posts about AI coding seriously:

- must be using the latest state of the art model from the big US labs

- must be on a three digit USD per month plan

- must be using the latest version of a full major harness like codex, opencode, pi

- agent must have access to linting, compilation tools and IDE feedback

- user must instruct agent to use test driven development and write tests for everything and only consider something done if tests pass

- user must give agent access to relevant documentation, ie by cloning relevant repositories etc

- user must use plan mode and iterate until happy before handing off to agent

- (list is growing every month)

---

if the author of a blog post about AI coding doesnt respect all of these, reading his blog posts is a waste of time because he doesn't follow best practices

[greggyb]

As stated in the article, I have unlimited access to multiple frontier models and I use Claude Code, among other harnesses. The rest of your list is not directly addressed in the post, because it is irrelevant to the point being made, but I do all of those things and more. You will note that in the appendix on LLM usage, some of the things I constantly have to correct in LLM-generated code are testing mistakes. And if you care to ask, yes I have context files to address these mistakes, and I iterate them to try to improve the experience.

I would honestly appreciate constructive feedback on LLM usage, because, as I stated, I am constantly having to rework code that LLMs generate for me. The value I get from LLMs is not in code generation.

[demorro]

You're missing the point, and also demonstrating it. This blog isn't about personal experience, and it makes no claims about LLM capability at all. It is simply about whether code, in either volume or quality, should be used as a proof claim.

> LLMs entice us with code too quickly. We are easily led.

Arguably _is_ your argument. That people aren't doing the above and it's causing problems. You probably agree that just spinning up Claude code on the regular plan without doing the above can still generate a fuck-ton of code but that shouldn't be used as evidence either for or against AI effectiveness.

[sarchertech]

> All my comments are written by AI. Quite meta, isn't it, knowing you came here after I triggered you with my "guys, this is AI generated slop" comment?

Maybe knock it off since the rules changed to not allow AI comments.

[nubg]

It's irony ofc

[sarchertech]

That’s not what irony means.