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
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.
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.
> 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.
> I think it would typically have taken you longer.
That's actually highly doubtful to me.
Tons of studies and writing about how reading and debugging code is wildly more time consuming than writing it. That time goes up even more when you're not the one that wrote the code in the first place. It's why we've spent decades on how to write readable/maintainable code.
So either all this shit about reading/maintaining code being difficult was lies and we've spent decades wasting our time or AIs can only improve productivity if you stop verifying/debugging code.
So I find it very unlikely that it would have taken more than a couple hours to just write it the first time.
2. If God were to descend and give you a very good, reality-tested spec:
3. Would you be done faster? Of course, because as every AI doomer says, writing code was never the bottleneck!!1!
4. So the only bottleneck is getting to the spec.
5. Guess what AI can help you with as well, because you can iterate out multiple versions with little mental effort and no emotional sunk cost investment?
Man, if this were true we’d see a crazy, massive explosion of quality products being written, and launched. While we see some use, i don’t perceive an acceleration. In fact, i see a lot of trivial bugs being deployed to prod.
And then it turns out God wrote the spec in code because that’s what any spec sufficient to produce the same program from 2 different teams/LLMs would be.
> Not because it’s trendy. Because ACID compliance
> The CAP theorem is not a suggestion. It’s a mathematical proof.
> this is not a philosophical inconvenience. It’s a compliance violation.
@dang, i thought we just banned AI written blog posts?
I know I will get downvoted for this everytime I post this comment, but I'm sick and tired of 80% of everything being written/heavily edited by AI.
Can the author not just use his own words? Is he really that bad at writing? But somehow the content is supposed to ne worth my time?
I'd rather read the prompt instead, so I know which parts the author actually meant to say, rather than wondering which parts the AI invented to make it sound flowery and punchy, the Tiktok style of prose.
In this case, Trump is the "worst person you know," but he's doing something that a lot of people across the left [¹] and right [²] political spectrums have wanted to do for years by repealing (edit: suspending) the Jones Act.
[¹] Free trade dems, Clintonite think tankers, third-way corporatists, the self-styled "neoliberal" dems, abundance dems, etc. Not all dems support doing away with it because in theory that hurts the maritime unions.
[²] Much broader support on the right because it maps onto the free market and anti-regulation ideologies.
i'm confused by your footnotes, because in the first one you clarify you mean right-wingers and "free market" fanatics and then seemingly restating it in your second footnote while presenting it as a separate group haha
> Free trade dems, Clintonite think tankers, third-way corporatists, the self-styled "neoliberal" dems, abundance dems, etc.
These are all very much part of the democrat coalition, i.e. the "left" on the US political spectrum. It's not the entire democrat coalition, but they are a large part of it. I'd call them centrist for sure, but certainly not right wing.
Again we're talking about US politics here, i.e. left is the democrat coalition and right is the republican coalition. You'd have to be a very online blue sky user to call e.g. abundance dems anywhere near the right in the US.
> IRAN has claimed responsibility for an attack on two oil tankers anchored in Iraqi territorial waters, as conflicts in the region continue to escalate and strikes on commercial shipping spread beyond the Strait of Hormuz.
It's not that it's impossible to go through it, but you have to do something specific in order to do so beyond just trying to go through, or you're going to walk straight into getting a bloodey nose.
But yeah, these ships weren't anywhere near the strait.
I gave a similar presentation in January which covers the AI features that emerged in 2025 that culminated in the step-function in capability in Nov'25 and where I went from there.... (certainly my GitHub activity is bright green since)
The presentation was created with Claude Code to prove itself; never going back to Keynote/PowerPoint. Press 'X' key to disable "safe mode". Prompts are in the repo.
- 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
reply