Plebian

Life finds a way.

@jaredef @htmx_org Me too, it begins with `git clone`...

@htmx_org This is probably a good time to tell you that I created a single prompt that deterministically builds the entire htmx library htxlang.org/seed/htmx

it is truly idiotic to compare a explicitly deterministic process like compilation to an explicitly stochastic process like LLM inference

@omninomsky @javarevisited I have found that for hard problems, an LLM agent is a great rubber duck! Because they excel at fuzzy pattern matching, they do well in non-deterministic systems. However, they won't always solve it. They'll find the issues and possibly even implement a solution, if you guide it.

@javarevisited Lets assume for the sake of argument that this is true. (I doubt it) How does that solve the hard problems regularly being faced in production grade software systems? You know, the things senior engineers get paid to deal with regularly.

Serious question. If ChatGPT can explain system design better than most seniors… What is the real value of experience now?

@javarevisited Whatever textbooks are going for these days. That's how I learned. But putting out fires in production, that's how I cut my teeth! Running GPT willy nilly is like driving everywhere in second gear. You'll make it, eventually. Maybe. You have AI, you have no excuse not to learn.

@PeterBell @plainionist If it's hard for a human, it's hard for an LLM. They aren't any smarter than us, but they are good haulers. Sometimes.

Why would human comprehension be the bar? Current SoTA models are writing mathematical proofs that make no sense to me but I am told are elegant and correct. No reason to believe that in the near future they'd write code that was beyond my comprehension as well. Just need to make sure they can fix it as well!

Can it even be "perfect code" when humans don't understand it? 🤔

@vikrambuilds It runs on philosophy, and nobody knows how that works!

Worst dev nightmare in 2026: AI gives perfect code… but you have no idea how it works 😭

@FormerlyNoAcc @bobby_mcgi54028 @IceSolst @bitsadventures If assembly is the level you're on, you'd better review it. But even if you are operating on higher abstractions, it's still useful to know the underlying details. Knowing how Java bytecode works helps me tune JVMs to C-level performance in the cloud but I'd rather use Clojure!

@bobby_mcgi54028 @IceSolst @bitsadventures Usual deflection is having sufficient red-green, but that's just writing it yourself with extra steps. Truth is it won't produce great code like 99% of the time and if you're not reading it, well... Not being able to read the assembly is a way way way lesser sin.

Interesting article on treating agent output like compiler output (and why) skiplabs.io/blog/codegen_a…

@IceSolst It was reviewed. Rigorously. By the compiler contributors. That's why I don't need to review it. However I still learn about how the compiler works because all abstractions are leaky. It's not even about "determinism", it's about knowing the underlying implementation details.

@S_N_SH_E_ Suck less

In the last 12 months, AI reduced build time by ~70% but didn’t increase success rate the same way. So here’s the real question: If everyone can build faster, what actually creates an edge now?

@oskarth Pales in comparison to dev / develop.

so many years wasted because some SJWs insisted we change default git branch from master to main

@adrianmcli @oskarth snowflake.sh

@oskarth You can write a bash script for this bro. Completely abstract it away.

@itsolelehmann Tell it hack your site like Mythos. Chaos engineering.

POV: claude traveled 6 months into the future and told you exactly how your next move failed. it's called a premortem. daniel kahneman (nobel prize-winning psychologist behind "thinking fast and slow") called it his single most valuable decision-making technique. google, goldman sachs, and procter & gamble all use it before major launches. here's the problem it solves. when you ask claude "is this a good plan?" it finds all the reasons to say yes. that's what it was trained to do. so you walk away feeling confident. you execute, and spend weeks / months building on top of that plan. then it blows up. and you realize the problem was obvious in hindsight, you just never stress-tested it because claude told you it was solid. a premortem fixes this by flipping the frame. instead of asking "what could go wrong?" you tell claude "it's 6 months from now and this is already dead. tell me how it died." that shift turns off claude's optimism because there's nothing to be optimistic about. the premise already says it failed. so claude stops looking for reasons your plan will work and starts explaining how it fell apart. claude comes back with every way your plan could die, each one with a full failure story and the early warning signs to watch for. then a synthesis pulls it all together: > which failure is most likely > which failure is most dangerous > the single biggest hidden assumption you're making (often the most valuable part) > a revised version of your plan with the gaps closed you say "premortem this" and give it your plan. the skill handles the rest.

@ibuildthecloud I can code anything manually but I'm having great success with small composable primitives. Think computer science and philosophy, all the way down. Define the system as nodes and edges in a graph. Set it all up in topological order, and watch any model knock it down.

We're definitely going to get to the point where handwriting code doesn't make sense. Similar to handwriting assembly doesn't make sense. But the analogy breaks down. Developers don't really know assembly and they can't really review assembly. On top of it, the assembly that's produced by compilers is so optimized and confusing that you can't even understand it. For AI the code that's being produced, the developer needs to understand it could have written it themselves but it's just not as efficient or safe to do so. AI is not a new abstraction layer. It is a tool to create something you understand. We still need people to understand the mechanics of coding. We still need people to learn these skills and these are important skills. And even with the power of AI, you can't get away with not knowing these skills if you want to be effective. As always, there'll be people who try to take shortcuts and they'll get short-term gains and then it will fall apart.

@ThielenKalani @zuess05 @mboya_angina But what I don't get is why don't you have Claude decide the halting problem while you're at it? Or a host of other provably impossible problems? You're just lazy.

@Plebian_2 @zuess05 @mboya_angina “Claude don’t reinvent the wheel.”

We spent the last decade treating software engineering like some elite, highly complex art form. Now random teenagers are copy-pasting entire SaaS platforms out of Claude on a Tuesday morning. What happens to our entire identity when our hardest skill becomes a cheap commodity?

@GarysXad @rushicrypto No, I'm not calling you a liar. We're talking about two different things. I hope your friend is successful.

@Plebian_2 @rushicrypto My Friend is in the AI business AI can be used to conquer small jobs where it don’t need the “learning” that the internet provides You are a stupid piece of shit and I do not lie

Here's a thought: Why don't the AI companies put the AI to work on how to reduce the energy demand of AI by 90%?

@jordan_patten @OmarShahine It's a signal that you may need to take a different tack. Often I find there is a fundamental misassumption undergirding the inquiry. Sometimes it's a one-line change. You have many tools now. Don't forget the battle-tested rubber duck or hammock-oriented engineering.

@OmarShahine Yeah I thought that was bizarre too, or like “you’ve tried fixing this a few times now and it’s still not working how about we take a break.” Ummm no we need to fix it

What the heck is going on with Claude Code. It spends a lot of time telling me things are going to take hours days or weeks to do and wants to stop working a lot. "Or if you'd rather I sleep on it and pick up tomorrow with fresh eyes, that's fine too." Like why would I want you to sleep on it? You don't sleep.

@OmarShahine How would you think under such constraints? Lucid or anxious? Give it time and space. It puts it in a calmer more rational state. Impatience is a virtue of a great programmer, but it's also a leaky abstraction in a harsh topological space. Gotta find the sweet spot.

@zuess05 @wenchodev It's always been possible to buy slop for cheap. You get what you pay for. Honestly though, the cost of software should be next to nothing. The value is in orchestrating systems and knowing how to patch leaky abstractions (the LLM is hit or miss).

@wenchodev Yeah I guess that’s worth paying for haha

I am actually curious. If junior devs are using Claude to write code at a senior level... And senior devs are using Claude to write code at a senior level... What exactly is the difference between the two roles right now, other than a $100k gap in salary?

@devXritesh I program manually too, I taught myself in the 90s before Internet. But I am embracing the LLM as a philosophical tool and a chaos engineer. If you can write code and computer science theory you can set up an agent to run with your ideas. It is messy but fast.

Will society accept me as a real developer if I don’t use AI? Everyone acts like if you’re not using Claude/Cursor/Copilot 24/7, you’re outdated. I still code manually. No AI autocomplete. Just me, the editor, and deep thinking. Am I falling behind… or actually building real skill? 2026 is splitting developers into two types: • Ones who prompt • Ones who understand Where do you stand? Be honest 👇

@GarysXad @rushicrypto On the contrary. Nobody is running state of art models at home, and if they are using them, it's only through an API call to data centers that are running the inference on GPU clusters, not home computers. Exposing to the net? True, but depends on what they are doing.

@Plebian_2 @rushicrypto Not exactly…. AI don’t really use a lot of power .. I have a friend who runs AI models on his computer at h on me What take the power is exposing it to the internet