causal

Right, so they probably meant bullish

Wow great writeup and holy cow that's bad - I'm still trying to understand what OpenClaw/Moltbot can do that makes it worth this to so many people.

Yeah it's wild we blame GDPR instead of the companies that decide tracking us is worth our inconvenience

This is the 5th OpenAI product called Codex if I'm counting correctly

Yeah use Flash 3 for easy + fast stuff, but it can't hold the plot like Opus or Codex 5

I like this way of framing the problem, and it might even be a good way to self-evaluate your use of AI: Try vibe-writing a novel and see how coherent it is.

I suspect part of the reason we see such a wide range of testimonies about vibe-coding is some people are actually better at it, and it would be useful to have some way of measuring that effectiveness.

To be clear, since I'm past the edit window: I think Ozempic is an amazing drug, it's just unfortunate how popular it is to hate on people for using a drug they need.

Not really, probably a majority of Americans look down on people using Ozempic

I've had the same suspicion for various providers - if I had time and motivation I would put together a private benchmark that runs hourly and chart performance over time. If anyone wants to do that I'll upvote your Show HN :)

Yeah, lots of "change your habits" type responses that won't change the reason we're here.

Sigh. Lines of code is not execution man. Having functional apps is not execution. A full stack, AWS-deployed, multi-stage scaleable microservices wonder is not execution!

LLMs make it a lot easier to build MVPs, but the hard work of VALIDATING problems and their solutions, which IMO was always >80% of the work for a successful founder, is harder than ever. With AI we now get 100 almost-useful solutions for every real problem.

My friends love this game, but we've found a bug: The reported time in the win modal differs from the time copied to clipboard when you share.

Same. Abandoned fly for Digital Ocean when I found myself hitting my head against the wall trying to get their "just works" to work too often.

Huh. That doesn't sound like a browser at all. But okay, thanks for the summary!

What is Pinokio? The website just says "Your PC is the Cloud" - what?

Unsure why you were downvoted; I'm curious to understand this comment. Playing finance and accounting games I presume you mean.

Ah you're right, scrolled past that - the most salient contrast in the chart is still just GPT-5 vs GPT-4, and it feels easy to contrive such results by pinning one model's response as "ideal" and making that a benchmark for everything else.

And the example given was specific to OpenAI models, yet the title is a blanket statement.

I agree with the author that GPT-5 models are much more fixated on solving exactly the problem given and not as good at taking a step back and thinking about the big picture. The author also needs to take a step back and realize other providers still do this just fine.

I don't understand how a component library would be AI-proof in a way CSS templates are not.

I expect a lot of business disruption because of AI. Agree it's not the same as employee replacement, but it adds to the sort of fog of war around what effect AI is really having.

The Comrade Trump memes would go crazy

In my experience GPT-5 is also much more effective in the Cursor context than the Codex context. Cursor deserves props for doing something right under the hood.

Yeah sounds unhealthy, at least self-aware?

These numbers don't pass sanity check for me. With 4x300W cards you can get a 1K token DeepSeek R1 output in about 10 seconds. That's just 3.3Wh right? And that's before you even consider batching.

Hmm so I can't really tell if this falls under "Ahead of the game" or "AI psychosis". The latter is usually accompanied by impact to quality of life, with hints of that where he talks about coding on his phone around friends (which thankfully he recognized is unhealthy).

Working with startups, I meet a LOT of people who obsessively cannot stop using LLMs. People who jump on MAX plans to produce as much as possible- and in the startup scene it's often for the worst ideas.

LLMs are slot machines- it's fun to pull the lever and see what you get. But the hard problem of knowing what is actually needed gets harder as we sift through ten-thousand almost-useful outputs.

You get points for suggesting edits and badges for completing review activities (votes to close, triage, etc). I thought you got points for the latter as well but looks like that's not the case.

I'm kind of confused too. I spend way more time testing and reviewing code than I could possibly keep up with 4 agents

And half the garbage is from people "moderating"! You are literally rewarded points for doing moderating activities, so of course every post is flooded with BS edits, votes to close, etc.. Cobra effect and whatnot.

Oh absolutely - when it becomes clear you have high engagement somewhere, adapt that feature to facilitate the engagement! They could have made comments threaded or embedded ways to expand it into the right forum, but instead they literally shut down engagement. Bonkers.

Big fan of Linux. Use various distros daily.

That said, tech folk routinely underestimate how much they rely on their own technical skill. Try using Linux for a week without ever opening a terminal. Terminal is a "f this I'm going back to Windows" button for most people.