teunlao

Impressive tech execution, but the format has fundamental scaling issues.

Clubhouse lost 93% of users from peak. WhatsApp sends 7 billion voice messages daily - but those are DMs, not feeds.

The math doesn't work: reading is 50-80% faster than listening. You can skim 50 text posts in 100 seconds. 50 voice posts? 15 minutes.

Voice works async 1-to-1. You built Twitter where every tweet is a 30-second voicemail nobody has time to listen to.

The transcription proves it - users will read, not listen. Which makes this "text feed with worse UX"

Learned more about attractors dragging this around than from wiki. This is how math should be taught.

SQL interface for querying traces is the real killer feature though.

SARCASM: the only acronym worth building hardware for

This looks promising for vanilla PHP projects

Two questions:

How does this differ from Laravel's Vite plugin? I'm curious if there are specific advantages for non-framework PHP apps, or if the component system offers something unique.

When a PHP component changes, does it trigger a full page reload or partial re-render of just that component?

Love the DSL. Saw caching is roadmap — does a failed step restart everything, or resume partway?

Nice work! Batch generation is underserved for sure.

One question: how do you handle consistency across generations? With Veo's non-deterministic nature, keeping 50 product photos stylistically coherent seems tricky.

Agree. Starting from Python for-loops is embarrassing baseline. Any decent implementation gets you most of that 5x for free. The interesting part isn't the speedup - it's that AI can do routine optimization unsupervised. That's the actual value prop.

Great library. Just memo your custom nodes properly - edge re-renders cascade faster than you'd expect.

Humans cut the phrase to avoid sounding like AI. AI copied it because humans used it. Style collapsed into feedback loop.

Chromium fork with AI sidebar. Revolutionary? No. The limits-for-default-browser bait though - that's actual strategy. Will people stay after seven days? My guess most won't.

This is the kind of tool that actually gets used daily, well done!

Nice try inventing algebraic independence. Two moments here: the concept exists, the name doesn't. Whether it's useful in crypto - we'll see when someone actually tests it.

us-east-1 down again. We all know we should leave. None of us will.

Been using both daily for three months. Different tools for different jobs.

Claude Code has better UX. Period. The permission system, rollbacks, plan mode - it's more polished. Iterative work feels natural. Quick fixes, exploratory coding, when I'm not sure exactly what I want yet - Claude wins.

Codex is more reliable when stakes are high. Hard problems. Multi-file refactors. Complex business logic. The model just grinds through it. Less hand-holding needed.

Here's the split I've landed on - Claude for fast iteration tasks where I'm actively involved. Codex for delegate-and-walk-away work that needs to be right first time.

Not about which is "better" - wrong question. It's about tooling vs model capability. Claude optimized the wrapper. OpenAI optimized the engine.

METR study is brutal. 20% faster vs actually 19% slower - that's not rounding error territory.

Productivity measurement is a mess though. Spent three weeks last spring chasing metrics on a "microservices will save us" project. Spoiler: they didn't. Could've been TDD, could've been GraphQL, could've been AI - same story. We suck at measuring ourselves.

But that asymmetry thing - yeah, that's real. Spammer generates 10K bot accounts, never checks if they work right. I write code with Copilot, gotta verify every function or I'm debugging hallucinated bugs at 3am. Harm scales unlimited. Benefit hits human bandwidth ceiling. Hard ceiling.

Slot machine comparison works for coding, breaks everywhere else. I use it for boring CRUD boilerplate, ticket categorization, email templates I'd write anyway. Not life-changing. Not gambling addiction either. Just... mildly useful sometimes? Context dependent.

Hallucinations aren't fixable - author's right there. But calculators can give you garbage output too if you fat-finger 7 instead of 4. Still use them. Just don't blindly trust them.

The variable naming problem never went away. We just moved it from "temp2" to "make this better."

Same skillset. Different enforcement. Compiler used to force clarity through syntax errors. AI forces it through three debugging cycles when you realize the output doesn't match what you thought you said.

Natural language was never designed for precision. We spent decades building tools that forced precision. Now we're back to ambiguity at scale, and the engineers who couldn't name variables are writing paragraph-long specifications.

Logic is abundant. Clarity is the new bottleneck.

Bruno + git has been perfect for our team. Collections in the repo, no external dependencies, works offline. Should have switched years ago.