2,055 karma · joined October 8, 2013
But also we've just witnessed LLMs go from being a glorified line auto-complete tool to it writing a C compiler in ~3 years. And I think that's something. And noting how we keep moving the goal post.
- Ohh look it can [write small function / do a small rocket hop] but it can't [ write a compiler / get to orbit]!
- Ohh look it can [write a toy compiler / get to orbit] but it can't [compile linux / be reusable]
- Ohh look it can [compile linux / get reusable orbital rocket] but it can't [build a compiler that rivals GCC / turn the rockets around fast enough]
- <Denial despite the insane rate of progress>
There's no reason to keep building this compiler just to prove this point. But I bet it would catch up real fast to GCC with a fraction of the resources if it was guided by a few compiler engineers in the loop.
We're going to see a lot of disruption come from AI assisted development.
I think these days the $200 Max subscription wouldn't be needed. I bet with these latest models you can make due with mixing two $20/mo subscriptions.
Real time was 2 weeks of watching the agents while watching TV and playing games, waiting for limit resets, etc... Very little decided focused time.
This was built using old versions of Codex, Gemini and Claude. I'll probably work on it more soon to try the latest models.
I wish they would share the full conversation, token counts and more. I'd like to have a better sense of how they normalize these comparisons across version. Is this a 3-prompt 10m token game? a 30-prompt 100m token game? Are both models using similar prompts/token counts?
I vibe coded a small factorio web clone [1] that got pretty far using the models from last summer. I'd love to compare against this.
It doesn't have to be malicious. If my workflow is to send a prompt once and hopefully accept the result, then degradation matters a lot. If degradation is causing me to silently get worse code output on some of my commits it matters to me.
I care about -expected- performance when picking which model to use, not optimal benchmark performance.
When you're in an abusive relationship they say intentions don't matter, only impact does. Because victims often focus on the intentions of their abuser and stay in the cycle of abuse.
Let me repeat it, intentions don't matter, only impact does.
I predict in 2026 we're going to see agents get better at running their own QA, and also get better at not just disabling failing tests. We'll continue to see advancements that will improve quality.
I've had a 15 year+ successful career as a SWE so far. I don't think I've had a single idea so novel that today's LLM could not have come up with it.
I've been able to help setup cross app automation for my partner's business, remodel my house, plan a trip of Japan and assist with the cultural barrier, vibe code apps, technical support and so much more.
The threads says this was 4 hours of work and they billed for things that weren't even used.