Doing "almost exactly the same thing" is fubdamental to competition and capitalism. The ones doing it better will survive, that's how we improve.

About climate, I think you overplay it. China is already investing heavily in nuclear, and we should be doing the same.

Alibaba (who makes Kimi) and Moonshot AI (who makes Qwen) collaborate closely. Qwen relies on Alibaba's infrastructure for training.

Look at your iPhone and you'll see all the greatest inventions have been born out of collaboration not competition:

* GPS was created by the Department of Defense

* the internet was created through the collaboration of many international research institutions

* speech recognition came from MIT and DARPA

* AI voice assistants were created by DARPA. Apple immediately hired the head of the program after it was finished to create Siri

* accelerometers (MEMS) is another DARPA innovation from the 90s

* touchscreens were invented by CERN

* digital cameras came out of Bell Labs which had a government-mandated monopoly that required them to fund research like this

* lithium-ion batteries were created through a collaboration between British, Japanese, and American organizations

All of these have only become transformative technologies because they were created with public funding and released to the public. It's collaboration and public funding that drives innovation

I'm using firejail to sandbox Opencode, for security and to keep the agents from personal data. I documented it in my blog [1].

[1] https://blog.fidelramos.net/software/how-i-sandbox-ai-agents...

Thank you - great blog post.

https://github.com/haplo/pelican-theme-reflex — My Pelican [1] theme. While I started it before LLMs were a thing, I was recently able to tackle all my open TODO items with AI.

https://github.com/haplo/pelican-copy-code — A Pelican [1] plugin to add a Copy button to all source code blocks. It happens at build time, so no DOM manipulation at page load time.

https://github.com/haplo/venice-kde-widget — A KDE Plasma widget to display Venice.ai [2] remaining

I'm currently writing a series of blog posts about these that I call Vibe-coding Escapades [3]. I will also write about bugs and technical problems that AI helps me debug and solve, some are non-trivial and have been pestering me for years.

[1] https://getpelican.com/

[2] https://venice.ai/

[3] https://blog.fidelramos.net/software/vibe-coding-escapades

I've been eyeing the MNT Pocket Reform for a while:

https://shop.mntre.com/products/mnt-pocket-reform

I posted the link to the Spanish article by mistake, please notice that there is an English version available:

https://blog.fidelramos.net/software/tailscale-3-syncthing

You are correct, this article is part of my self-hosted Tailscale series using Headscale, which is why I'm receiving the full benefits of moving Syncthing to it. Otherwise Tailscale would keep the same metadata that Syncthing discovery and relay servers were. I talk about that in Part 1 of the series, it's a main motivator for self-hosting Headscale.

This is so cool that we can port useful stuff like this in minutes, thanks for sharing!

I packaged it for Arch Linux AUR: https://aur.archlinux.org/packages/whatcable-linux

Is that accounting for SpaceX stages being reusable? Honest question.

Make it Monero mining, it's CPU-efficient and private.