Hey, Ghostr is pretty cool! Have you seen it yet? ⚡️👻👀👻👀⚡️
You can configure what tools it has access to, and only give it read access to your drive. My concern would be privacy. I’d be curious to try that using LM Studio with a local model.
I’m bullish on eventual Bolt12 receive support on Fedimint, but that requires PTLCs on lightning 😞
But whenever that comes around, it will strike a very nice balance of trust and convenience
Have you used AWS? It ain’t exactly miles ahead. I can’t wrap my head around just why cloud computing DX is so unbelievably bad.
Partly true? I’m curious what you mean. In what way do you think it could be good vs bad?
I disagree that prior tech deflation was linear rather than exponential. Regardless, I appreciate your respectful disagreement with me. I’ve been generally disappointed with the rudeness I often get on Nostr for voicing my honest thoughts. Your response is refreshing, thank you!
If tech deflation reduces the debt ratio, why has the debt ratio gone up in the past half century even though we’ve had staggering tech deflation? IMO if the rate of tech deflation increases, the rate of money creation will simply be increased to absorb it. The only thing new this time around is bitcoin. The tech deflation of AI is a difference of degree, not of kind.
I’m with you. I quit a past job over having to try wrangle a ridiculously convoluted Python codebase and running into exactly the issues you’re describing. I don’t know how or why anyone ships any production software with Python. Sure, it works for quick and dirty one-off scripts, but anything more than that and you run into crazy amounts of untestability and non-determinism.
Have you looked into METR at all? They’ve done some interesting work on measuring AI model capability and autonomy against a benchmark of software engineering tasks which have baseline measurements of how long they take a human to do.
TL;DR, frontier models currently have a ~50% success rate when performing tasks that would take humans a little over two hours, and that time has been doubling every 7 months. Assuming that trajectory, they’ll have the autonomy to complete a human 40 hour work week worth of work in less than three years.
https://metr.org/blog/2025-03-19-measuring-ai-ability-to-complete-long-tasks/
CrowdHealth has done right by me. Been a member for two years, and they paid for my daughter’s birth. Super painless to get the funds sent to my bank account. Sadly now that I have insurance through work, I’m probably going to leave. Would love to see a way for employers to offer CrowdHealth to employees as a substitute for health insurance.
You don’t get insurance cards, because CrowdHealth is not insurance (it’s better IMO).
As far as coverage, check out this page on their website: https://www.joincrowdhealth.com/how-it-works-v2
TL;DR, you pay the first $500 for each health event. I believe a health event can range from a broken arm to brain cancer.
Real bitcoiners celebrate whitepaper day instead
I use both Damus and Primal, and can’t get zapping to work on either. Damus gives me a cryptic error, and Primal asks for a bunch of personal information that I’m not giving out. Idk exactly how zapping works under the hood, but shouldn’t Nostr clients be able to spit out a Bolt11 that you can pay out of band?
Oh snap, I got everything I need already! Will give it a shot tomorrow, thanks man
Still, it’s somehow less terrible than Python. I genuinely have no idea how anyone gets anything done in that language.
Next step is sat-denominated 😁
First-class nix support in OpenAI Codex would be a game changer
Anyone else using Claude Code? I started using it over the past few days on a React project and it’s kinda blowing my mind. Still need to see how well it works in a more pedantic language like Rust.




