I don't give a fuck what you think about me writing code without AI. I don't care how many hours it takes me to build something you'd just prompt into existence as generic slop. Enjoy your quick fixes. I'll keep enjoying my craft. I don't chase the final product. I live for the process, the logic, the friction, the grind. The output is just a byproduct. The work is the point. I don't write code for you, for money, or to be "helpful". I write it because I love the act of building it. If you're chasing efficiency or profit, do whatever works for you. I do what I like. Why hand the process over to a model just to chase "productivity"? I don't work for you. I do this because I like it, and it doesn't hurt anyone.

I shared my Tor relay logs with an LLM and got its analysis back, very proud of my #Tor relay. > Over the past 10 days, your relay has safely transferred **611.48 GB** of sensitive user data while handling **1.2 million total connections** from both IPv4 and IPv6 users. You have protected an average of **1,022 unique connections per hour**, consistently maintaining over **1,000 active privacy circuits** at any given time. Traffic shows a healthy balance of **48.5% uploads and 51.5% downloads**. > The real-world impact is substantial: your daily average of **378 GB** in traffic has enabled the equivalent of **1,260 hours** of educational video streaming, **126,000** uncensored article downloads, and **12.6 million** secure message transmissions. > Security remains flawless, with perfect forward secrecy on all v3 handshakes, **zero malicious circuits blocked**, and automatic prevention of **151,608** potential attacks. Technically, your relay demonstrates excellent dual-stack IPv4/IPv6 support, a healthy **18.4% growth** in daily traffic, and clean synchronization with **9,901** network descriptors.

Weirdly, Jumble filters out onion relays by default. Disable it in settings? Nope, it'll still filter them. Why? You haven't enabled 'insecure' connections, which is Jumble's way of saying `wss` and `https` without the secure marker

I like this concept, a hybrid Nostr model where users run onion relays directly on their devices. This approach makes Nostr P2P. User uptime is the challenge. We still need always-online relays. But local relays on user devices can sync their data when they come online. Here's a scenario. A Nostr client first tries to send an event to the user's WRITE relays. If these are all offline, the client then tries to send the event to the READ relays of the user's contacts, specifically those contacts who also have the sender in their contact lists. When the user comes back online, their device can fetch pending events from those contact relays. I think these local relays should be WoT relays, not just personal ones. View article →

I'm now a Tor relay operator, my relay fingerprint is `C61B3104A7871EED8CCDA43D6BDC5BA54EC140F4`. #asknostr any tips?

Before AI became powerful, I used to show people what I built and how, the code itself, and it interested them because the code looked smart. That moment when you read code and say, "wow, this is smart." So they learned from it. Now, I don't feel like showing them anymore. I feel like they won't be as interested as before, because AI can do it now. They might learn from my code, but they don't need to learn from some random person's project when they can just ask AI. I don't really know. I don't write code to impress people, I write it for myself, or for a goal I want to achieve and for fun, of course. But the AI thing is somehow upsetting me, like it ruined something I enjoy doing.

I don't remember the last time I used the internet without a VPN