Basanta Goswami

Introducing designers to Bitcoin. These sessions have been the most successful orange pilling events that I have personally seen Paper and Veronika have been doing some real high quality work in Bengaluru 📝 af38ee77…

the greatest gifs i have seen in 2026

Starting a keto/carnivore diet from today. Primarily due to some constant health issues that I have been dealing with for a couple of years now, with no traditional treatment helping much. I saw my father dealing with the same issues all his life. Hundreds of doctor visits and whatnot couldn't fix anything. This was my yesterday's dinner, ate a bunch of junk one last time. 🤞 Hopefully things get better now

🤸‍♀️ 🦽🏌️

Yesterday when talking to my brother over the phone, we both heard some other people whispering things in the background when no one else was present with either of us. This isn't the first time either, I had a similar experience when talking to my mother years ago. We typically only make Signal calls but because of network issues, decided to make a regular call. I'm so done with this shit 🌜

@YakiHonne Yakihonne has become my favourite new client, both on Android and web, but I noticed a few things that are annoying me a little. I sent a DM but posting publicly as well On web: 1. I have my kind 10063 blossom server list configured, but when I try to upload a file, Yakihonne seems to not use it 2. I have a +99 after my notifications menu for a couple of weeks not and it doesn't seem to go away even after opening them 3. 2. When I was prompted to setup my `kind: 10050` relay list, there was no deduplication in the tags and I was prompted to sign an event containing the same relay multiple times

Testing ghostr

Andrej Karparhy said something interesting in his interview with Dwarkesh. We know that training these models on large amounts of data produces cognitive capabilities, that's why larger models are better at doing things. But we don't exactly know what parts of the model is actually responsible for it. If we can figure it out, then we can make smaller models much more capable even though they'd be much less knowledgeable If that's the direction it takes, then eventually I expect to have some standard smallish models that can be fed huge datasets according to your needs, and companies will use these to create knowledge specific models that can run on your regular devices. Or models that can learn stuff on demand on your own system 📝 9d4cea86…

Confidential computing is underappreciated. It enables traditional applications and software products to provide privacy for their users, while maintaining the same UI/UX, without impacting performance. However, it still requires a level of trust in the hardware being used and the infrastructure provider, such as AWS. If the application developers own the hardware, you must trust that they are genuinely utilizing secure enclaves. This trust model is simpler compared to the traditional one, where you have to rely on the integrity of system administrators, the application itself, and the infrastructure provider to ensure that your data remains private. You also have to consider the risks of rogue employees leaking data or malicious actors gaining access to your servers. With confidential computing, the focus shifts to trusting that the hardware manufacturers have produced reliable CPU and GPU components and that the infrastructure provider is actually operating them. Complete client side end to end encryption is obviously better for privacy, but then you can't do AI or analysis stuff on the server side Most existing app don't need to integrate with nostr and Bitcoin to provide better privacy and security to users. Just letting users chose a different server if needed is good enough. Users can choose to self host a server, or use one based on CC and that whole thing would be better than what we have today. The model used by @OpenSecret is also a good way of utilizing CC