working on this

Swap DNS/VPN/Tailscale for TOR maybe.

But the point is that the nsec can stay on a hardware device, and traditional authentication patterns can be used to gain permission to use it.

Nostr's customer is... Nostr? Refugees looking for alternatives? Cypherpunks and bitcoiners?

No one really knows. If I make something compatible with Nostr, am I just circle jerking Nostr development or am I making something that can be broadly useful?

Thanks for posting this. I think the space needs more open ended questions and thoughtful end products than off the cuff answers and buggy drafts.

Do you self custody your food as it grows?

I can't believe I'm asking this, but...does anyone know of a good learn Microsoft Excel tutorial/ series? Like, by the end of it you could confidently say you are proficient in Excel when applying for bullshit jobs...?

#asknostr #fml

Then again the hardware device can fail or the secure element can become obsolete and vulnerable to attacks.

I’m in England, and I just can’t understand how they manage to drive completely the other way around!! 🤣🤯😱🤓🚙🫂💜

My comprehensive COLDCARD Q Review:

3 times as expensive as building a SeedSigner, and it can't scan SeedQRs that my SeedSigner can scan.

Seed QRs seem terrible to me

But I might be overweighted on evil maid risk

SS?

An LLM is basically a bunch of layers that transform data into other data

Some of those probably at some step encode some “knowledge” of sorts

The problem is that it will work whatever you throw at it, even if it is a nonexistent thing, and return the most sensible “knowledge” according to the training data

Ou defnitly. I would go as far as claiming that AI (artificial intelligence) is a totally misleading naming.

Therefore I would be so happy if people just rely more on saying LargeLanguageModel.

Since this is what it is. A machine is given Terrabytes of Text and or maby image data. With a huge amount of metrics it creats a model that defines dependencies between words, recognizes patterns and so on. And with all this training data it tries to answer your question.

What most models I tried lack today is to know their incompetence. I mean this in no discriminating way. But most of the time this model answer any question very confidently. Independent if their answer is true or false.

And since they lack conciousnes and the ability to be very critical about their output seems to me that until now there is no Model that would have anything to do with intelligence. These are only probability based answers.

Like the students in the class that are very good in assuming what answer the professer wants to hear, without understanding the topic.

HWWs be like

Maybe it's a sign we've punished with enough hellthread notifications 😂