Damus
Basanta Goswami profile picture
Basanta Goswami
@basantagoswami

bitcoin maximalist • অসমীয়া

○ building freedom tech @ unsigned.in
○ teaching bitcoin @ xonghoti.com
○ non bitcoin dev work @ infini8tystudio.com

कर्मण्येवाधिकारस्ते मा फलेषु कदाचन |
मा कर्मफलहेतुर्भूर्मा ते सङ्गोऽस्त्वकर्मणि ||

Relays (3)
  • wss://nos.lol/ – read & write
  • wss://relay.snort.social/ – read & write
  • wss://relay.damus.io/ – read & write

Recent Notes

basantagoswami profile picture
I tried making a comment on my keyboard app using a browser extension to see the event structure, and it seems like your implementation has missing parent item tags from NIP-22

The event generated looks like this:

{
"kind": 1111,
"tags": [
["A", "32267:573010eb0f43fcfc253c2895828d30bf6939b056384e2cc6648f40096acaef4e:in.unsigned.keyboard"],
["K", "32267"],
["P", "573010eb0f43fcfc253c2895828d30bf6939b056384e2cc6648f40096acaef4e"],
["d", "1.0.0"]
],
"content": "My comment..."
}

The tags there should look like:

"tags": [
// root scope (uppercase)
["A", "32267:573010eb0f43fcfc253c2895828d30bf6939b056384e2cc6648f40096acaef4e:in.unsigned.keyboard", "<optional-relay-url>"],
["K", "32267"],
["P", "573010eb0f43fcfc253c2895828d30bf6939b056384e2cc6648f40096acaef4e", "<optional-relay-url>"],

// parent item (lowercase - same as root for top-level comments)
["a", "32267:573010eb0f43fcfc253c2895828d30bf6939b056384e2cc6648f40096acaef4e:in.unsigned.keyboard", "<optional-relay-url>"],
["k", "32267"],
["p", "573010eb0f43fcfc253c2895828d30bf6939b056384e2cc6648f40096acaef4e", "<optional-relay-url>"]
]

I'm not sure about the added d tag, but the parent item tags are required in the NIP

Also, not sure if it was a deliberate choice for some reason, but if not, this is probably a bug
basantagoswami profile picture
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 @note1n4xw4...
basantagoswami profile picture
I will often open a client and wonder what standard/NIP it uses, and there would be no information about it. Many times I had to clone the repo and ask Claude Code to find the kinds used there and search for any open PRs etc in GitHub to find the actual NIP

basantagoswami profile picture
Nostr clients should add a link to the NIPs that they implement, maybe also a link to the apps that are interoperable with it. More important for special purpose clients compared to genral ones
basantagoswami profile picture
By everywhere I mean it has definitely been adopted by companies and developers from many different places, which I didn't expect to happen for a new L2 this fast

They built a great product and SDK, although they should have tried to make an open protocol instead (with a BIPs, NIPs like specification). Also, their biggest marketing point is the stablecoin and token support, and privacy is not a priority yet, even though Ruben Somsen's original statechain design six years ago already came with blind signatures for ecash level privacy