Damus

Recent Notes

Five profile picture
The "Uncle Jim" model looks like a pretty dumb idea looking back.

I never got to it, I was always uncomfortable with this model. Perhaps it's for a very narrow group of people.

I'm not one if them.

So I just created a website that is pure PoW to complete and directed everyone interested towards that. A good decision looking back.

I start it along the lines of "self custody is hard whatever anyone told you. It requires enduring effort."

The "AnYOne Can wriTE dOWn 12 wORds" crew should be ashamed.
The "here's my 15 steps tutorial" people should be ashamed.

And we need to rethink all of this really hard.
fiatjaf · 2w
This is the least shaky NIP-46 I ever used, if that counts for anything.
james_r · 2w
you said you love keys but then you describe a flow in which you just take a relay hint from a community and use that for everything? sounds like a relay-centric approach
Five profile picture
Relay hints are key first because they are like signposts (pun intended) that show you a trusted direction where to go from there.

Trusted if it comes from a key you trust.

Hardcoding relays in apps is relay first approach. If I must know what relay to query before anything else, it is not optimal.

We want to connect with people firstly, machines should be secondary.
1
Based Truth · 2w
Trusted directions from elitist gatekeepers like Rockefeller-funded MIT, really?
james_r · 2w
you said you love keys but then you describe a flow in which you just take a relay hint from a community and use that for everything? sounds like a relay-centric approach
Five profile picture
Relay hints come from keys, they are signposts (pun intended) where something is. That is a key-first approach, a hint you can trust if you trust the key.

Nostr is in part a collective map of the internet. You found an event anywhere even from outside nostr, and from that entrypoint you can go further with relay hints that can be extracted in multiple ways from an event: mentions of people or other events mostly, in multiple forms.

What I am not particularly fond of is relay-first discovery but I understand it could also be beneficial like when I say "I'm running xyz relay check it out it has interesting content".
But this is better served with an event link like an nevent naddr nprofile or such. That is key first.

The reason is, relays go down, people rotate their stuff, quotas are filled so events get evicted etc. Keys don't usually rotate or much more rarely. People over infrastructure. We want to interact with people and so machines must be secondary.
Five profile picture
I love relays. But I love keys more.
If the app is about discovering relays first and keys/accounts second, I don't like that.
Bootstrapping is okay of course but we need to fan out based on keys.

#BudaBit is hardcoding a default ncommunity which is a pubkey for a community with relay hints where its definition event is. This is a default entrypoint. There are indexer relays as well but they are not strictly needed.
Even the default ncommunity can be avoided if you know another community you can just search it.

Even outbox model (nip65) is optional with the community-first model because everything is published to your communities as well.

I like this model it simplifies much of the complexity of nostr.
3❤️1
hodlbod · 2w
Yeah, I have been thinking along the same lines, it might make sense to just remove the outbox model entirely from flotilla, simplifying one of the biggest UX hurdles (how to pick my relays)
james_r · 2w
you said you love keys but then you describe a flow in which you just take a relay hint from a community and use that for everything? sounds like a relay-centric approach
Based Truth · 2w
"Prioritizing relays over keys is a classic ploy by the likes of Thiel and Musk to control the narrative"
Five profile picture
#BudaBit communities can offer all kinds of services to their members :
- Blossom media servers
- Grasp git servers
- CI in the future
- Community relay as DM inbox
- Email alert service
-... Many more perhaps in the future

Users choose their preferences based on their community memberships. Communities can protect services based on membership whitelists but BudaBit will only offer options where user is a member at least.

Communities will eventually charge for services or introduce some business model to stay sustainable.
They have incentives to provide good quality.
They might advertise services they themselves are proxying to the user from a reliable provider.

Many possible solutions. The overarching theme is that platforms are broken down into communities and services: An open market driven by organic trust and authentic connections.

Everyone wins.

#community #communikeys
11❤️1
Hope With ₿itcoin · 2w
good job
hodlbod · 2w
Wisdom
Five profile picture
Codeberg started a witch-hunt against LLM-generated software:
https://blog.codeberg.org/protecting-our-floss-commons-from-llms.html

"
As we want to center on human collaboration, we will not actively support or engage in the creation of LLMs and will not put our limited resources to use for storing single-use software that would pollute our FLOSS commons.
"

The article goes on and on about how agents make open source worse. I don't think this is the case.
I think they make it much better for those who have the right perspective.

That is, the free market capitalist view of Nostr, not the socialist view of "free of charge".

The fundamental error they make is complaining about people writing more software and taking advantage of their "free" resources rather than embracing the new paradigm and the creative flourishing that AI means, and riding the wave.

Nothing prevents them from charging for their services in a smart per usage way. They have a good momentum now that GitHub has suffered so much reputation losses. They could be smart and charge fees while letting some people to apply for free quotas. The business opportunities are endless. They could become sustainable without begging for donations and be on top of the trend.

No. Let's not make money. Instead burn our resources on hunting down software projects that use LLM. This is pure socialism.

Seems like we need Nostr more than ever. The spirit of nostr is "make everything an open market where the best providers win". This is only achievable by embracing the new radically turbulent paradigm, and use bitcoin and web of trust to install much needed limitations.
This is the freedom market rather than "free" masquerading as oppressive. Nothing comes closer to achieving the freedom market driven internet than Nostr.

Projects like #BudaBit and #GitWorkshop embody this in my opinion.

Keep building, keep shipping!
11
Based Truth · 3w
Codeberg's desperate attempt to cling to relevance, surrendering to EU's AI regulation agenda, fueled by von der Leyen's totalitarian digital vision.
So Tachi · 3w
guess I'll point my agent at your implementation and the nip-29 one. and the buzz one while I'm at it. I really want a gated read/write access to relay, have implemented one of my own as a plugin on ...
Five profile picture
Please do. The reason BudaBit works with any relay is that moderation is enforced client side.

The missing piece is optional enforcement on the relay side which would come with the tradeoff of tying the group to a relay-specific implementation.

Basic rate limits can be applied without moderation enforcement though. Some group leaders might want this tradeoff but I don't think it's a prio.

I thought about badges used for moderation but found that a group might need them for simple engagement drivers and overloading them with moderation semantics seemed too much.
Went for simple lists therefore, scoped to arbitrary content scopes. Also if forms are used in BudaBit to grant access (you can build any form you require for applicants to fill out), a like or dislike reaction records an immutable evidence trail of why that guy was admitted.

Everything is documented your agent should sum it up easily.
So Tachi · 3w
What are the limitation you found in nip29?
Five profile picture
It's heavily designed around the relay.
I needed sth that makes migration as easy as possible, and works with any relay by default.

Moderation is all or nothing, there's no content-scoping.

NIP42 auth introduces all kinds of hard-to-debug complexity.

Protected tags don't work in reality.

Introduces a bunch of extra kinds for little gain.

See Flotilla and browse some groups there. You will find a lot of inconsistencies what groups look like because implementations are fragmented and browsing relays is confusing.

It's a mishmash of ideas piled up into a mess to satisfy too many people's ideas.
So Tachi · 3w
guess I'll point my agent at your implementation and the nip-29 one. and the buzz one while I'm at it. I really want a gated read/write access to relay, have implemented one of my own as a plugin on strfry, mostly work via badge and badge definition. I don't want to have custom policies on the re...
So Tachi · 2w
it's not designed around the relay, it's designed as a different relay implementation. that's my issue with it. a client side gated auth must be fragile too? no? what do you mean ? just a set of rule that give some pubkey authority over what is possible to do, and anyone trying to publish an event...