Because nobody has convinced me that nostr is more decentralized than a DHT
https://twitter.com/peterktodd/status/1759151430336737645?s=46
Discussion
If I understand, the only benefit to keets protocol is that all clients are also relays. However that is a feature of nostr, not a bug. I host my own client and relay on my sever, but I don't want my phone to have to contain all of that data too.
My measuring stick is which platform is more decentralized. I'm not a protocol developer, but i can sort of see how a DHT might be better. Bittorrent proved that DHTs work, so long as the DHT keys are network endpoints, yada yada. What he said in that post. Again, i'm no expert here.
I went through most of the comments on the actual stacker news post, and it seems like the authors fears are majorly solved by nip65. Considering I run my own relay and client, the protocols are equally decentralized to me. I just the requiring all users to run relays is extra overhead. Do I think it's nice that some clients also store your data locally? Yes, but that should be up the the consumer. The user should be free to decide if it's valuable to them to store their notes on their phone or if they trust their relays.
I don't necessarily care about how decentralized a protocol is in early stages, but how decentralized it could be. The benefit of nostr is that the client and the relay is separated, so the user has a choice for both. There is a single client/relay on hole punch for messaging, so until there are more than that, it's not decentralized right?
The number of implementations of the protocol may grow in time if this thing catches on. It's too early to tell. One thing i have noticed is that nobody wants to run their own hardware. I'm waiting for all the nosr relays to live on aws, azure etc. Is that decentralized?
If it opens up to the general public, I would anticipate 75% of relays will run in cloud providers because home hardware only goes so far.
That's a separate issue though, that is on alternative cloud providers to solve (for example by implementing more terraform apis).
People will only want to run their own hardware when stuff like startos, umbrella, etc give a more iPhone style UX. Idk if you tried them, but I am a devops engineer and startos is still pretty shitty from a UX perspective that I've been considering running umbrel side by side to compare.
I have not tried startos, etc, only kubernetes. Umbrel i hear is easy, but also haven't tried. I don't see the incentive to run a node. Especially in the cloud. Seems expensive.
Again, the protocol has nothing that requires anyone to store anything. It’s completely agnostic. Your concern isn’t something that is actually happening, everyone can choose what they do or don’t want to host/share/save.
You don't have to store your own messages/data?
There's nothing about the protocol that requires it, but practically every message app you do does this because always loading remotely would be really slow. Keet does store everything locally just because it works better, but that is one app and they simply haven't added granular controls in that way. Right now you can delete any file that you've downloaded if you don't want to hang onto it. But there's nothing indicative of the protocol that says anything about how a user must handle the data. Same as you can just "leech" a torrent file, any app can choose their degree of data storage or network reliance as they would like.
That makes sense, but then nostr:npub1tsrh6g3etzdahj0jgnvec86z0sqw7yhkcq9l578m4tyv8v5976yseatmmg 'a point was that pear is more decentralized, if users don't even have to store their own data, how does that make it more decentralized? Doesn't that just mean it's like traditional social media or apps with the option to be decentralized? Won't making it optional just make things trend towards centralization?
Obviously let's not make the decentralization measurement a false dichotomy. Maybe we are splitting hairs? I was more focused on the DHT side of things, but i still don't understand that fully.
I definitely think both are decentralized. Just trying to discuss the incentives in either protocol. I guess technically both protocols allow leaches. But I'm failing the see the benefit of one VS the other.
Well, based on my lack of understanding the DHT, the DHT is harder to attack. I am not aware of any attacks yet on nostr, but that doesn't mean it hasn't. Bittorrent has withstood the test of time. I don't see why nostr can, or maybe already has implemented the features of bittorrent.
The massive benefit of the pear model is the ability to distribute a file or application to millions of people with ZERO infrastructure cost.
What’s the biggest issue with relays and making a global network with #Nostr? Monetization. Even right now the costs of having a popular relay are very significant. This means that the network can’t “passively” exist at scale, and it’s a huge centralization pressure because nobody will donate $5,000 per month to keep relays running just to help the network.
In the Holepunch alternative, this cost is essentially nonexistent. Every individual feed will act like a live torrent and you simply download from all of the other followers, and they can selectively store old content as they wish and it’s automatically available from those who do.
The more people who join/follow/chat/etc, the faster and more responsive and less costly all of it becomes. Whereas in the relay model, the bigger it gets the more expensive it is and the worse the experience is without aggressive scaling, which means they need money.
In short, the client/server model is a major reason why the internet centralized around platforms. The Holepunch protocol fixes the client/server barrier.
Hmm, that sounds good. But maybe when it's open sourced it'll gain traction on nostr protocols and they can be integrated. But also, couldn't a nostr client replicate this by having a relay in the client?
Holepunch is already FOSS, keet isn't yet. Holepunch is built with DHT (P2P), Nostr is built with relays and clients (decentralized client/server)
That’s the beauty of P2P, you can easily choose. I think you have the dynamic the opposite. One of the biggest centralizing forces on the web is having to use specialized software, skills, and networking to host a web facing server. Relays don’t fix this issue unfortunately, it only divides up the work, but it keeps the fundamental friction with being a host.
With Pear, anyone can freely host as much or as little as they want, and there’s zero setup or special skills required to be a “seeder.” It is literally as if you built a live, dynamic web experience on top of BitTorrent. Mobile can simply view whatever it chooses from the network of seeders/peers, and keep it if desired, or not.
I think #Nostr is a fascinating project with enormous potential, but I think Pear/Holepunch goes one layer deeper, and more fundamentally changes the entire nature of the web infrastructure. It’s already far better for me in transferring files, sensitive data, and doing live video calls between my devices and everyone I work with. And it’s not even much of a comparison. This is just the very beginning, the more it grows, the *better* it will work. That’s going to change the game.