nostr is stateless it’s just a streaming protocol for messages, there is no ordering (the same way there is no packet ordering in UDP). Looks like what you need is something built on top of it. Right now nostr relays are just that, real time message relays, they are not even supposed to store the messages.
I've brainstormed ways to solve this on top of Nostr. Perhaps I could write a program that downloads all of the history of Nostr (or perhaps just the users I'm interested in?) and then my server would sign when I received the messages and create order out of that.
Then if I get a message that claims to be from the past at some point in the future, I'd instantly be able to see that. I'd also be able to periodically ping relays and see what messages that are no longer being relayed and I'd have some curiosity about why that is happening.
I just kind of wish this was baked into the protocol. But I sailed here from the Scuttleverse. I'm open to ideas about how to do this well.
Is this even possible? Won't relays throw away older messages at some point? Or are they supposed to scale infinitely?
There are some relays that have no storage at all, they only relay messages between connected clients and then discard the messages.
I'm betting on webtorrent paired with seedboxes - your peer just has to serve whatever mp3/mp4 you're hosting to the first few clients, and as long as their browser windows are open they can help seed any viral growth, handling DDoS/hugs-of-death without issue. If there are multiple hosts in a community they can enter peering agreements and sync their seedboxes to keep media available, solving bitrot
if only I could figure out how to get paid to implement it
The financial system at large is built to discourage people transferring funds anonymously for myriad reasons, and most people aren't interested in keeping all their cash inside the black market (no exit nodes in torspeak)
from my view it's been settled for near 2 decades now, IPFS just gets more press in certain circles
They were using Gun db at some point too maybe but it's been some time since i checked, maybe it was a clone.
Just post the SHA256 hash on the note and when publishing keep sending the file to some file hosting server (as done today).
The file host will be gone one day, hopefully some other server will have the same hash.
SHA256 is not enough? Just use a tag like file:sha256:blablablabla and you now support better hashes in the future.
Nostr isn't about blockchain nor a crazy fear of losing data. It is intended for sharing notes that someone, somewhere might have interest to piece back together in a few centuries from now.
Maybe time for a NIP on this topic?
If we don't want a global scope for blob sharing maybe the relay only responds if we are friends/members of the relay.
I think the main difference is that in IPFS the content ids (equivalent to torrent hashes) support multiple encodings, etc, and peer connections are more flexible than bittorrent "nodes" in their addressing - i.e. multiple protocols/transports, etc.
> First, I want a replication strategy. Nostr messages get lost in time, and many of the clients end up just blasting an entire message history at your client. Because there's no clue in the protocol how messages are related other than a timestamp this also means you can fake timestamps and write fake messages in the future or back in time
You can do this with email or git too and it doesn't make it any less useful. I actually like the backdating feature as it allows you to copy your account to a new key.
As for replication, at damus I am working on https://github.com/damus-io/nostrdb which is intended to be a "sqlite for nostr". I plan on implementing set-reconciliation based syncing with strfry relays (using a technique called negentropy), so that replication is very efficient.
> Second, I don't like that many Nostr clients are using the same signing key for messages as they do for lightning transactions.
This is simply not true.
> Third, someone needs to delete some of these NIPS. The arms race to make Nostr as complex and difficult as possible to implement is not going to do much for the ecosystem in the long run. In the beginning Nostr was simple to implement from scratch, they should get back to that!
All nips are optional except for nip01, you can ignore them all for the most part.
> Fourth, it needs a dedicated blob store protocol. Yah, I know IPFS isn't great but someone should come up with something that is simple and works.
It does not, in the same way email or git or any text-based protocol doesn't need a dedicated blob store. These are separate concerns and they should be a separate protocol. nostr clients can of course integrate and link to any blob store it wants via new NIPs that describe this. I believe there are a few already in the nips repo.
Cheers!
> This is simply not true.
I'm sorry if I'm wrong about this one. How does it work?
I guess I assumed it was based on the private key you generated when you got started on Nostr and if that private key became compromised then the Zaps could go to whoever had the private key.
Perhaps there is a document somewhere that explains this.
The basic idea is that a lightning node will detect when the invoice with a nostr note inside is paid, and then send the receipt to nostr as a nostr note, with the original bolt11 invoice inside with the signature from the user who sent the zap.
It's all described by NIP-57, a spec I put together to support this:
https://github.com/nostr-protocol/nips/blob/master/57.md
I was working on c-lightning at the time and I thought it would be really cool to replace the "like" button with an instant bitcoin micro-payment. I think it worked out quite well! There are many sites utilizing zaps in all aspects of the protocol, such as a decentralized market for AI job requests (data vending machines), zapgoals and zap fundraisers. All built on this note type. protocol synergy!
I guess what confused me is I've used Nostr clients where everyone has a Zap button. Who is holding onto those lighting receive addresses until they specify where they want the funds to go?
Like if someone Zaps me right now, I haven't specified a place for those Zaps to go. Do I call the Damus staff and they send the total of the micropayments Zapped to me over Nostr to a Bitcoin receive address? I don't think anyone has Zapped me, I'm just trying to wrap my mind around how it works since I was wrong earlier.
Maybe I need to set aside an afternoon and read about how Lighting works, perhaps I just don't get that protocol.
then you can't be zapped
you can only be zapped if your profile has a tag in it that tells people where to send your zaps to
without that, a zap button shouldn't show up, and if it does show up, it shouldn't do anything
I didn't realize every single person who downloads a zap-supporting client had to also set up a separate app to handle the micro-transactions and host that app somewhere where it is available 24/7 in case their post gets zapped.
Haha, like Nostradamus, very clever name!
That remembers me off project.ioni.st and now iam sad.
> You can do this with email or git too and it doesn't make it any less useful.
In the sense that you can fake timestamps, you're of course correct on git and anything else beyond public blockchains.
In the other sen, git is a counter-example of what I take the author's main point to be, in that commits do have an mandated relationship and linkage that Nostr noted don't.
1) With regard to message replication/ordering, the relays are supposed to reject messages that are received with a timestamp that deviates from the server clock beyond some max delta, but it's not clear they always do this. Nostr in its current state (multiple relays hosting everything and letting clients sort it out) is optimizing for redundancy over efficiency — as the ecosystem matures I think this will get sorted out. I have one idea in particular that I think could really help, but it's beyond the scope of this comment.
2) In the very early days you had to copy/paste your key into some clients, but this is not the case anymore. There are several web extensions now that hold your key securely and expose a limited interface for clients to sign messages.
3) You're right - there are too many NIPS (that's "nostr implementation possibility" for those unaware). But fortunately the only required NIP is NIP-01 which is basically just a standard for how to sign and exchange json between clients and relays. I don't think any client implements all the NIPs. In fact, there has been a lot of talk on nostr about 'microapps' (apps that specialize in a narrow use case) and that's fine — maybe because nostr was birthed in the context of being a Twitter alternative we've been thinking that all clients need to support the whole protocol, but I don't really think it will turn out that way. Since all apps are interoperable with a shared user identity, it's possible for clients to be complementary to each other in a way that hasn't previously been incentivized.
4) Regrading blob storage — that's what NIP-94 was created for. You can in principle create a "file" message containing whatever metadata you want. This can be a torrent magnet link or IPFS cid or whatever.
For what it's worth, I've personally never had as much coding anything than I have working on nostr. There's a core group of people that are very excited about this, and it's really motivating to get immediate feedback every time I push an update. No doubt the whole thing is very messy compared to a corporate-sponsored project like (for example) Bluesky. My bias is toward ecological systems tending to win in the long term. We'll find out.
I believe Nostr is doing way better than Bluesky at achieving the goal of being a distributed social network.
If everyone cares about privacy then privacy focused places ought to be full of on average smart people having intelligent conversations. If few people feel the need to bother you end of serving largely the paranoid and the odious and this can easily be a self re-enforcing trend as bad drives out the good.
I feel like the space could use some innovation if its produce communities worth attending to without strong centralized moderation. Instead of bubbling up the top 100 things the community as a whole thinks are worth reading or the top 100 things that someone thinks I might be willing to engage with including 60 things selected because they are liable to make me angry or succeed in wasting my time why not have a personalized analysis done on MY side or on a computer I rent to pull in 10,000 things and figure out which 100 I actually want to engage with based on my own aims not advertisers.
I would probably pay for that especially if it could be plugged into a lot of common platforms to filter out the crap.
Maybe running a local spam filter on one of these decentralized platforms is the way to go. I wouldn't mind running LLMs and removing all the content from Trump crime family fans. Would be nice to detect and reject LLM-generated adtech bullshit while we are at it.
> Second, I don't like that many Nostr clients are using the same signing key for messages as they do for lightning transactions.
> Fourth, it needs a dedicated blob store protocol.
You should make some Nostr Improvement Proposals
> First, I want a replication strategy.
xx Network has message replication built in.
> Third, someone needs to delete some of these NIPS.
xx Network lets you delete messages.
> Fourth, it needs a dedicated blob store protocol.
For blobs, clients should be able to plug in to any 3rd party blob storage (e.g. Crust Network). This secondary storage is needed only for large attachments and isn't strictly required in messaging. If you think about it, you could upload large attachments anywhere and send links in messages. Look how Teams or Outlook work with One Drive.
First, I want a replication strategy. Nostr messages get lost in time, and many of the clients end up just blasting an entire message history at your client. Because there's no clue in the protocol how messages are related other than a timestamp this also means you can fake timestamps and write fake messages in the future or back in time. This doesn't have to be an append-only log, but you need some idea of message order to avoid wasting bandwidth to get someone's timeline and detect when a message has been posted out of order.
Second, I don't like that many Nostr clients are using the same signing key for messages as they do for lightning transactions. We don't know how many of these web Nostr clients are secretly sending your private keys back to their servers, and everyone will run from Nostr as soon as some untrustworthy dev starts emptying lightning wallets.
Third, someone needs to delete some of these NIPS. The arms race to make Nostr as complex and difficult as possible to implement is not going to do much for the ecosystem in the long run. In the beginning Nostr was simple to implement from scratch, they should get back to that!
Fourth, it needs a dedicated blob store protocol. Yah, I know IPFS isn't great but someone should come up with something that is simple and works.
Anyway best of luck to the people who are working on the project. If anything we learned more about how these protocols can work, even if Nostr isn't going to be the winner in this space in the long term.