4 ms·

I see a problem. I'd say a majority of the posts on Nostr are media posts (mostly images) and the network relies on Imgur and other image hosting services for a

by mikae1 4y ago

I see a problem. I'd say a majority of the posts on Nostr are media posts (mostly images) and the network relies on Imgur and other image hosting services for all content. Not very decentralized in practice.

leesalminen 4y ago
It's weird how "decentralized" means something different to every individual. Having the ability to choose where to host my images is what I'd call decentralized.
- mikae1 4y ago
  Perhaps the greatest problem with this setup is that the post content is separate from the post itself. I hope that most of us remember what a giant mess hosting images for forums on free image hosting services (that were later shut down) caused.
- kvathupo 4y ago
  As I understand it, the raison d'être of nostr is to use relays to store your data, which is presented on clients. So if a front-end bans you, then you can grab your data off any relay. If imgur shuts down, you're out of luck! Bluesky goes further with the AT protocol for more general data and algorithms iirc
  - kevingadd 4y ago
    And as it happens, imgur is revising their TOS and deleting a bunch of content
  - drexlspivey 4y ago
    Relays don't (have to) store your data, they relay in real time via websockets. Clients should store the data if they care about persistence
    - derefr 4y ago
      How is this not just IRC then? The thing that makes Twitter/Mastadon/etc not just IRC, is that in these, you can scroll back as far as you like—specifically, past where your client joined the network.
      - partialzero 4y ago
        Yeah, seems like a (barebones/underspecified) IRC.
      - nunobrito 4y ago
        Because IRC still needs a lot of things to keep running (be usable) and works mostly for chat messages. NOSTR isn't twitter nor IRC, you can take your posts which can contain just about any type of data inside and build an art gallery, a forum or dump for meteorological information. You can scroll back in the past. It is up to the relay if they have interest in keeping your data or not. If nobody cares about your data, you can still preserve it yourself without barriers for that.
        derefr 4y ago
        I’m not a publisher/journalist; I don’t have any posts. I’m a historian and archivist. And in those roles, I care about being able to retroactively access data that neither publishers nor relays considered to have value / be worth preserving at time of publication, but which instead was preserved due only to the “accident of history” of them publishing in a durable medium. Compare and contrast: digitized newspapers available on microfiche.
        nunobrito 3y ago
        I'm also an archivist. I've created http://arquiva.me/ http://arquiva.me/ from scratch which is archiving tweets (and other sources) related to people in Portugal. That archive preserves history in open format that anyone in the next centuries can access. As archivist, we both need to pick our sources manually at some point. Nostr is quite a good tool for archives, since its format is flexible (long term preservation) and can work even fully offline.
anarchogeek 4y ago
There's a new nip to cover the metadata for images and media including being able to use torrents for file hosting. https://github.com/nostr-protocol/nips/blob/master/94.md https://github.com/nostr-protocol/nips/blob/master/94.md
psychlops 4y ago
The problem is that a decentralized service links to a centralized service?
- mikae1 4y ago
  https://news.ycombinator.com/item?id=35692399 https://news.ycombinator.com/item?id=35692399
rektide 4y ago
Is it a problem? Right now it doesn't seem to be a practical problem. I agree that where we're headed, it seems like services like this will only continually degrade, that a huge amount of the corporate-run internet is undergoing radical #enshittification at an alarming rate, & the nation-states have strapped afterburners onto this hellbound-sled by starting to make the terms of use for these already imperfect the feudal- oops i mean corporate- data-keeps. New age-verification identity-verification to post stuff has scaled up with shock & awe speeds, with Imgur just burning down huge swarths of the internet. So I guess yeah it has become a practical problem alraedy. Oh and there's other signs of horror/intensification all about. Specs like Mobile Document Request opening up the Jevon's paradox of making it easier to request government id online are going to make a very shitty 203X's that greatly piss over the internet legacy we have. https://github.com/WICG/mobile-document-request-api/issues/6 https://github.com/WICG/mobile-document-request-api/issues/6 But also... we have the saints of human history, https://archive.org https://archive.org, ticking along doing the good deeds. The more rag-tag ArchiveTeam folks. They keep saying it's not for archiving, but I really hope WebPackage / WebBundle specs take off, that we build a norm of take-away sites that we can retain. (Caveat: right now Chrome has zero interest in letting you use old snapshots, but I have zero faith this limited security totalitarianism will hold, given that Certificate Transparency lets us know that indeed this content did come from X site at Y time & had the right cert then.) In general, it's all the web, so it only sort of matters that the thing goes away. We need to update the maxim, "Cool URIs Dont Change" (https://www.w3.org/Provider/Style/URI https://www.w3.org/Provider/Style/URI). Sometimes the resources go away. But the URI remains. And we can spread backups, share the content, even when the hosts vanish, because the web is so cool like that.
- mikae1 4y ago
  > Is it a problem? Right now it doesn't seem to be a practical problem. It will be the day these image hosting services die (and they do, all the time). This setup separates the post content from the post itself. I hope that most of us remember what a giant mess hosting images for forums on free image hosting services (that were later shut down) caused. Apparently there was a discussion just a few days ago about Imgur deleting images that are not associated with a user account[1]. Just imagine all the broken image embeds that will cause… [1] https://news.ycombinator.com/item?id=35636190 https://news.ycombinator.com/item?id=35636190
yellowapple 4y ago
IPFS seems promising as a solution here.

by mikae1 4y ago

leesalminen 4y ago

It's weird how "decentralized" means something different to every individual. Having the ability to choose where to host my images is what I'd call decentralized.

mikae1 4y ago
Perhaps the greatest problem with this setup is that the post content is separate from the post itself. I hope that most of us remember what a giant mess hosting images for forums on free image hosting services (that were later shut down) caused.
kvathupo 4y ago
As I understand it, the raison d'être of nostr is to use relays to store your data, which is presented on clients. So if a front-end bans you, then you can grab your data off any relay. If imgur shuts down, you're out of luck! Bluesky goes further with the AT protocol for more general data and algorithms iirc
- kevingadd 4y ago
  And as it happens, imgur is revising their TOS and deleting a bunch of content
- drexlspivey 4y ago
  Relays don't (have to) store your data, they relay in real time via websockets. Clients should store the data if they care about persistence
  - derefr 4y ago
    How is this not just IRC then? The thing that makes Twitter/Mastadon/etc not just IRC, is that in these, you can scroll back as far as you like—specifically, past where your client joined the network.
    - partialzero 4y ago
      Yeah, seems like a (barebones/underspecified) IRC.
    - nunobrito 4y ago
      Because IRC still needs a lot of things to keep running (be usable) and works mostly for chat messages. NOSTR isn't twitter nor IRC, you can take your posts which can contain just about any type of data inside and build an art gallery, a forum or dump for meteorological information. You can scroll back in the past. It is up to the relay if they have interest in keeping your data or not. If nobody cares about your data, you can still preserve it yourself without barriers for that.
      - derefr 4y ago
        I’m not a publisher/journalist; I don’t have any posts. I’m a historian and archivist. And in those roles, I care about being able to retroactively access data that neither publishers nor relays considered to have value / be worth preserving at time of publication, but which instead was preserved due only to the “accident of history” of them publishing in a durable medium. Compare and contrast: digitized newspapers available on microfiche.
        nunobrito 3y ago
        I'm also an archivist. I've created http://arquiva.me/ http://arquiva.me/ from scratch which is archiving tweets (and other sources) related to people in Portugal. That archive preserves history in open format that anyone in the next centuries can access. As archivist, we both need to pick our sources manually at some point. Nostr is quite a good tool for archives, since its format is flexible (long term preservation) and can work even fully offline.

anarchogeek 4y ago

There's a new nip to cover the metadata for images and media including being able to use torrents for file hosting. https://github.com/nostr-protocol/nips/blob/master/94.md https://github.com/nostr-protocol/nips/blob/master/94.md

psychlops 4y ago

The problem is that a decentralized service links to a centralized service?

mikae1 4y ago
https://news.ycombinator.com/item?id=35692399 https://news.ycombinator.com/item?id=35692399

rektide 4y ago

Is it a problem? Right now it doesn't seem to be a practical problem. I agree that where we're headed, it seems like services like this will only continually degrade, that a huge amount of the corporate-run internet is undergoing radical #enshittification at an alarming rate, & the nation-states have strapped afterburners onto this hellbound-sled by starting to make the terms of use for these already imperfect the feudal- oops i mean corporate- data-keeps. New age-verification identity-verification to post stuff has scaled up with shock & awe speeds, with Imgur just burning down huge swarths of the internet. So I guess yeah it has become a practical problem alraedy. Oh and there's other signs of horror/intensification all about. Specs like Mobile Document Request opening up the Jevon's paradox of making it easier to request government id online are going to make a very shitty 203X's that greatly piss over the internet legacy we have. https://github.com/WICG/mobile-document-request-api/issues/6 https://github.com/WICG/mobile-document-request-api/issues/6 But also... we have the saints of human history, https://archive.org https://archive.org, ticking along doing the good deeds. The more rag-tag ArchiveTeam folks. They keep saying it's not for archiving, but I really hope WebPackage / WebBundle specs take off, that we build a norm of take-away sites that we can retain. (Caveat: right now Chrome has zero interest in letting you use old snapshots, but I have zero faith this limited security totalitarianism will hold, given that Certificate Transparency lets us know that indeed this content did come from X site at Y time & had the right cert then.) In general, it's all the web, so it only sort of matters that the thing goes away. We need to update the maxim, "Cool URIs Dont Change" (https://www.w3.org/Provider/Style/URI https://www.w3.org/Provider/Style/URI). Sometimes the resources go away. But the URI remains. And we can spread backups, share the content, even when the hosts vanish, because the web is so cool like that.

mikae1 4y ago
> Is it a problem? Right now it doesn't seem to be a practical problem. It will be the day these image hosting services die (and they do, all the time). This setup separates the post content from the post itself. I hope that most of us remember what a giant mess hosting images for forums on free image hosting services (that were later shut down) caused. Apparently there was a discussion just a few days ago about Imgur deleting images that are not associated with a user account[1]. Just imagine all the broken image embeds that will cause… [1] https://news.ycombinator.com/item?id=35636190 https://news.ycombinator.com/item?id=35636190

yellowapple 4y ago

IPFS seems promising as a solution here.