[dmantis]

Tor and I2P is a better fit. Nostr is very weird. It sells itself as decentralized, but basically all frontends provide same several relays.

When those relays get subpoenas and remove your resource, you're done. You can use some unknown relays to publish, but who's gonna use them as clients outside of the defaults? It's effectively designed for shadowbans.

[sanskritical]

Many clients automatically seek, or prompt an action in 1 click to retrieve content from additional relays that a Nostr pubkey announces if said content is referenced but not available on already subscribed relays. As a publisher, you announce what relays you are currently publishing to in your identity metadata. So even if you don't specifically subscribe to a smaller relay, you can still access the content on it.

Tor and I2P are great technologies. ZLibrary, for example, runs an excellent Tor hidden service and it is usually the most reliable way to access news from the site. However, this did not remain true for a while when two of the operators were arrested. Tor and I2P require you to have infrastructure online. The point of "check Wikipedia for news" is that you can have something persist even if you do not have your servers online. Nostr is the best technology available in this category.

[dmantis]

Where does the nostr pubkey announce it? Let's say you are banned from main default relays. Is there any side channel?

[fiatjaf]

See https://how-nostr-works.pages.dev/#/pathological

[sanskritical]

You announce all the relays you publish to to the relays you are publishing to. If someone quotes your post in a post of their own, in many clients subsequent readers will be able to retrieve your content from the relays which they don't currently follow. Relay discovery mechanisms grow progressively better. I don't know of any pubkey ever banned from major relays. The operator of Damus, one of the more successful clients in history and one of the main default relays, openly engages with dissident personalities in a welcoming manner. You could probably get a "filtering transparency report" from him and others and ask if they've blacklisted any specific pubkeys and why. I am unaware of any that are currently blacklisted, and generally the network seems to defer to WoT mechanisms by clients to blacklist content.

Regardless, this remains a far more resilient persistent source of information that you don't operate than "check Wikipedia".

[dmantis]

I still don't get how to bypass a theoretical block, you need to access at least one of the current major relays or a side channel to find a follower, only аfter that you can re-translate the new set of relays. The autosub is good, but IMO the current major operators have an Elusive Joe situation because nostr is very small. Things will change as soon as people with money and government connections see it as a problem.

I totally agree with the Wikipedia argument though.

[sanskritical]

I think relay discovery by clients has improved by leaps and bounds and will continue to do so. Previously you had to give some complicated "follow me at such and such relay" thing. These days you can just tell people your npub and things seem to "just work". I don't know what's going on under the hood, I can only say that clients figuring out which relays have content seems to have improved drastically.

[cxplay]

Check NIP-65[1], this is the key to the latest Nostr client's automatic discovery feature.

[1] https://nips.nostr.com/65

[cxplay]

To improve user experience, Nostr clients typically pre-load several large relays. In fact, Nostr also supports using NIP-19[1] pointers to pass custom relay hints to the client, similar to the tracker in BitTorrent and Magnet. Furthermore, I believe that with Let's Encrypt now offering free and widespread IP certificates, domain dependency issues will be further alleviated on Nostr.

[1]: https://nips.nostr.com/19

[hypeatei]

Tor, sure. i2p requires some proxy config in your browser and you need to run a service in the background explicitly. I wish they'd release a dedicated client like Tor does.

[heavyset_go]

Both require daemons that offer SOCKS (other other protocols) proxies for browsers to route through.

If you're talking about the Tor Browser bundle, I2P has this: https://geti2p.net/en/download/easyinstall

[yapyap]

Pretty sure the whole concept of i2p is that every user also contributes to the network, which is why the config

[hypeatei]

I'm not sure why that would require more friction or why my comment was downvoted: i2p is harder to use than Tor. Adoption will be hindered by it, even in tech circles. I want to fire up an executable and maybe click through one prompt but not have to remember all the configuration steps the next time I want to use i2p.

[dmantis]

They work on that, https://geti2p.net/el/blog/post/2025/10/16/new-i2p-routers

The current focus are new rust and go implementations, but embedding lib for applications is also in the roadmap.

I also agree on hindrance. I don't understand why they don't provide a simple docker-compose at least for daemon deployment for immutable management with controlled scopes. There is an image in the dockerhub, but no proper instructions. People have to spend several hours to ensure that everything works correctly.

[FuriouslyAdrift]

There's always Freenet...

[flipped]

Freenet is promising but it's not even fully done. Experimental stage. I am also skeptic of their more usage of LLM generated code these days.

[FuriouslyAdrift]

Freenet (aka Hyphanet) is 25 years old... it's been "stable" for decades. Biggest problem is that content is sort-of hosted locally and considering how much child porn and other criminal content is in its darker corners tha can be risky.

It's also incredibly slow.

[flipped]

Stable and yet never used in live env. It's being rewritten and not at all stable. I don't expect it to become the standard anyway.

[FuriouslyAdrift]

I'm old enough to remember when there were more nodes for Freenet than Tor. I don't hink you have much experience with darknets.

[flipped]

Darknet term is mostly used to refer onion services, which have been around since 2004. Freenet was never meant for anonymity, both have separate goals.