Just to share some experience: 100% privacy on these kind of networks is more a sympton of talking to software engineers than the radio reality observed on mesh networks.
Very often what comes as top priority is communication itself, which means radio-amateur radio frequencies where encryption is only available for signing messages but not the content.
Reticulum can be 100% private as it can choose to send packets in plain text. Can also broadcast messages without a specific target receiver, these are features that are necessary for radio transmission on amateur frequencies.
On LoRa: attention that default reticulum is still "too talkative". Meaning that in some LoRa frequencies there is a legal limit to how many messages (transmission time) your device can send per hour and even the automated announcements (to say "Hello, I'm here") quickly waste that available space.
On Internet: reticulum is great. I'm seeing about 700 connected devices across different hubs.
People often compare reticulum to Meshcore or Meshtastic but they often neglect one important aspect: those are exclusive LoRa implementations. Reticulum is the same protocol (language) no matter the network underneath. This means that you can connect a phone via BLE to a LoRa link which on the other side is connected on the Internet and then flies down to another BLE somewhere. This is not common to find on reticulum but possible.
There is a lot of good work on the topic. Maybe it will surface here on HN soon when more stable. In October there is an yearly event in Coimbra, Portugal where these topics are presented and tested in deep.
I had a lot of hope for reticulum, but it's a one man project with a near burned out maintainer, taking on an absolutely monstrous task.
I'm now cautiously optimistic. This or something similar is likely the long term future of mesh networks, but in the short to medium term, meshcore is the way to go.
The protocol itself is more mature than the maintainer situation suggests. I built a complete independent implementation from the public spec in a few sessions. If one person can do that without touching the reference codebase, the foundation is solid. The risk is to Mark's implementation, not to Reticulum as a protocol.
> Reticulum does not use source addresses. No packets transmitted include information about the address, place, machine or person they originated from.
Sure, but if it’s anything like MeshCore, a few observer nodes will be able to see which repeaters the message first entered the network from and from that know roughly where it was located.
Reticulum packets include a hop count of how many nodes it’s been through (to prevent packets looping indefinitely) and the destination address. Two way links use tables on each node to reverse the route back to the origin and don’t use origin addresses.
The worst part privacy wise could be the announce packets which include a hop count and the address for the origin so if you had enough interfaces you could roughly say where a node is on a network graph (geo locating that is an entire other problem)
Privacy also depends on what interfaces you communicate over, eg TCP reveals IP addresses to the nodes you directly connect to (combine with hop counts to see if a packet definitely originated from that node), LoRa gives an idea of how close it might be because of its range limits and the fact radio emissions can be located.
Mark accepts outside contributions over LXMF using git patches. The main communication method is over Reticulum Relay Chat using clients like NomadNet (made by Mark, unintuitive but works well) which is basically IRC and is purely sending messages to everyone in that channel at that time and is centralised around self hosted servers.
While the project is interesting from a technical perspective, there are some serious issues. The main implementation has a vague, non-standard licence, but this is not overly problematic as alternative implementations can simply be used instead. A bigger problem with Reticulum is that it does not appear to offer robust protection against abuse. Messages are protected by cryptography, but there are many other ways to render a network unusable, which will inevitably happen on a larger scale.
With the Artificial Inanity systems and the Rampant Orphan Botnet Ecologies (ROBE) beginning to spew tremendous amounts of crap (a technical term) onto the Ret, things are getting tricky for us ITA.
Rayfish, Tailscale, ZeroTier, Netbird etc are a different category of tool, designed to facilitate connectivity between you and your own pool of machines. They don't generally do much in the way of multi-hop routing, but rather they orchestrate VPN tunnels on top of another routed network (like the internet).
Reticulum would be better compared to Yggdrasil, cjdns etc as routing schemes for larger networks or true meshes, which can work independently of (and have no dependency upon) the internet.
The author of Reticulum's take is that the implementation is the specification, so if another implementation works with the standard Python one and implements enough of it's features then it implements Reticulum.
I suppose it means it's easier to develop the protocol only having to update one Python codebase but it does mean other implementations are kinda in a state where there isn't anything to properly verify they work.
There's also a issue with quite a few vibe coded implementations that don't actually properly work filling the space.
There is a published spec, I built a clean-room implementation from it. And mine makes at least two Python implementations. The protocol itself is solid, the spec just isn't marketed.
People don't seem to get this was a genuine question. What is a network without people using it? Anyone here on reticulum?
The internet may be ripe for disruption, as it has been domesticated for corporate use. It is no longer ours. Browsers have become an operating system unto themselves. Something much simpler would work also. Corporate bloat is continuing to pile up, while markdown evolves to do everything we want.
[Gnomic comments go over people's heads (I'm guilty of writing these too). Gotta think in UX terms - write to make your audience's lives easier, not harder.]
You're right, and I think a lot of these projects hit the same problem: "what can this offer me that centralized platforms can't?"
Take the Gemini protocol: if someone invents a "Gemini killer app", it can be replicated over the web because the web's features are a superset of Gemini's features. Medium is the centralized equivalent of traditional blogging. The whole indie web could be emulated by Meta (it's not big enough for them to care). Much like capitalism eating subcultures, the existing internet will consume, commoditize and re-sell all alternatives.
The only exceptions I can think of are decentralization itself (centralization can only offer an emulation of decentralization) and freedom from corporate censorship, and not enough people care about them to make the switch.
I'd love to hear counterexamples if anyone has them, because I don't like this conclusion.
People do use it but it's definitely a small community. It has the NomadNet Page protocol which is basically small no client side script webpages (forms and links only) in a fairly bandwidth efficient form.
Just to share some experience: 100% privacy on these kind of networks is more a sympton of talking to software engineers than the radio reality observed on mesh networks.
Very often what comes as top priority is communication itself, which means radio-amateur radio frequencies where encryption is only available for signing messages but not the content.
Reticulum can be 100% private as it can choose to send packets in plain text. Can also broadcast messages without a specific target receiver, these are features that are necessary for radio transmission on amateur frequencies.
On LoRa: attention that default reticulum is still "too talkative". Meaning that in some LoRa frequencies there is a legal limit to how many messages (transmission time) your device can send per hour and even the automated announcements (to say "Hello, I'm here") quickly waste that available space.
On Internet: reticulum is great. I'm seeing about 700 connected devices across different hubs.
People often compare reticulum to Meshcore or Meshtastic but they often neglect one important aspect: those are exclusive LoRa implementations. Reticulum is the same protocol (language) no matter the network underneath. This means that you can connect a phone via BLE to a LoRa link which on the other side is connected on the Internet and then flies down to another BLE somewhere. This is not common to find on reticulum but possible.
There is a lot of good work on the topic. Maybe it will surface here on HN soon when more stable. In October there is an yearly event in Coimbra, Portugal where these topics are presented and tested in deep.
https://ratspeak.org/ seems like a modern fork of Reticulum built in Rust and with a fairly active community
Unfortunate name for the project.
I had a lot of hope for reticulum, but it's a one man project with a near burned out maintainer, taking on an absolutely monstrous task.
I'm now cautiously optimistic. This or something similar is likely the long term future of mesh networks, but in the short to medium term, meshcore is the way to go.
The protocol itself is more mature than the maintainer situation suggests. I built a complete independent implementation from the public spec in a few sessions. If one person can do that without touching the reference codebase, the foundation is solid. The risk is to Mark's implementation, not to Reticulum as a protocol.
> Reticulum does not use source addresses. No packets transmitted include information about the address, place, machine or person they originated from.
Sure, but if it’s anything like MeshCore, a few observer nodes will be able to see which repeaters the message first entered the network from and from that know roughly where it was located.
Reticulum packets include a hop count of how many nodes it’s been through (to prevent packets looping indefinitely) and the destination address. Two way links use tables on each node to reverse the route back to the origin and don’t use origin addresses.
The worst part privacy wise could be the announce packets which include a hop count and the address for the origin so if you had enough interfaces you could roughly say where a node is on a network graph (geo locating that is an entire other problem)
Privacy also depends on what interfaces you communicate over, eg TCP reveals IP addresses to the nodes you directly connect to (combine with hop counts to see if a packet definitely originated from that node), LoRa gives an idea of how close it might be because of its range limits and the fact radio emissions can be located.
reticulum don't just propagate over lora. IT can practically propagate over anything it supports (network, lora, bluetooth, etc).
it isn’t
Why are there not more developers? Is ratspeak where they all prefer to be?
Also reticulum and crypto seem like they go together. Non-http crypto transactions?
What about entirely-non-http blockchains?
A new bitcoin that is completely off the internet as we know it??
Mark accepts outside contributions over LXMF using git patches. The main communication method is over Reticulum Relay Chat using clients like NomadNet (made by Mark, unintuitive but works well) which is basically IRC and is purely sending messages to everyone in that channel at that time and is centralised around self hosted servers.
What would be the benefit of not using tcp or http ? It is used only as a way to transport binary data.
While the project is interesting from a technical perspective, there are some serious issues. The main implementation has a vague, non-standard licence, but this is not overly problematic as alternative implementations can simply be used instead. A bigger problem with Reticulum is that it does not appear to offer robust protection against abuse. Messages are protected by cryptography, but there are many other ways to render a network unusable, which will inevitably happen on a larger scale.
Unfortunately whenever I see the name, I am always reminded of part of the anatomy
I think it was named so for a reason: https://en.wikipedia.org/wiki/Endoplasmic_reticulum
personally i find it a beautiful match
Reticulum just means net or net-like structure.
Was also the name of the internet in Neal Stephenson's Anathem.
Haha I immediately looked for this comment and wondered if the name was inspired by Anathem.
Yes it is.
With the Artificial Inanity systems and the Rampant Orphan Botnet Ecologies (ROBE) beginning to spew tremendous amounts of crap (a technical term) onto the Ret, things are getting tricky for us ITA.
One of my favorite books that I have yet to convince any of my friends to read. So many cool ideas explored.
it'd be nice to see a comparison to rayfish.xyz
Different animal altogether.
Rayfish, Tailscale, ZeroTier, Netbird etc are a different category of tool, designed to facilitate connectivity between you and your own pool of machines. They don't generally do much in the way of multi-hop routing, but rather they orchestrate VPN tunnels on top of another routed network (like the internet).
Reticulum would be better compared to Yggdrasil, cjdns etc as routing schemes for larger networks or true meshes, which can work independently of (and have no dependency upon) the internet.
Maybe a better comparison is iroh.computer
Iroh is a framework that use QUIC to help create p2p software. It is not a mesh network.
It has nothing to do with that, like at all.
Not in a technical sense, no. But at a high level, in spirit, I can see some overlap.
In the same way that you can hack apart a loaf of bread with a spoon, yes, but they are very different things with different goals.
No published specification, single Python implementation. Shame, as this solves a problem that badly needs to be solved well.
The author of Reticulum's take is that the implementation is the specification, so if another implementation works with the standard Python one and implements enough of it's features then it implements Reticulum.
I suppose it means it's easier to develop the protocol only having to update one Python codebase but it does mean other implementations are kinda in a state where there isn't anything to properly verify they work.
There's also a issue with quite a few vibe coded implementations that don't actually properly work filling the space.
There is a published spec, I built a clean-room implementation from it. And mine makes at least two Python implementations. The protocol itself is solid, the spec just isn't marketed.
Where is the spec?
It seems like there's 14 different mesh networks that all are out there claiming to be decentralized and this and that. This is ridiculous.
We need to develop one universal standard that covers everyone's use cases.
For anyone who missed the reference: https://xkcd.com/927/
The resistance is out there somewhere?
People don't seem to get this was a genuine question. What is a network without people using it? Anyone here on reticulum?
The internet may be ripe for disruption, as it has been domesticated for corporate use. It is no longer ours. Browsers have become an operating system unto themselves. Something much simpler would work also. Corporate bloat is continuing to pile up, while markdown evolves to do everything we want.
[Gnomic comments go over people's heads (I'm guilty of writing these too). Gotta think in UX terms - write to make your audience's lives easier, not harder.]
You're right, and I think a lot of these projects hit the same problem: "what can this offer me that centralized platforms can't?"
Take the Gemini protocol: if someone invents a "Gemini killer app", it can be replicated over the web because the web's features are a superset of Gemini's features. Medium is the centralized equivalent of traditional blogging. The whole indie web could be emulated by Meta (it's not big enough for them to care). Much like capitalism eating subcultures, the existing internet will consume, commoditize and re-sell all alternatives.
The only exceptions I can think of are decentralization itself (centralization can only offer an emulation of decentralization) and freedom from corporate censorship, and not enough people care about them to make the switch.
I'd love to hear counterexamples if anyone has them, because I don't like this conclusion.
People do use it but it's definitely a small community. It has the NomadNet Page protocol which is basically small no client side script webpages (forms and links only) in a fairly bandwidth efficient form.
> Anyone here on reticulum?
Is that even the right question?
I only read the intro message but the framing I arrived/guessed at was:
"open, distributed alternative to tailscale"
So I think my first usecase would be to use this instead of tailscale to connect to my Pi?
Have you guys heard of Netbird? It's exactly what I was looking for as a Tailscale alternative.
How is that related?
Bros sponsored lol. Nah but good recommendation, looked it up and it seems pretty interesting