A cryptography-backed networking project that doesn't know about FIPS?? That's an impressive credencial.
There's some interesting stuff here, I find nostr to be a bit too seat of the pants for my comfort, but that doesn't mean there aren't good ideas to be found. I may have to write up a comparison between this project, yggdrasil, and maybe named-data networking if I ever get around to running a sim.
I don't know and I don't think anyone should care about this other FIPS you mentioned, but it's not my project, it's from this guy apparently. He seems to be someone who would know about the other FIPS if it's that important, you could ask him.
I'm not a big fan of this name anyway, would be good if it changed to something nicer.
Count yourself lucky, then. If you've had anything to do with selling commercial software to US customers, FIPS is the proverbial sand in the oil, gumming up all kinds of progress.
Likely the biggest differentiator is the same as in Nostr vs SSB: it has Bitcoin's signatures slapped on top, so it smells to the Bitcoin crowd better.
Yggdrasil seems to be only an overlay network that assumes an TCP/IP network, while FIPS is intended to run from the lowest level of the stack so it's possible to do real-world mesh networks with it and connect them to build an alternative internet.
Aside from that I think Yggdrasil uses (or used to use) a DHT, which likely works great for their size and scenario, but can't really be used for a large scale real mesh network.
This is really cool. At first, it read purely aspirational, but, as I read on, I got the impression that the author of this has put a ton of time into working this out on a technical level. I will definitely try it out and play with it.
Is it just me, or does this project seem vibecoded? It's a month old, but one contributor has contributed 100k lines. I know a lot of those lines are documentation, but that seems kind of suspicious itself. I'm not usually one to point em dashes, but there are a lot of em dashes in the documentation, and there's just a lot of documentation. So either the creator is really good at writing technical documentation and loves a good em dash, or it's LLM slop.
There's also commits with the title like "Session 51: Create fips-intro.md protocol introduction" and "Update design docs for session 142 implementation changes", which read to me as vibecoding.
You make good points. I was unsure of it myself. I've only read some of the docs so far, and it doesn't give me the impression of vibed copy too much, but who knows. Also the fact that the author apparently is an old cryptographer gave me some confidence this isn't just throwaway slop.
I've been looking for a protocol that allows multiple independent mesh networks to exist without central authorities and later merge automatically as soon as a link is established.
Hopefully this is it, although I don't fully understand it yet.
I've discarded previous similar protocols that relied on flooding or DHTs as unscalable, but this one looks better from a first glance.
I hope one of these mesh networks takes off. I don’t know enough about network design to know if any of these are actually good but hopefully one or more of them pan out.
It's very unlikely. E.g. yggdrasil exist and works fine. I've been running it for years. But is practically useless - I can just use normal Internet that works just fine, and encrypt connections in the higher layer.
Nowadays p2p application/transport level connectivity is a way better approach to accomplishing same thing that weird networking solutions would do. E.g. Iroh.
Mesh networks by their very nature typically need to make core architectural decisions that make them integrate poorly with general purpose networks. If you like mesh networks I would recommend playing with Meshtastic. The hardware is affordable, and there's surprisingly lot of people running it already, so chances are you'll get connected to existing users and it will work. Check the public maps.
I think you're confusing things. From my understanding FIPS isn't about doing p2p connections on top of existing TCP/IP connections. That would indeed be useless.
Meshtastic is a great hobby but it cannot possibly ever scale to more than a few hundred nodes, so it's not a serious project if one really wants to take a shot at building a parallel internet.
Oh, thank you for pointing it out. If the emphasis is on actual physical connectivity indeed, then maybe my criticism is premature. Though I suspect it will end up running over the existing Internet anyway, and the big claims will probably not pan out. But who knows. Good luck anyway.
it cannot possibly ever scale to more than a few hundred nodes
The big question in my mind (beyond how) is the why. What compelling feature would make mesh technology worthwhile. The most compelling use case I can think of personally is self organizing wifi walkie talkies.
The way I think about it is a combination of a fun hobby and accessible disaster utility. With meshtastic, for less than $100 you can have a setup that pairs with your phone, runs from baterry power and allows you to exchange messages with people in your area up to donno... hundreds of miles.That's valuable in many scenarios: civil unrest, war, natural disaster. And when things are fine you can check on the local channel and look and goofy messages by strangers.
The way the internet is organized currently provides a lot of single points of failure where one can snip off a large chunk of the network at will
Mesh networks won’t fix the problem of the physical infrastructure still being organized that way, but it might help with creating (probably low bandwidth and probably slow) connections between locations that would otherwise be physically disconnected.
dmbaturin | 5 hours ago
The protocol seems interesting but why would anyone take a name that clashes with an incredibly well-established acronym?
[OP] fiatjaf | 4 hours ago
I had never heard of that, so I'm not sure it's that well known.
In any case, most good acronyms or names have already been used before, so it's not anyone's fault if clashes occur.
aag | 3 hours ago
You're giving yourself too much credit. FIPS has been around and widely known for ages.
No, the person who comes second is at fault, and should at least do a web search.
jrwren | an hour ago
it tells us a lot about the author and either their hubris or ignorance.
it tells me enough to know that I'm not interested in whatever it is.
yes, naming things is hard. do better.
xjix | 3 hours ago
A cryptography-backed networking project that doesn't know about FIPS?? That's an impressive credencial.
There's some interesting stuff here, I find nostr to be a bit too seat of the pants for my comfort, but that doesn't mean there aren't good ideas to be found. I may have to write up a comparison between this project, yggdrasil, and maybe named-data networking if I ever get around to running a sim.
[OP] fiatjaf | 3 hours ago
I don't know and I don't think anyone should care about this other FIPS you mentioned, but it's not my project, it's from this guy apparently. He seems to be someone who would know about the other FIPS if it's that important, you could ask him.
I'm not a big fan of this name anyway, would be good if it changed to something nicer.
viraptor | an hour ago
viraptor | an hour ago
It's really well known. It's on the level of calling something POSIX, or ANSI. For security/crypto area, I promise it's not a good look to reuse it.
[OP] fiatjaf | 44 minutes ago
OK, I'm convinced. The name should be changed.
gerikson | 3 minutes ago
Count yourself lucky, then. If you've had anything to do with selling commercial software to US customers, FIPS is the proverbial sand in the oil, gumming up all kinds of progress.
drumbox | 5 hours ago
how does it compare to yggdrasil?
dpc_pw | 4 hours ago
Likely the biggest differentiator is the same as in Nostr vs SSB: it has Bitcoin's signatures slapped on top, so it smells to the Bitcoin crowd better.
[OP] fiatjaf | 3 hours ago
Yggdrasil seems to be only an overlay network that assumes an TCP/IP network, while FIPS is intended to run from the lowest level of the stack so it's possible to do real-world mesh networks with it and connect them to build an alternative internet.
Aside from that I think Yggdrasil uses (or used to use) a DHT, which likely works great for their size and scenario, but can't really be used for a large scale real mesh network.
thasso | 2 hours ago
This is really cool. At first, it read purely aspirational, but, as I read on, I got the impression that the author of this has put a ton of time into working this out on a technical level. I will definitely try it out and play with it.
Unreal | an hour ago
Is it just me, or does this project seem vibecoded? It's a month old, but one contributor has contributed 100k lines. I know a lot of those lines are documentation, but that seems kind of suspicious itself. I'm not usually one to point em dashes, but there are a lot of em dashes in the documentation, and there's just a lot of documentation. So either the creator is really good at writing technical documentation and loves a good em dash, or it's LLM slop.
There's also commits with the title like "Session 51: Create fips-intro.md protocol introduction" and "Update design docs for session 142 implementation changes", which read to me as vibecoding.
[OP] fiatjaf | an hour ago
You make good points. I was unsure of it myself. I've only read some of the docs so far, and it doesn't give me the impression of vibed copy too much, but who knows. Also the fact that the author apparently is an old cryptographer gave me some confidence this isn't just throwaway slop.
wrl | 16 minutes ago
Check the
.gitignore.[OP] fiatjaf | 5 hours ago
I've been looking for a protocol that allows multiple independent mesh networks to exist without central authorities and later merge automatically as soon as a link is established.
Hopefully this is it, although I don't fully understand it yet.
I've discarded previous similar protocols that relied on flooding or DHTs as unscalable, but this one looks better from a first glance.
dubiouslittlecreature | 4 hours ago
I hope one of these mesh networks takes off. I don’t know enough about network design to know if any of these are actually good but hopefully one or more of them pan out.
dpc_pw | 4 hours ago
It's very unlikely. E.g. yggdrasil exist and works fine. I've been running it for years. But is practically useless - I can just use normal Internet that works just fine, and encrypt connections in the higher layer.
Nowadays p2p application/transport level connectivity is a way better approach to accomplishing same thing that weird networking solutions would do. E.g. Iroh.
Mesh networks by their very nature typically need to make core architectural decisions that make them integrate poorly with general purpose networks. If you like mesh networks I would recommend playing with Meshtastic. The hardware is affordable, and there's surprisingly lot of people running it already, so chances are you'll get connected to existing users and it will work. Check the public maps.
[OP] fiatjaf | 3 hours ago
I think you're confusing things. From my understanding FIPS isn't about doing p2p connections on top of existing TCP/IP connections. That would indeed be useless.
Meshtastic is a great hobby but it cannot possibly ever scale to more than a few hundred nodes, so it's not a serious project if one really wants to take a shot at building a parallel internet.
dpc_pw | an hour ago
Oh, thank you for pointing it out. If the emphasis is on actual physical connectivity indeed, then maybe my criticism is premature. Though I suspect it will end up running over the existing Internet anyway, and the big claims will probably not pan out. But who knows. Good luck anyway.
BTW. In the same category there is https://reticulum.network/ already and https://www.scion.org/, though SCION is more ... industrial business minded, less of a scrappy grass-root.
I was just recently watching https://www.youtube.com/watch?v=XTnYVh7K6xQ which is very relevant, and if FIPS is serious about actual physical connection than Meshtastic should very much tie into the plan due to the sheer popularity.
xjix | 3 hours ago
The big question in my mind (beyond how) is the why. What compelling feature would make mesh technology worthwhile. The most compelling use case I can think of personally is self organizing wifi walkie talkies.
dpc_pw | an hour ago
The way I think about it is a combination of a fun hobby and accessible disaster utility. With meshtastic, for less than $100 you can have a setup that pairs with your phone, runs from baterry power and allows you to exchange messages with people in your area up to donno... hundreds of miles.That's valuable in many scenarios: civil unrest, war, natural disaster. And when things are fine you can check on the local channel and look and goofy messages by strangers.
dubiouslittlecreature | an hour ago
The way the internet is organized currently provides a lot of single points of failure where one can snip off a large chunk of the network at will
Mesh networks won’t fix the problem of the physical infrastructure still being organized that way, but it might help with creating (probably low bandwidth and probably slow) connections between locations that would otherwise be physically disconnected.