Be careful when installing Yggdrasil Network on your device - your device address will be available in a network/peer explorers and if your firewall is not configured to reject incoming connections from the yggdrasil ipv6 interface - your locally running services could get exposed.
this should be standard advise, always configure your host firewall correctly. Especially if you use something like ArchLinux where it doesn't come pre-configured
Spam can be filtered effectively client-side with a good spam filter. This has worked well for me for decades without the need for any server-side spam filtering.
You mean like LLM filters? Right now it's all reputation based on IP and domain with a whole ecosystem of anti-spam companies like Spamhaus, SenderScore, ProofPoint, etc.
Using NLP / LLM spam filtering would presumably be either inaccurate or expensive or both. Someone would have to pay for it.
Bayesian filters are basically just a cheaper / worse version of what an LLM filter would do. Very easy to beat. Especially if the spammer is using an LLM to write a semi-unique email for each recipient.
What I'm trying to tell you is hat this has de facto worked for me during the past 20+ years. I get ca. 100 spam mails a day and they all get neatly sorted in the spam folder. There is no server-side filtering at all, my email provider allows users to switch that off entirely (and better should because it's very faulty).
As I've said, I'm not interested in theoretical arguments. All of my domains wildcard forward to the same email address, too. Filtering client-side has never been a problem.
That's pretty interesting if true. It must be because you're the only one still doing it. Using techniques just for your idiosyncratic inbox doesn't make sense to a spammer. But if every inbox was only doing this, then your experience would be much different. Nothing but unstoppable spam. Exposure = money. Why would they not spam you? They don't like money?
Knuth once said there’s two kinds of programming languages in the world: Languages everyone complains about, and languages nobody uses. I feel like there’s some corollary here…
I was surprised to see this on the HN homepage, I didn't create Tyr but I did create Yggmail (https://github.com/neilalexander/yggmail) which it is based on. There is no store-and-forward as such, the sending node will keep the message in its outbox and will keep retrying until the destination is online.
> There is no store-and-forward as such, the sending node will keep the message in its outbox and will keep retrying until the destination is online.
Yes I might be wrong but my understanding is there is no point in creating another system where messages hop from one peer to the other like Meshtastic or Reticulum (what make sense for their use case).
Let's say users have their "email server" running on both on there mobile phone and a home server and in sync.
We can expect 2 of the 4 servers will be online at the same time to send the message. I personally like those odds, Internet is pretty reliable in our days.
I believe we have spent too long trying to solve very hard trilema in messaging, trying to have it all: confidentiality, anonymity and uncensorability ... and reliability ... and ease of use.
The result is in practice most people use GMail, Outlook and Whatsapp.
Yggdrasil is fantastic, it goes back some original ideas of the Internet we have almost forgotten, and in practice solves a lot of problem we have been dealing with for too long.
Tyr is probably overkill with Deltachat on top of yggdrasil. The network already is encrypted so it's fine to send plaintext emails as long as there's no 3rd party email hubs.
back in the day a few of us used to run ssb (secure-scuttlebot) over yggdrasil (and cjdns before that) and that system would distribute the private messages to all of the peers within 3 hops. offline peers would just sync up when online and then decrypt the messages sent to them.
ssb's been broken for around five years, but now that it's working again it'd be fun try this experiment again.
2026 could be the year mesh networks finally take off!
Curious why you believe it was broken, and is now fixed. What new development are you referring to? I agree that Patchwork kinda took a dive, and functionality started to bitrot with each new maintainer...but it still replicates feeds.
I couldn't get any of their latest versions working. The ssb-server was still functioning, but had no working client that I could find. https://github.com/evbogue/ssbc is a working fork with a patchbay lite client from circa 2015/16 live at https://ssb.evbogue.com/ (with git-ssb!). I'm also recreating pfrazee's original Phoenix client from scratch.
Let's talk more on a more appropriate channel. Are you on bsky? we're having a small discussion there about "bringing open source projects back from the dead with AI" right now.
I kinda understand the point, but e-mail of all things… the one thing in the current tech stack that is in fact "P2P"… technically all you need is either a VPN that allows incoming connections to you on a fixed address on tcp/25, or a dyndns and any ISP with inbound tcp/25 open…
Also, E2E encryption >> "the network handles that".
"Ok, we got an overlay network going. What protocol should we use for standard comms?"
Scenario 1:
"Email?" "Email I guess." "Email."
Scenario 2:
"IRC?" "Nah, XMPP." "Matrix!"
"Oh, we're suggesting chat protocols? Here's a #11th draft standard that ticks all the boxes for modern secure private chat."
You’re OG. My first was some unknown distro that installed in DOS on my Win95 machine and dual booted that way. Totally confused me. Second was Red Hat 6.0 in 1999. That one, I was a little more successful with.
I did try that weird linux on fat32 distribution but like you I completely forgot its name. I remember that I installed it because I wanted to run bitchx and be able to send ping of death!
The reference server is an Android app so yes, that is probably the point of the default design, but reading the README I believe you can also use a more traditional server-to-server setup:
DeltaChat/ArcaneChat Integration
DeltaChat and ArcaneChat are perfect companions for Tyr. These are messengers that use email protocols but provide modern chat interfaces. When you configure DeltaChat/ArcaneChat to use Tyr's local server:
1. DeltaChat/ArcaneChat sends messages via SMTP to Tyr
2. Tyr wraps them in Yggmail protocol and sends through Yggdrasil
3. The recipient's Tyr receives the message via Yggdrasil
4. Their DeltaChat/ArcaneChat fetches it via IMAP from their local Tyr
5. All this happens peer-to-peer, with no central servers
If you run Tyr on a VPS/RPi/old smartphone, you can still exchange messages decentralised this way, as long as your server and the device/server you're communicating to are both online, and have DeltaChat/ArcaneChat fetch the messages later.
Such a setup could be useful if you find people around you using Tyr and you're losing messages because your phone kills the app, though a PoC like this probably won't have much of a network effect.
Be careful when installing Yggdrasil Network on your device - your device address will be available in a network/peer explorers and if your firewall is not configured to reject incoming connections from the yggdrasil ipv6 interface - your locally running services could get exposed.
Should be noted that this does not apply if you install Yggmail only without mainline Yggdrasil, as Yggmail embeds its own node and does not use TUN.
this should be standard advise, always configure your host firewall correctly. Especially if you use something like ArchLinux where it doesn't come pre-configured
Systems can be so simple and elegant when you just assume no one will use them to send spam.
Spam can be filtered effectively client-side with a good spam filter. This has worked well for me for decades without the need for any server-side spam filtering.
You mean like LLM filters? Right now it's all reputation based on IP and domain with a whole ecosystem of anti-spam companies like Spamhaus, SenderScore, ProofPoint, etc.
Using NLP / LLM spam filtering would presumably be either inaccurate or expensive or both. Someone would have to pay for it.
No, I'm using Bogofilter and it works perfectly. I'm not talking hypothetically. AFAIK, it does some Bayesian statistical analysis.
That doesn't work alone https://en.wikipedia.org/wiki/Bayesian_poisoning
Bayesian filters are basically just a cheaper / worse version of what an LLM filter would do. Very easy to beat. Especially if the spammer is using an LLM to write a semi-unique email for each recipient.
What I'm trying to tell you is hat this has de facto worked for me during the past 20+ years. I get ca. 100 spam mails a day and they all get neatly sorted in the spam folder. There is no server-side filtering at all, my email provider allows users to switch that off entirely (and better should because it's very faulty).
As I've said, I'm not interested in theoretical arguments. All of my domains wildcard forward to the same email address, too. Filtering client-side has never been a problem.
That's pretty interesting if true. It must be because you're the only one still doing it. Using techniques just for your idiosyncratic inbox doesn't make sense to a spammer. But if every inbox was only doing this, then your experience would be much different. Nothing but unstoppable spam. Exposure = money. Why would they not spam you? They don't like money?
To be fair there have been multiple popular e-mail networks on Tor like SIGINT in the past and I never received spam there.
Knuth once said there’s two kinds of programming languages in the world: Languages everyone complains about, and languages nobody uses. I feel like there’s some corollary here…
FWIW it was Bjarne Stroustrup (of C++) who said that, not Knuth.
There are actually a lot of networks that people would assume get spammed to death, but they seem to work fine for me when I have used them.
Another example is BitMessage. It worked fine for me I never had random spam or anything.
I think your definition of "popular" here is... unpopular.
Very cool. How does this deal with offline recipients? Do the messages just get dropped, or does Yggdrasil somehow store and deliver them?
I was surprised to see this on the HN homepage, I didn't create Tyr but I did create Yggmail (https://github.com/neilalexander/yggmail) which it is based on. There is no store-and-forward as such, the sending node will keep the message in its outbox and will keep retrying until the destination is online.
> There is no store-and-forward as such, the sending node will keep the message in its outbox and will keep retrying until the destination is online.
Yes I might be wrong but my understanding is there is no point in creating another system where messages hop from one peer to the other like Meshtastic or Reticulum (what make sense for their use case).
Let's say users have their "email server" running on both on there mobile phone and a home server and in sync. We can expect 2 of the 4 servers will be online at the same time to send the message. I personally like those odds, Internet is pretty reliable in our days.
I believe we have spent too long trying to solve very hard trilema in messaging, trying to have it all: confidentiality, anonymity and uncensorability ... and reliability ... and ease of use. The result is in practice most people use GMail, Outlook and Whatsapp.
Yggdrasil is fantastic, it goes back some original ideas of the Internet we have almost forgotten, and in practice solves a lot of problem we have been dealing with for too long.
Neat
“End-to-end encrypted email for the mesh networking age”
Perhaps wish we weren’t headed for such an age but glad Yggmail is here for it!
interesting, i kinda wish we were headed for such an age but i doubt we are
Tyr is probably overkill with Deltachat on top of yggdrasil. The network already is encrypted so it's fine to send plaintext emails as long as there's no 3rd party email hubs.
I think DeltaChat in this case is used mainly for spam disincentivization via mandatory encryption.
back in the day a few of us used to run ssb (secure-scuttlebot) over yggdrasil (and cjdns before that) and that system would distribute the private messages to all of the peers within 3 hops. offline peers would just sync up when online and then decrypt the messages sent to them.
ssb's been broken for around five years, but now that it's working again it'd be fun try this experiment again.
2026 could be the year mesh networks finally take off!
Curious why you believe it was broken, and is now fixed. What new development are you referring to? I agree that Patchwork kinda took a dive, and functionality started to bitrot with each new maintainer...but it still replicates feeds.
I couldn't get any of their latest versions working. The ssb-server was still functioning, but had no working client that I could find. https://github.com/evbogue/ssbc is a working fork with a patchbay lite client from circa 2015/16 live at https://ssb.evbogue.com/ (with git-ssb!). I'm also recreating pfrazee's original Phoenix client from scratch.
Let's talk more on a more appropriate channel. Are you on bsky? we're having a small discussion there about "bringing open source projects back from the dead with AI" right now.
as a side note, nano do you still have a working pub with all of the historical data on it?
I kinda understand the point, but e-mail of all things… the one thing in the current tech stack that is in fact "P2P"… technically all you need is either a VPN that allows incoming connections to you on a fixed address on tcp/25, or a dyndns and any ISP with inbound tcp/25 open…
Also, E2E encryption >> "the network handles that".
My first Linux install was Yggdrasil, just for that, this interests me…
You’re OG. My first was some unknown distro that installed in DOS on my Win95 machine and dual booted that way. Totally confused me. Second was Red Hat 6.0 in 1999. That one, I was a little more successful with.
I did try that weird linux on fat32 distribution but like you I completely forgot its name. I remember that I installed it because I wanted to run bitchx and be able to send ping of death!
> unknown distro that installed in DOS on my Win95 machine
I had that from some magazine included cd ; my father thought I had completely messed up the family PC.
Is my understanding correct that all involved parties must be online?
The reference server is an Android app so yes, that is probably the point of the default design, but reading the README I believe you can also use a more traditional server-to-server setup:
If you run Tyr on a VPS/RPi/old smartphone, you can still exchange messages decentralised this way, as long as your server and the device/server you're communicating to are both online, and have DeltaChat/ArcaneChat fetch the messages later.Such a setup could be useful if you find people around you using Tyr and you're losing messages because your phone kills the app, though a PoC like this probably won't have much of a network effect.
If I were to run an yggmail server and configure delta-chat to talk to it, would I get a similar result?
Yes.
> Because the Internet was built around centralized infrastructure.
Yeah, well... No.
> We finally have this possibility to use true P2P email.
I mean, email is literally peer to peer.
> Your mail address is derived from your public key: <64-hex-characters>@yggmail
The email address is literally 64 hex characters? How am I supposed to use this on a daily basis?
I have serious doubts who are the target of this or what is the point of this to exist at all. It is literally worse than email.
> I have serious doubts who are the target of this or what is the point of this to exist at all. It is literally worse than email.
Yggmail is a fun proof-of-concept and that’s about it. It isn’t perfect, nor does it have to be.
“It’s email, but not as you know it.”