Delta Chat – Email Based PGP Encrypted Chat
delta.chat
delta.chat
> After almost two years of collaboration with the wonderful Iroh team, and years of discussions with numerous experts in the decentralization space, we are happy to announce that Delta Chat 1.48 apps on all platforms contain state-of-the-art Peer-to-Peer networking support, including hole punching and forward-secret end-to-end encryption. Concretely, Delta Chat now establishes private Peer-to-Peer gossipping networks between users who start a webxdc app that uses the new joinRealtimeChannel() API.
https://fosdem.org/2025/schedule/event/fosdem-2025-5853-delt...
https://fosdem.org/2025/schedule/event/fosdem-2025-5217-chat...
Does this solve all the other problems with encrypted email, which is not widely used for a reason?
Here's a discussion
https://www.latacora.com/blog/2020/02/19/stop-using-encrypte...
I honestly don't care about what the people I communicate with do, as long as I have the capability to at least own my persistent identifier (i.e. my TLD).
Just having that capability exerts just the right type of pressure on large service providers to maintain a baseline quality of service, regardless of whether the majority actually makes use of it or not, just like phone number portability has done in that domain.
but that is the case with every messenger that stores messages on your behalf. telegram, whatsapp, signal... with those ALL people are not running their own servers.
whatsapp and signal store encrypted copies of the messages, and so does deltachat.
beyond that each client also stores a copy of each message locally. as does deltachat.
in difference to all others at least with deltachat i have a choice to use a different server that i trust. with whatsapp and signal i don't have that choice.
it doesn't solve all of them but a few at least. i don't believe using pgp itself is a problem. if it was deltachat could replace it with another better encryption.
metadata is a problem. that could be solved by removing messages from the email server and storing them elsewhere. (or storing them back on the server with the metadata removed). that doesn't prevent intermediate mail forwarders from keeping a copy though. i think that is the real problem. deltachat doesn't control the communication channel and can't prevent leaking. i don't know how matrix or jabber compare here. if they use intermediate servers to forward messages the problem would remain.
message content leaking is not a problem because messages won't ever be decrypted on the server.
long term secret leaking is a problem tied to the current implementation of pgp. deltachat could change that (and maybe it does?)
> Reliable instant messaging
If it's based on e-mail, it's just not. (if we assume generally close-to-realtime delivery and not being blocked by random recipients)
WhatsApp lets people find each other with phone numbers instead email addresses. This has a profound difference in real-world usability because it decreases friction.
Before WhatsApp, people texted each other using cell companies' SMS service.
Why did Person A (who already had an email address) send a SMS text to Person B (who also already had an email address) to chat?!? Because both people already had each others' phone number in their smartphone's address books. Recording each other's email address is less likely so it won't be used as "ids" to chat -- especially between friends & family.
To wit, my mother has already memorized my brand new phone number that I've only had for a few years but she doesn't know my email address at all even though I've sent it and re-sent it to her multiple times. To normal people, the phone numbers are more sacred than email addresses. WhatsApp was a continuation of the conveniences of SMS -- without paying 10 cents per message to the mobile phone carriers.
Friction from cognitive overhead is a big deal in technology adoption.
On the same note, isn't phone numbers being more sensitive information than email nowadays, hence you might not want to share it.
Look if it is saved in the smartphone's address book, why does it matter if it is email or a phone number. I bet that people don't actually memorize even the phone numbers of people close to them these days.
>Why did Person A (who already had an email address) send a SMS text to Person B (who also already had an email address) to chat?!?
I think it is because we didn't have near universal internet access in smartphones for a long time. If Whatsapp (or something like that) was a little bit late to appear, and the tech crowd was actually a bit more smarter, things like Delta Chat had a much better chance to be in Whatsapps current place.
Because when people interact with each other in informal situations, it's the phone numbers that are shared. Not email addresses. Thus, email addresses are often not in the Contacts app at all.
Arguably primarily because they want to chat with them and not email them. The two have vastly different UX beyond just contact discovery.
Before WhatsApp there were ICQ (also number-based!), MSN, Skype... WhatsApp's main contribution over these was indeed using phone numbers and contacts access for automated contact discovery, but I don't think it makes sense at all to characterize it as a usability improvement on email.
Another case in point: iMessage supports email addresses as identifiers too. I don't even have my phone number on there because I consider it a strictly inferior identifier (it changes every time I move countries, I lose it if payments to my operator ever lapse, unlike a TLD I have no way of owning it independently from a phone plan etc.)
Chats in general (not WhatsApp specifically) have "better" usability than email for some people because:
+ chats skipp the extra keystrokes of email workflow such as "Compose new email" and then enter an extra "Subject:" line which makes normal people put useless things in there such as "a quick question..." and then put the real conversation in the email body. It's a bunch of extra friction for no reason when communicating informally between friends & family.
+ chat ids such as phone numbers are not given out as freely as email addresses which makes chats have less spam and clutter and more easily isolated to personal communications. Email inboxes are clogged up with a bunch of non-chat mails like shipment notifications from Amazon.
I actually prefer email comms and have told my family to send me emails instead of text chats because I'm always at my desk and can use my full keyboard to type out a reply -- but -- they ignore me and always just send text chats. Normal people have a mental model that's geared toward text chats.
E.g. I saw my aunt again 30 years after she last saw me and one of the first things she asked was "What's your phone #? I want to send you a photo when you were little." And because I received her text, I now have my aunt's phone # in my Contacts app. But I don't have her email address. She never gave it to me; and I never asked for it.
Exchanging phone #s is the natural thing to do between friends & family. On the other hand, exchanging emails is often more natural when interacting in business settings or dealing with people at arms length.
Definitely, but that's not WhatsApp's innovation.
i do the same as you as i dont carry a phone anymore (fucking with the police).
Hell, Jabber can still do this. I'm not convinced of the other modern alternatives that don't have half the functional capabilities that Jabber do.
The only "downside" of Jabber is that it's XML based, but really if you think about it, that's a strength. Anyone here could probably parse the protocol effortlessly. There's also lots of fully functional clients out today, and servers that scale like nobody's business.
I'm more ashamed we don't invest more into XMPP: Google Talk, Facebook Chat and WhatsApp were built around it. These are companies with insane to scale userbases, tried and tested.
Here's a 2008 article from Facebook on using their chat with a Jabber client:
Nope.
Another downside is the X. The protocol being extensible means that no two clients implement the same subset of extensions making it useless. I’ve been scolded in the past for using the "wrong" client, making my messages look weird for people using the "right" client. Just make a good protocol.
This reminds me on this nice comment thread: https://news.ycombinator.com/item?id=19217818
that generic server backend that i am describing in my linked comment is something i am using here: https://news.ycombinator.com/item?id=42159045
Email is the completely wrong protocol choice for instant messaging. It's just a completely different use case. You wouldn't send a letter to the fire department if your house is on fire either.
When you use XMPP or Matrix you know your message will be delivered instantly. When you use email you don't. But that's only because we haven't defined an extension that tells the client whether the server can deliver messages instantly. It's not really that inherent to the system. Surely you've had at least one real-time conversation by email because both of you happened to be online at the same time and nothing went wrong necessitating a message delay.
In particular, Email has a lot of assumptions about acceptable delivery delays, bidirectional (non-)reachability etc. baked in that would be very hard to globally undo.
SMTP is optimized for the use case of email; XMPP is optimized for one-to-one instant messaging; Matrix is optimized for "Slack-like" chat.
I highly doubt that these three use cases are similar enough to warrant introducing an additional common protocol/layer.
Are there some compromises? Of course. But creating new accounts, or using new services isn't one of them. Email-speed instant messaging is good enough for me.
And to be honest, as much as I like Matrix, it has plenty of compromise of its own - including speed/reliability when using their own servers.
Even though I am part of this secure messaging group chat on matrix thing and I had even created a whole tierlist of security of protocols etc. and saw delta chat & have it running and fiddled with it.
The fact that it is written by creator of pypy is wholesome!!
However this comes with serious trade-offs. PGP lacks forward secrecy: if a key leaks, all past messages can be decrypted. Also IMAP offers no metadata privacy, anyone can see who you email and how often.
Signal and WhatsApp are likely a step ahead in terms of privacy with their double ratchet encryption (https://en.wikipedia.org/wiki/Double_Ratchet_Algorithm).
WhatsApp is closed source, so whatever they claim to can't be proven. And remember that both WhatsApp and Signal are legally required not to disclose to you whether they are spying on you or not.
E2E only requires a correctly-designed client, not a correctly-designed server.
Since any binary can always be deobfuscated, deliberately putting in a backdoor in the client would be an extremely risky PR move, especially in an app as big as WhatsApp, and one surely receiving lots of attention from security researchers.
Not to mention that OpenWhisperSystems (the creators of Signal) worked with Meta on their E2E implementation, and if you don't trust them to do right, you shouldn't trust anybody.
Let's face it, people over here just don't like Meta and love to spread FUD about them.
If you live in an authoritarian state with a track record of spying on individual citizens, and without legal protection for the privacy of foreign citizens, it's legitimate to decide not to trust a company like Meta headquartered in the USA.
That is not strictly true. It is possible to reverse-engineer the client. It's unlikely that anyone will find the motivation to do so, but there's no law of physics that says they can't. You can theoretically review the code that is running on your device - it's just harder than with an open source reproducible build.
It's called punching upwards.
>Since any binary can always be deobfuscated, deliberately putting in a backdoor in the client would be an extremely risky PR move, especially in an app as big as WhatsApp, and one surely receiving lots of attention from security researchers
Correct me if I'm wrong, but can't a proprietary delivery mechanism such as the App or Play stores deliver a backdoored build to an individual device, then later autoupdate it to a clean one?
Are you talking about decompilation? This, is in general very difficult, and that's when the author and compiler don't want to stop you from doing it.
Go fire up jadx or ghidra or whatever you like and see if you can find the bits of code you know are there. Now try demonstrating that a bit of code isn't there.
Yep, don't like it. They already have proven to not be trustworthy by various privacy scandals in the past
Not to mention, if you revoke a key (maybe because you lost your laptop and want to be proactive about security), without any authenticated timestamping service in the mix, all past messages and signatures can no longer be trusted, regardless of the revocation date. That's why when you revoke a key on github, all your previous commits' signatures turn red.
I've never understood why no one's succeeded in doing anything about this after all these years.
its the rest of the account and metadata that is at risk
Furthermore metadata transparency can be made less problematic by using email aliases.
now what's interesting here is a potential future development of deltachat where SMTP is only used to negotiate the peer to peer connection, and the actual chat messages are sent over the direct connection. although it sounds kind of weird to build a chat solution on top of SMTP and then bypass SMTP in the end.
the remaining benefit would be the ability to talk to people who don't use deltachat. not needing another account on a new service. not needing to develop or maintain server infrastructure.
wait, in a way it's already there: https://github.com/ArcaneCircle/live-chat
now adapt that to storing messages... ;-)
24-jan-2021 https://news.ycombinator.com/item?id=25893626 148 comments
07-jan-2021 https://news.ycombinator.com/item?id=25674894 4 commments
27-feb-2019 https://news.ycombinator.com/item?id=19263357 11 comments
21-feb-2019 https://news.ycombinator.com/item?id=19216827 56 comments
03-feb-2017 https://news.ycombinator.com/item?id=13560279 1 comment
The provider was GMX and it was their fault, not DeltaChat's. But I had to call off the experiment.
Noting that the team have put together "Chatmail" a minimal email suite designed for speed, security and convenience.
Also helps with the whole "onboarding to chat" and multiple identities not linked to existing accounts.
https://delta.chat/en/chatmail
There are multiple public servers which may be an option, instead of using an existing email account at provider X.
The chances of your email messages being 'man in the middled' are almost non-existent. Its a micro-percent of live investigations that does this.
The police RELY ON finding messages in either your inbox or the recipients. It doesnt matter if they are encrypted or not in reality. Of course they'd like to read them and if they are this deep into you. Its likely they still will (probably from the recipients lesser password hygiene than yours).
Almost no evidence is every 'plucked' out of the air and read. It just doesnt work that way.
However, software like Delta is better than nothing esp for normies. Its just limited in use for people who really need ways to frustrate LE.
Source: Ive been under NCA and FBI investigation before. Protonmail with the two password method stopped them every time. Memorise the 2nd password and password manager the first.
Unless you exclusively use proton-bridge and disabled auto-update, then proton controls all endpoints in which your passwords live.
So what is preventing the FBI from forcing Proton to serve a special UI just to you that will be used to exfiltrate your second password?
https://en.wikipedia.org/wiki/Apple%E2%80%93FBI_encryption_d...
no charges but lots of attempted snooping.
fuck them
The e-mail part made me think of this. I have no use for an e-mail based chat but an online chess game that's easy to use without registration would be cool. He's 82 and he can't use complex sites.
Lichess lets you play without registration, I believe (or at least one player; not sure if both sides can be anonymous).
This is a known Delta chat issue, but they deem it too niche to consider properly supporting aliases.
In an ideal world, when an email arrives, it should create a conversation between a sender and a recipient (alias) without adding a main account as a member. Sending a new message in that conversation will automatically use an alias as a sender. For new chats/conversations, I should be able to choose which alias to use.
Then I would be able to use it as a chat UI for my mail messages.
if you want to use different aliases in deltachat, i think the best option for now is to create different deltachat profiles for each. you can have multiple profiles active at once and you get notifications for all of them, you only have to switch profiles to see all the messages. switching profiles is a single click and feels like switching folders in traditional mail clients.
Alias is a standard email feature. If Delta Chat want's to be usable as a general mail client, aliases should just work.
https://delta.chat/en/help#why-can-i-choose-not-to-watch-the...
https://delta.chat/en/help#why-can-i-choose-to-only-watch-th...
Thank you for painting this beautiful picture.
PGP is intended for the classic "used by a human to encrypt emails manually" flow and is actually insecure if you automate around it.