Skip to main content


Lots of people are looking for secure chat platforms and stuff like that. So I thought I'd create a poster.

I excluded Telegram because it's pretty much like WhatsApp. And this https://iddqd.press/2019/12/11/telegram-is-an-obvious-honeypot.html

I would've included Signal, but I'm being skeptical here and Signal looks a bit suspicious since it requires your phone number etc.

What are your thoughts on this?

#tech
#technology
#security
#privacy
in reply to Gilbert Busana

good call on Telegram. It's actually worse than WhatsApp, WhatsApp is end-to-end encrypted by default, Telegram is not, but the way they portray themselves makes users think it is. Also, no way to end-to-end encrypt groups.

Not sure about Session here. They have a cryptocurrency token tie-in that somehow is supposed to make the network "safer" (than Tor), but some small amount of mined tokens is hard-coded to always go to the organization behind Session:
https://mastodon.social/@rysiek/106542118359065380

🤔
in reply to Gilbert Busana

in reply to Gilbert Busana

we can debate this for hours, but the long and short of it is: Telegram makes it *easy* to make a vary serious mistake, and think one is communicating in an end-to-end encrypted way, when one isn't.

And I have seen this happen.

There is no good reason to do that.

I think The Grugq put it best:
https://medium.com/@thegrugq/operational-telegram-cbbaadb9013a
in reply to Gilbert Busana

Telegram makes it easy to make a vary serious mistake
This is one of my biggest gripes about TG honestly. People should be better educated on how to use the tool within it's confines. I mean, all the info is there, but someone has to go looking to read it, which rarely happens. Good point.
in reply to Gilbert Busana

it's not about "all the info is there", it's about *misleading messaging* around this from Telegram itself. Go to their website, you'll read that "Telegram messages are heavily encrypted".

Making such claims in the context of groups not being end-to-end encrypted at all, and private chats not e2e encrypted by default, is actively harmful.

And sure, they can say "well, on page 20 of our FAQ you can read that you need to enable encrypted private chats". Doesn't fix it.
in reply to Gilbert Busana

and then there's this bit:

> Telegram keeps your messages safe from hacker attacks.

...also from their website. In e2e encrypted systems there are no messages that system operators need to "keep safe from hacker attacks". And that's how IM systems should work in AD 2022.

If Telegram team really cared about people's privacy they would deploy e2ee by default as soon as possible, and in the meantime have *super-clear* messaging about the current shortfalls. They don't.
in reply to Gilbert Busana

and to me that means that they *don't* care about user's privacy. They have some other, more important things to focus on. What those things are is anyone's guess. But that's enough for me to be very wary about all Telegram's claims about how they protect privacy, encrypt stuff, split keys, etc etc.

They are clearly not 100% honest with their users about e2ee, why should we trust them on anything else?
in reply to Gilbert Busana

in reply to Gilbert Busana

In the end, regardless which of these services we use, there's a level of trust that has to be given by the end users.
Not with XMPP! Self hosted, federated, true e2e chatting my beloved
in reply to Gilbert Busana

ah, I was waiting for the XMPP crowd to butt in.

I've set-up and run five XMPP servers. I've been a pretty heavy XMPP user, and used both OpenPGP and OTR encryption on XMPP.

XMPP is unusable for most people, because the matrix of which client/server software implements which XEPs is a kilometer deep and a mile long.

This means I cannot reliably know if the person I will be talking to will have the particular combination of XEPs available.
in reply to Gilbert Busana

> XMPP is unusable for most people, because the matrix of which client/server software implements which XEPs is a kilometer deep and a mile long.

> This means I cannot reliably know if the person I will be talking to will have the particular combination of XEPs available.

Wait what? Why would you need to know? It doesn't matter what XEPs their client or server support, you can still easily communicate with them. That's the entire point of the "eXtensible" in the name.

I've had everyone I know on XMPP since 2013ish and have never once had to know or care about what XEPs their software supported.
in reply to Gilbert Busana

yes, but what if I expect end to end encryption, for example? That's what this thread is about, after all.

The point is: certain XEPs have become important enough that "who knows if the other side implements it" is not an acceptable approach. You end up with a lowest-common-denominator thing.
in reply to Gilbert Busana

That one is clearly indicated in your client, you can't enable OMEMO unless your contact supports it, and then you can choose whether to send a message anyway knowing it's encrypted the whole way with TLS. What's the problem again?
in reply to Gilbert Busana

I'm still unclear on the problem.

You said differing XEP support made XMPP unusable, I said it absolutely did not and doesn't matter. You brought up OMEMO which requires all clients to clearly indicate support, and are now linking lists of XEPs, which again, don't matter...
in reply to Gilbert Busana

please show me where exactly did I bring up OMEMO.

My point was broader and it was about the incompatibilities between different combinations of XMPP clients and servers stemming from spotty implementation of XEPs in said clients and servers. And about the usability challenges that stem from that.

This whole branch of the thread is a pretty good example of proponents of XMPP not accepting criticism, instead of fixing things.
in reply to Gilbert Busana

I'm trying to point out there are no incompatibilities. Virtually the entire public federated XMPP network runs either ejabberd or prosody as a server, both are well maintained and support anything anyone would want.

*Most* everyone runs a modern well supported client, like Conversations, Dino, Siskin, Gajim to name a few, but even if your contact wants the pain of running pidgin that doesn't affect you.

XMPP's only problem is combatting decades of misinformation from people that connected to gtalk using pidgin once in 2006 and found it to be a bad experience (it was terrible!), but it's been the best IM experience for well over a decade at this point, and the only one that is a standard with wide adoption and multiple independent implementations that you can run yourself.
in reply to Gilbert Busana

cool, cool. Meanwhile I look at those XEP implementation tables and they ain't pretty.

We can run in circles for days.

Wake me up when XMPP creates some kind of a *standard* suite of XEPs that *have to* be implemented in clients and/or servers, and some form of "certification" process, so that when I see that a client is (say) "XMPP 2022"-compatible, I know what to expect.

Instead of me having to consult said tables of XEP doom.
in reply to Gilbert Busana

@Rysiekúr Memesson 🇺🇦 @moparisthebest @ThatOneCalculator :calcdumpy: :calckey: @DarkSky 💙💛 @Peter Sanchez

I think that the standard suite of XEPs is in https://xmpp.org/extensions/xep-0459.html and there is a server compliance suite for those at https://compliance.conversations.im/ which publishes the results.

I have a vague memory of a similar test suite for clients, but I can't find anything, so maybe I'm remembering it wrong.

Anyway, most people are using one of the clients with good support, these days, unless they have very specific requirements, so things aren't as bad as they were in the gtalk era.

Roberto Resoli reshared this.

This website uses cookies to recognize revisiting and logged in users. You accept the usage of these cookies by continue browsing this website.