Gotta say the way Kaidan documents their work is probably the most impressive I have encountered in the FOSS world. The list of new features is long and detailed, as are the bug fixes. The packaging status at the end of the page is brilliant. We can quickly see from knowing our distro whether we have the changes.

it’s about who I can reach.

Only if you play the pushover. I will not install Signal - full stop. I hand out a contact card that has my XMPP JID and OMEMO fingerprints. Others install XMPP for the purpose of reaching me.

I miss when Google and Facebook were XMPP

XMPP is a good option precisely because those companies are not involved.

My bad lol. I read the whole article before commenting. was probably just sleepy

Depends on who you ask. Generally most privacy concerns associated with XMPP can be mitigated by hosting your own server (or using a trusted one).

I find it always strange when people reply as if the person posting an article was the person who wrote it. Any particular reason why you did so?

I don't think Daniel Gultsch actively subscribes to this community, although he does have his own Mastodon instance.

More power to you man. I can imagine how difficult it must be to give up on a regular income stream.

Any privacy concerns?

Never bought the app on the Play Store (got it through F-Droid), but support Daniel via Liberapay. Probably the most solid, reliable, no-bullshit messenger app that I have used.

Monocles uses a lot of generated code these days.

Anyway, I've switched to Cheogram and so far it seems more reliable.

EDIT: And the only reason I use Cheogram in addition to Conversations is because I need to seperate my group chat account from the account with which I talk to friends, so that I do not miss messages. :D

Is that a webxdc app? edit: chess I mean

i had monocles for a bit but i like cheogram better.

can play chess!

Very nice. Not to crash a thread but anybody running Monocles (one of many forks)?

Ah-ha, I see. Thanks!

When the author started using it didn't support edits yet

On this topic deltachat has come up with a beautiful solution I believe. Your identity is your public key, and servers are used solely to relay stuff from one end to the other. You can have 10 "accounts" on 10 different servers, servers can fall and it's still ok because there are 9 still up.

The con is that you don't have a simple, publicly usable pointer to you like a mail address or a phone number. Turns out it's the best defense against spam, creepy weirdos, trolls, so it's not that problematic.

Signal supports editing messages. I don't get it.

I think this Brazilian service wants to add Movim soon: https://isacloud.com.br/

That could be it tbh. I kinda forgot I was dealing with a server on europe, but I haven't really seen any trusworthy instances in south america and most countries here don't have well developed privacy laws.

This kind of silly reasons are exactly what people say why they can't use xmpp and prefer Telegram for example.

I had acquired a habit of typing a message, sending it, and only then reading it again and checking for typos. The lack of such an essential feature caused me to send incomprehensible messages, making the app almost unusable for me.

I'm sorry, what? This is the reason he starts with for saying Signal sucks? I can't even bend my head around that logic.

Well, at least he backtracks at the end.

Signal doesn’t actually suck. It’s fine. It gets the job done. It lets me communicate with my friends. We are quick to dismiss a new thing after two minutes of trying it out just because it doesn’t meet our expectations. This is especially true if we are pressured into something instead of following intrinsic motivation. We obsess over the little things that ultimately don’t matter. However, if we earnestly give it a shot, we get used to it. We adapt.

Nice ! Feeds desperately need to be implemented more

Australia has had phone number portability for well over 20 years. This means that for basically everyone, the phone number they had when they first got a mobile phone is the one they use today. Phone numbers are amazing in that they are simultaneously a stable code from which we can contact someone we know, but they are also abstract and therefore deniable (AKA "new phone who dis?").

What we have today is not a network effect problem, but an identity problem. Not only does XMPP give you a pretty unreliable identity, that identity, well, identifies you to everyone. Everyone knows dillekant@slrpnk.net is me, even people who aren't logged in.

Signal gets around this by piggy backing off your phone number, a thing people trust. Basically any traction Signal has is because of this piggy-backing. It can also use this technique to prevent a bunch of spamming / aid in moderation. Matrix, which the article also mentions, also gets around a bunch of issues through centralisation. XMPP servers need to do this manually, have moderation, etc. You also cannot guarantee portability the way you can with phone numbers.

One possibility is to do the same trick. Piggy back off phone numbers. Create some sort of attestation framework with phone numbers, then have some list of servers which will automatically create a UUID-named account based on that attestation. Now servers can come and go, and the client just needs to create a new attestation. This sort of works like a sea of servers and the client just uses the nearest one to create an account and fetch attestation information. However, it also has one of the biggest downsides of Signal: Needing a phone number.

The other trick is to use some sort of externalised identity framework, maybe openid or Solid, which means someone has to pay for the Solid ID, but not for the actual messaging framework. I wonder if that just shifts the problem one step over though. The big disadvantage here is that the deniability / anonymity is completely gone.

The root of the problem feels like DNS, and XMPP anchors off it. Maybe the right solution is something peer to peer, but then why not just use that as the messaging platform.

Cool to read that they plan to have it fall back on email when the address doesn't work via a xmpp.

I am curious to try it, but when thinking about it I realized my OIDC SSO solution uses email addresses as identifiers, which makes this a bit more complicated 😓

They actually started publishing their newsletter as a Movim/Monocles compatible xmpp pubsub blog a few months ago.

The same applies to publishing content: if you send newsletters or updates via email and messaging apps, do it via XMPP as well

Maybe the xsf should read this and do it