> Nobody has completely nailed the UX for long-term long-form conversations, but services like Signal are designed with those kinds of conversations in mind.

No, they're actually not designed with any kind of "long termness" in mind! Case in point - one cannot move to a new device and still have access to conversations that happened on the older device with Signal. All old conversations just have to die with the old device, and the new device starts as if Signal was installed for the very first time - including starting conversations with people, getting into groups, etc. Backing up the old device and restoring that into the new one wouldn't help for this. There is no message migration feature available (separately) when one changes devices, and what's worse, in my limited observation, the Signal team just closes feature requests and issues filed on this front on Github. The ugliest part is that this is not even explained when someone installs the app or activates Signal on a new device.

So Signal is good only for ephemeral messaging where one doesn't care about past messages and their availability. Wire has the exact same problem too. One might as well treat apps like Signal and Wire like voice call apps rather than as text/photo messaging apps.

It seems like these apps make some naïve assumptions about the target user either not changing devices for many years (self-defeating in a way with smartphones, where security updates stop after 4-5 years, maximum) and/or not caring about having access to older messages whenever the users gets a new device.

Also, no way to have multiple separate conversations with the same person, or threaded conversations.

These may just be app problems -- there may be protocol support for future applications to provide these (this is where Matrix stands today). But the lack of interest in these things doesn't look good.

tbh there's significant interest in decentralised comms, at least from the geek community - the matrix.org server alone is pushing out around 1500 msg/s atm and we're having to panic around finding faster hosting hardware as a quick win whilst working on longer term homeserver scalability work. obviously this needs to transplant into mainstream interest, but we're hoping that clients like Riot will get the UX sufficiently good to bring decentralisation to the mainstream folks otherwise stuck with WhatsApp and friends.

I got a group of about 30 non-technies using Signal, not because they knew what what perfect forward secrecy is, but because they grew vaguely suspicious of the motivations of governments and Facebook, and people/Google-results they trusted were positive.

The details don't matter to ordinary people, but our approval matters much more than you think. I have very high hopes for Matrix.org.

I think your group of people would be highly disappointed whenever each person moves to a new device and finds that they have essentially "lost" all old messages and have to re-join groups afresh.

> The details don't matter to ordinary people, but our approval matters much more than you think. I have very high hopes for Matrix.org.

I agree on this part completely, and I try to introduce better tools to others too. But not having messages restored on a new device (even when a user backs up and restores all data) and not providing a way for that when changing devices is unacceptable for me. At this point I'm back to Telegram, which is a compromise, but provides a much better experience as far as usability and meeting the expectations of non-tech users go.

matrix e2e is still beta, and we just haven't implemented e2e session sharing yet. you can now export/import keys when migrating between devices as a workaround but proper sharing is coming soon.

I had a brief look at Matrix recently, and I'm really looking forward to it becoming mature. A decentralized solution with emphasis on security and privacy is what I've been praying for not just for myself, but for the sake of all humans!

You're right, losing messages is a bad experience for most people.