About a libre, decentralized, independent, secure and authentication-friendly platform-agnostic messaging app - for Christmas.

1 reply [Last post]
boba
Offline
Joined: 08/28/2017

(I could not figure out how to move a topic away from the trisquel-users forum to the general forum where it belongs, so here we go)

-- from a thread about the available version of Jami on Trisquel 8, and how to get the last version to install: https://trisquel.info/en/forum/jami-version-trisquel-8-repos-still-called-ring --

The point was, Jami seems to be installing fine on Trisquel 8 using the one-click install .deb package available on jami.net, then working fine for what it is supposed to be doing.

One friend who accepted to test it on his [propietary] OS was less lucky however, and has already decided to throw it away. It is already difficult to tell people to be patient with libre tools in the making, especially those who kindly accept to play around with our last choice of a messaging app, but the disappointment seems to have been triggered by the very nature of Ring/Jami: decentralized and DHT-based networking.

Average users might not understand, or accept, perceptible delays simply for finding someone on the network. Users with a higher level of patience might not find it much annoying given the huge benefits of a decentralized commmunication structure but still, being able to communicate easily with people on a slow internet connection, or with people who have no clue about any technical or social issues is also part of the goal. Which makes me wonder whether the ups of distributed networking might not be hindered by the difficulty of adoption.

Anyway, the Jami client will also be improving in time on all the platforms it is available for, so adoption might follow suit. The other concern I have about the way it works now is that it lacks a contact identification mechanism: someone tells me their username, I send a text message, and we both are added to our respective rosters. There is no way to check the identity of the actual account holder, both at first contact and any time later. Some stranger already (inadvertantly) tried to start chatting with me, and though I did not find that highly frightening I still find it surprising in the context of a secure messaging app. That decentralization is adding latency is only about the laws of physics, that softs have bugs is only human, but random people allowed by design to reach you directly by your user name?

Years ago I had a try at Retroshare, which seemed to go the right way for private and authenticated connections while also going the opposite way of public and anonymous connections. Since I am not interested in massive sharing of copyrighted material, I do not see the need for the latter and it somehow goes in the way of the former. Does anyone know of a Retroshare without the public/anonymous (large) bit?

boba
Offline
Joined: 08/28/2017

I am now trying the desktop client for Telegram, hoping that the Android client for Jami will improve over time.

Centralized services seem to be winning on the reliability side for the time being. Not sure whether this situation arises from the inherent challenges stemming from decentralized network management for real-time communication or from specific choices from the devs (i.e. focusing on desktop vs mobile clients). Tools developed for mobile devices seem to be easier to port to "sedentary" platforms than the other way round, but most rely on centralized servers for one supporting service or another.

Could we not imagine a desktop based decentralized application which could be ported to mobile devices where the app would behave as a "remote" extension of the desktop client? The Retroshare team seemed to be considering moving in that direction.