Oh but there's one more thing about Matrix
If you don't like it, the CEO of the company will come argue with you about it here
I don't check for replies to my comments on HN because this community is too toxic but I'll bet Arathorn eventually comments to tell me I'm wrong. Maybe he'll promise that they don't sell the vector.im identity data. I don't trust him, anyway.
It does work decently on my phone though.
> the identity server is centralized
Is that really a downside? I don't know a single newcomer who found authenticating with "Nickserv" was a natural thing to do.
Edit: Though I believe that matrix.org requires an e-mail address nowadays. But most homeservers don't.
Damn. Fuck that.
My phone number and email is none of their business.
It's so dispiriting to see every form of communication on the internet moving to ever more spying on their users.
Don't want them to know your phone number or email address? Simply don't give it to them. It won't break your Matrix experience in any way. Using the centralized identity server is entirely opt-in for this specific reason.
So there are no channels on Matrix that require you to identify yourself before you can join?
Third Party ID's (email etc) are only used for opt-in contact discovery.
I remember that I gave my email when I signed up for Matrix at element.io but haven't found a way how to find out if I'm using the identity server. If my email ended up there, how can I see if it is public? How do I actually query the identity system for information, especially which information it has about myself?
I have tried to find this out, but it is not obvious at all and I could not find much information except the repeated statement that the identity system is optional. Any hints about the identity system are appreciated.
These are optional recovery methods.
F.U.D.
Nicknames are ephemeral. It is designed to be this way.
You can use SASL and connect with a certificate to give yourself a fingerprint to validate your identity.
It's fun and simple. Don't throw out the baby with the bathwater.
SASL is definitely an improvement over NickServ.
But...the community itself will likely beat them here.
> Fewer clients, way more complex protocol
This is actually a valid complaint, though it's partly due to Matrix being pretty young. The other part is that Matrix does a lot more for you than IRC (eg e2ee and federation) and those extra features require complexity.
> The identity server is centralized
This is a completely optional part of Matrix. You don't need identity servers. You never have to provide your phone number. They don't even push you to when you sign up. It exists as a convenience feature for people who complain that they need to be able to discover their friends on Matrix by their friends‘ phone numbers. Also, you're free to run your own identity server for you and your friends or whomever.
> other attempts [at creating a homeserver] appear dead
This one is way off. Synapse, the "one complete server" you must be referring to, is going to be deprecated. The Matrix Foundation team is working on Dendrite to replace it. There are also several independent projects that are very active, notably Conduit, started by one guy unaffiliated with the Matrix Foundation team. Conduit is extremely far along and is under constant development.
> All of Matrix is controlled by New Vector LLC
The people who created Matrix were careful to separate their for-profit projects from FOSS community projects. Matthew Hodgson (Arathorn) explains it well here [0]. TLDR: The Matrix spec, the Element clients, and the Synapse/Dendrite homeservers are maintained by a nonprofit. Same team, different organization and funding. Everything is open source, so anyone is free to fork at any time.
Interesting, I've always been reticent to host my own instance because of how complex Synapse is to setup for the uninitiated. Hopefully Dendrite makes it simpler.
I hugely disagree.
I have tried to setup a "homeserver". Synapse is huge beast, complex, CPU and memory-hungry, hard to setup (no, a Docker or VM instance does not mean easy to setup). Dendrite is not ready and has been "not ready" for years, not to mention it is also huge and hard to setup, and from the same vendor, so it allays none of the fears. Construct is the _only_ one of all the 3rd party servers which I have found to be actually in a fully usable and federable state (the rest being all in different states of abandonment and uselessness), created by an author who is _openly_ critical of Matrix-the-company and their policy regarding 3rd party servers (incl. how most of the protocol is just defined by bug-by-bug Synapse compatibility and how the Matrix webpage to this day has no 3rd party servers list -- but happy to feature 3rd party paid-hosting list)
Compare this with Jabber. I install in seconds any of the _multiple_ servers that come with Debian stable. They are by now years old, but they still can connect with _everyone_ on the Jabber network. In Matrix, if I run a 2 year old Synapse, it will be like death-by-ostracization; same if I run today's Conduit.
After a couple hours I gave up on Synapse, having accomplished nothing and only being more confused than when I started out, as far as any path to a deployed config I'd be comfortable with. That they sell hosting, and it was such an incredible pain in the ass to run sanely on one's own, sets off some alarm bells.
It's also true that the new homeservers aren't yet ready for prime time. But they're coming along really fast.
It's true that homeservers by third parties had a bit of a rough time in the beginning due to finding "bugs" in the spec that no one noticed before because there had previously only been a single homeserver implementation. That situation has been improving though. I've seen Conduit development result in several different patches submitted to the upstream Matrix spec.
I have seen way too many being abandoned to believe that. In fact last time I researched I put Conduit in the abandoned list, though apparently I was wrong. Matrix is just a too complex protocol, and way too focused on one and only one server implementation. Which is ironic considering probably every other competitor protocol (save for perhaps IRC itself) is even more complex, but at least they were not that complex when they initially showed up -- so other people could actually start implementations that survived and growed along.
I'm somewhat of an layperson on this topic in that I haven't actually ever tried to work with the protocol on a low level. But my take on it is that if they can slow down the pace of Matrix evolution in the next couple years, the protocol will be in a good state. It's complex, but it's providing extremely complex features.
But if they keep going at this pace for too long, third party implementations may become less feasible.
Matrix 1
Matrix 2
Matrix 3
Matrix 4
Matrix 5
Matrix 6
Matrix 7
Matrix 8
Matrix 9
Matrix 10
Matrix ....
Matrix 999
Which clone of the matrix do you mean to use and how many messengers to install? XMPP has similar Matrix capabilities and modular architecture.
XMPP protocol surpasses the Matrix, it is clearly seen on the client for Android Conversations https://f-droid.org/en/packages/eu.siacs.conversations/
There is a project currently active to make some sort of replacement.
The identity server is completely optional (you can just login with username & password), and you can opt out or swap it in element -> settings -> general. Also, there's plans for getting it replaced.
There are quite a handful clients and servers. They are replacing their most featureful server (Synapse) with other implementations (Dendrite) to not have a monoculture. Heck, the German government is sponsoring development of yet another server (conduit.rs, in Rust).
Your approach sounds like wanting everything implemented right away, and giving no leeway. And I don't think your attack on their CEO is relevant even. Their whole approach is based on that you shouldn't need to trust them at all.
Given that your handle is "newvectorllc" (the previous company name of Element), it strikes me that you are so hellbent on trying to paint them in a bad light.