@clacke the spec development process for Matrix is managed by the MF, not NV, and happens in the open. So like XEP development at the XMPP Foundation, anyone who is interested can get involved. Not sure if the frequency of bits of the spec being frozen says anything much.
@clacke
> Is there good reason to consider the MF sufficiently independent from NV?
I may be drinking the Kool-aid, but my impression is that NV are aware of the dangers of taking the investor money, and have done everything they can to put MF in charge of the protocol. Not sure who owns the copyrights and trademarks on Riot, Synapse, and Dendrite, but they could be forked at any time by anyone unhappy with the dev direction NV is taking them in.
Synapse code being available and Matrix being a federated protocol satisfy @sir 's requirements in point 2). But ...
> Do multiple implementations actually exist
Yes. Not sure exactly what plans NV have for Dendrite and Synapse, but there are also Construct:
https://github.com/matrix-construct/construct
... and Conduit:
https://git.koesters.xyz/timo/conduit
@clacke Not sure how much progress has been made on getting Conduit and Construct feature-complete and production-ready, but Dendrite is neither yet. A lot of people argue that Synapse isn't really the latter, and even it's developers admit it's a rough-and-ready prototype, which is why they're developing Dendrite.
@af Really? I'll grant you point 3), but ...
1) a new standard instead of integrating with an existing one
2) server is free code and multiple homeserver implementations exist
4) Matrix Foundation takes no venture capital. New Vector does, so I'd be cautious about using modular.im . But the juiciest data (private chats) is now E2EE by default, so not much datafarming potential. NV don't own any of the "IP" that would allow future acquirers or investors to dominate Matrix development.
@sir