All that really shows is "the project has an active community". Have you seen how many XMPP has? Or Python[1]? This is just normal open development metadata.
Python has had 664 PEPs in 23 years (29 per year). XMPP has received 495 XEPs in 23 years (22½ per year).
Matrix has received about 650 in 8 years (>81 per year). Four times the change rate is quite annoying when writing code against the spec. Plus, most XMPP clients only support a fraction of the full spec, so by that comparison the impact of the rate of change is even worse. Furthermore, most XEPs and PEPs are mere (optional) extensions, whereas a lot of MSCs are alterations of existing APIs. Any JSON parser used for Matrix needs to anticipate fields changing or being added all over the place because you never know when random fields show up all over the input data because of a spec change.
The way the Matrix spec is developed feels a lot more like a proprietary company spec that happens to be published on Github than the IETF/XMPP/Python spec process. The rate of change is high and almost all changes are done to serve new features for the two or three major players that bought into the Matrix ecosystem.
One recent change that comes to mind is the move from secret, public URLs for media, to authenticated URLs. The setting to force that changeover won't apply everywhere for a while, but it'll completely break every media-supporting client written before the spec change.
Nothing wrong with extending the spec to improve the product, but with how fast the protocol is growing, I wouldn't want to be tasked with maintaining a Matrix client and I don't have much faith in the forward compatibility of the few Matrix bots I've written either.
It is a GOOD THING for people to open MSCs and try to evolve Matrix, and the number of open proposals shows the enthusiasm in the ecosystem for doing so and proposing ways to evolve the protocol.
> I wouldn't want to be tasked with maintaining a Matrix client and I don't have much faith in the forward compatibility of the few Matrix bots I've written either.
Authenticated media is literally the first time we've made a significant breaking change on the CS API in 10 years - and was effectively a security fix, to stop people abusing Matrix as a CDN. Bots I wrote 10 years ago still work today without changes (other than auth media).
> The way the Matrix spec is developed feels a lot more like a proprietary company spec that happens to be published on Github than the IETF/XMPP/Python spec process.
