The only section that doesn't end with "we suspect foul play but did not investigate" is a citation of [1], so might as well read that instead.

[1]: https://github.com/libremonde-org/paper-research-privacy-mat...

Perhaps because the vast majority of points in the article are obviously false or exaggerated out of context?

Needless to say Matrix servers and clients don't send any data (private or otherwise) by default to central servers. The only central services we have are the identity lookup service (i.e. map phone number & email to a matrix ID), which is strictly opt-in and various anonymous stats reporting (which is also strictly opt-in).

This doesn't seem to have any reasons to avoid matrix that I can see. The whole bridges section is strange since that's set up by each homeserver admin, and you can just.. not do that.

This article seems all about a particular implementation, not the protocol.

"Why would anyone use X after reading a critic blog post about it?"

Matrix/Element is the most promising messaging platform from a privacy standpoint, but I wish that they didn't chose Electron for building the Linux client...

That article makes it seem like private servers unconditionally sends metadata to official servers (is this really the case, seems odd?)

Other articles mentions sending metadata only to servers you have users from (which of course can be a problem, but at least is explainable).